UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Aspects of the progressive in English and Icelandic Johannsdottir, Kristin M. 2011

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2011_fall_johannsdottir_kristin.pdf [ 3.76MB ]
Metadata
JSON: 24-1.0072196.json
JSON-LD: 24-1.0072196-ld.json
RDF/XML (Pretty): 24-1.0072196-rdf.xml
RDF/JSON: 24-1.0072196-rdf.json
Turtle: 24-1.0072196-turtle.txt
N-Triples: 24-1.0072196-rdf-ntriples.txt
Original Record: 24-1.0072196-source.json
Full Text
24-1.0072196-fulltext.txt
Citation
24-1.0072196.ris

Full Text

ASPECTS OF THE PROGRESSIVE IN ENGLISH AND ICELANDIC  by  KRISTÍN M. JÓHANNSDÓTTIR B.A. University of Iceland, 1992 M.A. University of Iceland, 1996  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY  in  THE FACULTY OF GRADUATE STUDIES (Linguistics)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August 2011  © Kristín M. Jóhannsdóttir, 2011  Abstract This dissertation presents a semantic analysis of the progressive of both English and Icelandic, the only two Germanic languages that generally are considered to have fully grammaticalized progressive constructions. The progressive is an aspectual category where the focus is on a single, dynamic event being in progress at a certain time – the reference time. It is generally considered to be a sub-category of the imperfective aspect, just like the habitual aspect, and one of the descriptions typically given for the progressive is that it cannot have a habitual reading. Similarly, stative predicates are categorized as imperfective but non-progressive. Nevertheless, both habitual sentences and stative predicates occur in the progressive; they then appear to have a slightly different meaning from the one they have when they occur in the simple past/present. I argue that the subtle meaning difference between progressive and non-progressive statives and habituals is in fact an implicature. Stative verbs are shifted to being events in order to take on one or more of the prototypical eventive properties, and as events they can occur in the progressive. In such cases they usually imply dynamicity, control and/or temporariness. Habituals are essentially stative so when they occur in the progressive they too have been shifted to events, resulting in the same implicature of prototypical eventive properties, particularly temporariness. We then get the reading that the habit is temporary and it contrasts with the simple past/present that picks out a more general habit. Additionally I investigate another way to indicate that a series of events is in progress, namely the present participle progressive in Icelandic, which is a progressive construction with a presupposition for pluractionality. It usually occurs with iterative adverbials, in particular adverbs of quantity, which give additional information on the frequency of the series of events.  ii  Table of contents Abstract............................................................................................................................... ii Table of contents ...............................................................................................................iii List of tables .......................................................................................................................vi List of figures ................................................................................................................... vii Acknowledgements ......................................................................................................... viii 1  Overview and background ............................................................................................1 1.1 Introduction ................................................................................................................................1 1.2 Organization of the thesis.....................................................................................................2 1.3 Aspect ............................................................................................................................................3 1.3.1 A relation between times ......................................................................................................3 1.3.2 Imperfective aspect ...............................................................................................................5 1.4 The progressive ..........................................................................................................................9 1.4.1 The English progressive ........................................................................................................9 1.4.1.1 1.4.1.2  1.4.2  The origin of the English progressive ........................................................................................... 9 The use of the English progressive.............................................................................................. 11  The Icelandic progressive .................................................................................................. 14  1.4.2.1 The infinitival progressive ........................................................................................................... 15 1.4.2.1.1 The origin of the infinitival progressive.............................................................................. 15 1.4.2.1.2 The use of the progressive.................................................................................................... 16 1.4.2.2 Other progressive constructions in Icelandic............................................................................... 21 1.4.2.2.1 The present participle progressive........................................................................................ 21 1.4.2.2.2 The posture verb progressive ............................................................................................... 24  1.4.3 Conclusions .......................................................................................................................... 27 1.5 Theories of the progressive ................................................................................................. 27 1.5.1 Bennett and Partee ............................................................................................................. 28 1.5.2 Dowty ................................................................................................................................... 28 1.5.3 Parsons ................................................................................................................................. 29 1.5.4 Landman............................................................................................................................... 31 1.5.5 Portner ................................................................................................................................. 34 1.6 Aktionsart................................................................................................................................. 36 1.6.1 Aristotle and Kenny ........................................................................................................... 36 1.6.2 Vendler’s classification....................................................................................................... 36 1.6.3 The classification used in this dissertation ...................................................................... 38 1.6.4 Aktionsart and the progressive ......................................................................................... 38 1.7 Some final notes .................................................................................................................... 39  2  Stative verbs ................................................................................................................41 2.1 Introduction ............................................................................................................................. 41 2.2 The category of the stative .................................................................................................. 41 2.2.1 What constitutes a stative? ............................................................................................... 41 2.2.2 Stative tests.......................................................................................................................... 42 2.2.2.1 2.2.2.2 2.2.2.3 2.2.2.4  Progressive .................................................................................................................................... 42 Imperatives .................................................................................................................................... 44 Do something................................................................................................................................ 47 Occurrence in the complements of persuade and remind ........................................................... 49  iii  2.2.2.5 2.2.2.6 2.2.2.7 2.2.2.8 2.2.2.9 2.2.2.10  Occurrence with manner adverbs like deliberately, masterfully, etc......................................... 51 For phrases.................................................................................................................................... 52 Occurrence on either side of instead of........................................................................................ 52 Habitual meaning .......................................................................................................................... 53 Sequence of tense .......................................................................................................................... 57 The perfect................................................................................................................................... 61  2.2.3 The results of the tests....................................................................................................... 63 2.3 The difference between stative verbs and event verbs ............................................... 65 2.3.1 The Davidsonian and Neo-Davidsonian accounts .......................................................... 65 2.3.1.1 The Davidsonian account ............................................................................................................. 65 2.3.1.2 Neo-Davidsonian analysis ............................................................................................................ 67 2.3.1.3 In support of the Davidsonian argument ..................................................................................... 68 2.3.1.3.1 The modifier argument ......................................................................................................... 68 2.3.1.3.2 Perception verbs .................................................................................................................... 69  2.3.2  The Davidsonian argument: do stative verbs have it? ................................................... 69  2.3.2.1 Adverbial modification ................................................................................................................. 70 2.3.2.1.1 Manner adverbs...................................................................................................................... 70 2.3.2.1.2 Locatives................................................................................................................................ 77 2.3.2.2 Perception verbs ............................................................................................................................ 79 2.3.2.3 Anaphora........................................................................................................................................ 81 2.3.2.4 Results ........................................................................................................................................... 82  2.3.3 Kratzer’s analysis................................................................................................................ 83 2.4 Conclusion ............................................................................................................................... 85  3  States in the progressive ............................................................................................86 3.1 Introduction ............................................................................................................................. 86 3.2 The use of states in the progressive ................................................................................. 87 3.3 The semantics of states in the progressive .................................................................... 91 3.3.1 From lexical entry to time of evaluation ........................................................................ 91 3.3.2 Rothstein’s shift theory ..................................................................................................... 93 3.3.3 Coercion of states to events.............................................................................................. 95 3.4 Eventive implicatures ........................................................................................................... 96 3.4.1 Proto-Events ....................................................................................................................... 96 3.4.2 Eventive properties ............................................................................................................ 98 3.4.2.1 3.4.2.2 3.4.2.3 3.4.2.4 3.4.2.5  Control........................................................................................................................................... 98 Dynamicity .................................................................................................................................. 101 Temporariness ............................................................................................................................. 103 The implicatures ......................................................................................................................... 108 Results ......................................................................................................................................... 109  3.5 States in the progressive in Icelandic ........................................................................... 109 3.5.1 The project ........................................................................................................................ 109 3.5.2 Acceptability based on different categories................................................................... 110 3.6 Wh ere English and Icelandic differ ............................................................................... 113 3.6.1 Locatives and posture verbs ............................................................................................ 113 3.6.2 Weather verbs.................................................................................................................... 117 3.6.3 Be in the progressive......................................................................................................... 121 3.7 Conclusions ........................................................................................................................... 126  4  The progressive habitual ..........................................................................................128 4.1 Introduction ........................................................................................................................... 128 4.2 Habitual aspect ..................................................................................................................... 131 4.2.1 Habitual aspect vs. generics and iteratives..................................................................... 131 4.2.2 Habitual aspect in English and Icelandic ........................................................................ 133 4.2.3 Characteristics of the habitual aspect............................................................................. 135 4.2.3.1 4.2.3.2  Characteristic of an extended period of time ............................................................................. 135 General habit but not universal quantification .......................................................................... 136  iv  4.2.3.3 4.2.3.4  Event does not have to be ongoing at the reference time ......................................................... 137 Conclusion .................................................................................................................................. 137  4.2.4 Habituals as statives .......................................................................................................... 137 4.2.5 The semantics of habitual sentences .............................................................................. 139 4.3 The progressive habitual ................................................................................................... 141 4.3.1 Previous discussions of the progressive habitual........................................................... 141 4.3.2 Characteristics of the progressive habitual.................................................................... 143 4.3.2.1 4.3.2.2 4.3.2.3 4.3.2.4  Characteristic of an extended period of time ............................................................................. 143 General habit but not universal quantification .......................................................................... 144 Event does not have to be ongoing at the moment.................................................................. 145 Conclusions................................................................................................................................. 146  4.3.3 Temporariness and event time........................................................................................ 146 4.3.4 Do habitual events have stages?...................................................................................... 149 4.4 Formal analysis of the progressive habitual............................................................... 151 4.4.1 The semantics of the progressive habitual .................................................................... 151 4.4.2 The semantics of the habitual of the progressive ........................................................ 153 4.5 Gen eral conclusions ........................................................................................................... 154  5  The present participle progressive ..........................................................................155 5.1 Introduction ........................................................................................................................... 155 5.2 The present participle construction ............................................................................... 158 5.2.1 The present participle as a progressive construction .................................................. 158 5.2.2 The iterative nature of the present participle progressive ......................................... 160 5.2.3 The bare present participle construction ...................................................................... 165 5.3 The present participle vs. the progressive habitual .................................................. 168 5.3.1 Characteristic of an extended period of time................................................................ 168 5.3.2 General habit but not universal quantification .............................................................. 169 5.3.3 Event does not have to be ongoing at the moment..................................................... 169 5.3.4 Conclusion.......................................................................................................................... 169 5.4 Adverbs of quantity with progressive sentences ........................................................ 170 5.4.1 Adverbs of quantity .......................................................................................................... 170 5.4.2 The different behaviour of adverbs of quantity ............................................................ 171 5.4.2.1 5.4.2.2 5.4.2.3 5.4.2.4 5.4.2.5 5.4.2.6  5.4.3  Binding of indefinites (unselective binding) ............................................................................. 171 Preposing ..................................................................................................................................... 173 The adverbs in the progressive .................................................................................................. 174 In the scope of for-adverbials ..................................................................................................... 176 The ambiguity of alltaf .............................................................................................................. 177 Conclusion .................................................................................................................................. 180  The semantics of adverbs of quantity ............................................................................ 182  5.4.3.1 5.4.3.2  Adverbs of quantification ............................................................................................................ 182 Frequency adverbs....................................................................................................................... 183  5.4.4 General conclusions .......................................................................................................... 188 5.5 Pluractional presupposition ............................................................................................. 189 5.6 Conclusion ............................................................................................................................. 191  6  Closing Remarks ......................................................................................................193 6.1 6.2  Results ..................................................................................................................................... 193 Suggestions for further study .......................................................................................... 195  Bibliography....................................................................................................................197  v  List of tables Table 1:1: Progressive constructions of English and Icelandic. ............................................................................. 27 Table 1:2: The aspect classification used in this dissertation.................................................................................. 38 Table 2:1: Results of the stative tests......................................................................................................................... 64 Table 2:2: Arguments for an underlying state argument. ........................................................................................ 82 Table 3:1: Stative verbs in the first-person singular in COCA................................................................................ 88 Table 3:2 Event verbs in the first-person singular in COCA ................................................................................... 89 Table 3:3: Do posture verbs pattern with states or events?................................................................................... 117 Table 3:4: Do weather verbs pattern with states or events?.................................................................................. 121 Table 4:1: The co-occurrence of the progressive and the habitual ....................................................................... 131 Table 5:1: Adverbs of quantity with when-clauses ................................................................................................. 176  vi  List of figures Figure 1-1: Classification of the aspect system .......................................................................................................... 6 Figure 1-2: Landman’s (1992:27) continuation branch .......................................................................................... 33 Figure 4-1: Stages of the habit e’, ‘play well’......................................................................................................... 150  vii  Acknowledgements There are many people that I need to thank and I doubt I will manage to name everyone that made a positive impact on the writing of this dissertation. However, there are two people that I owe the greatest gratitude. These are my co-supervisors, Dr. Lisa Matthewson and Dr. Hotze Rullmann. Without them I would never have finished this huge project. Not only did they discuss with me every little detail, point me in the right direction when I was lost, and even suggest possible solutions, they also encouraged me throughout the whole project, and made me believe I could do this. If I were to mention them everywhere they contributed, their names would be the most common words in the dissertation. I also owe huge thanks to the third committee member, Dr. Gunnar Ól. Hansson, who gave me invaluable comments on the Icelandic data, in addition to raising many interesting questions that really made me think about the topic in a different way. Gunnar and his family have also remained dear friends throughout this whole process, which maybe says more than anything else. I would also like to thank the three external committee members, Drs. Paul Portner, Laurel Brinton and Michael Rochemont, who all contributed to making this thesis better. The audience at my defense also derserve thanks. As everyone knows, the thesis itself, although the final product, is only a part of the Ph.D. program and throughout the years at UBC I took many great courses from many fantastic professors. I would like to thank the amazing faculty at UBC and in particular Henry Davis, Rose-Marie Déchaine, Bryan Gick, Martina Wiltschko as well as Doug Pulleyblank who in addition to being my teacher also became a friend and a climbing buddy. Joe Stemberger and Patricia Shaw, although they never taught me, always had interesting things to say and for that I am thankful. The linguistic faculty at UBC cannot be thanked without special thanks going to Edna Dharmaratne who is the rock of the department. Masaru Kiyota and Shaine Mehji also deserve thanks for their hard work. When I first arrived in Vancouver I hardly knew anyone and so my fellow linguistics students became my social circle. Many became close friends, others good colleagues. My greatest thanks go to my best friends in the department, Marion Caldecott (my climbing buddy), Jeremy Perkins (my hiking buddy) and Leszek Barczak (my fellow European). I would also like to thank the students in my year, Diana Gibraiel, Yoko Ikegami, Noriko Yamane, Nahal Namdaran and James Thomson. Leora Bar-el was not only a fellow-student, floor-hockey buddy and squash-buddy, she also kept the social-life going her whole time in the department and is very much responsible for me having a great  viii  first year at UBC. Peter Jacobs taught me about his amazing culture but more importantly he showed me what a great speaker sounds like. Many thanks also to all the other linguistic students at UBC, Jeff and Clare Mühlbauer, Solveiga Armoskaite, Carrie Gillon, Amélia Reis Silva, Calisto Mudzingwa, Christiana Christodoulou, Donald Derrick, Heather Bliss, Jennifer Glougie, John Lyon, Mario E. Chávez Peón, Martin Oberg, Meagan Louie, Michael Barrie, Rachel Jarai, Sonja Thoma, Tyler Peterson, Ian Wilson, Beth Rogers, Carmela Irene Penner Toews, Hudu Fusheini, Mark Scott, Shujun Chin, Joel Dunham, Susan Jones, Ryan Waldie, Jason Brown, Yunhee Chung, Karsten Koch, Atsushi Fujimori, Olga Steriopolo, Oladiipo Ajiboye, and all the others. At UBC I learned the basics of elicitation and there I got the opportunity to work with two amazing sisters, Doreen Jensen and Barbara Sennott, who taught me about their language, their culture and life in general. Even though their language, Gitxsan, was not the topic of my thesis, I know that what I learned from them has made me a better linguist. I would never have been at UBC, studying linguistics, if it had not been for some amazing teachers back in Iceland who taught me about language and linguistics. These include high school teachers Valdimar Gunnarsson, Erlingur Sigurarson and late Gísli Jónsson, as well as all of my University of Iceland professors, but in particular Eiríkur Rögnvaldsson and Höskuldur ráinsson. During the last months of writing this thesis I was working for Eiríkur at the University of Iceland and without his generosity the final steps would have been a whole lot more difficult. In Manitoba two men encouraged me more than anyone else to continue my studies and do a Ph.D., Tim Schroeder and David Arnason. Thanks guys. Life in Vancouver was fortunately not just work, and many people helped make my life richer and more interesting. Without them it would have been a lot more difficult to sit in front of the computer and write. First I need to thank my Vancouver ‘mother’, Rosemary Payne and her family who always could lend an ear, whenever I needed, and who often invited me to dinner when I needed a break. Julianna Bjornson was an acquaintance from Winnipeg who became a dear friend in Vancouver. My VANOC buddies made life fun and interesting and I will never forget the adventure we shared. Liza Rogers, Elli Kim, Gus Labbé, Scott Marshall, Laetitia Sumison, Rina Mate, Erica Li, Maureen Sweeney, Sally Rehorick and so many others — I can’t name you all. This was one happy family. Finally, I have to mention my soccer buddies in Family Ties and Vancouver Presto and in particular coach Dave Sousa and his wife Lucy, as well as Benita Viera. You guys kept me in shape. In Italy Rut Valgarsdóttir was always ready with the right things to say, whenever I needed to vent, and in Iceland Gurún Helga Ívarsdóttir and Sigríur órardóttir have given me welcome breaks from the last steps of writing. My brothers, their wives and their kids remind me regularly of what life is all about.  ix  My final but dearest thanks go to my parents who not only have an endless belief in me but who are always there for me no matter what. I could not have done this with you.  x  1 Overview and background 1.1 Introduction In this thesis I look at the progressive constructions of two related languages, Icelandic and English, with a focus on the less studied uses of the progressive. These include stative verbs in the progressive and what I call the progressive habitual. In both languages the progressive is an imperfective construction, which ever since Reichenbach (1947) has been said to relate two different times, the reference time and the event time, where the reference time is inside of the event time.  (1)  John was eating when Paul came home.  Here the time of John eating his food is the event time (ET) and the moment when Paul comes home is the reference time (RT). As Paul comes home when John is in the middle of his eating, the time of Paul coming home, the reference time, falls inside the event time (see Klein 1994).  (2)  RT  ET  Two assumptions can be made on the basis of this description of the progressive; first, that only events can occur in the progressive (and not for instance states; see e.g. Poutsma (1926:339), Lakoff (1966, 1970), Visser (1973:1969)), and second, that the reference time is included in a time during which one relevant event takes place. Therefore, even if John eats three times during a particular day, the event time is not the time that it takes John to eat all his meals but only the time it takes him to eat the meal which he is in the middle of when Paul comes home. In this dissertation, however, I will show that neither of these assumptions is completely correct. Firstly, we do actually get verbs that usually are considered stative in the progressive:  (3)  I am really loving that song.  Secondly, even though usually the event time is the time during which one particular event takes place, it seems that it can also refer to an extended time during which many events of the same type take place, and which in that way resembles a habitual situation.  1  (4)  The Canucks are playing really well (these days).1  These additional uses of the progressive are what I will be focusing on in this dissertation. In addition I will look at the interaction between the progressive and temporal adverbials like always that seem to influence the use of the progressive construction, particularly in Icelandic, a language which has more than one way of showing that an event is in progress:  (5)  a. Jón er *borandi Jón is  /a bora.  eating (pres part) / to eat (inf)  ‘Jón is eating.’  c. Jón er Jón is  alltaf  borandi  /a bora.  always eat-ing (pres part) / to eat (inf)  ‘Jón is always eating.’  When we have one event in progress at the reference time in Icelandic, the infinitival progressive is used and the present participle progressive is not available, but when we have iterative events either progressive construction can be used. Lastly, even though the progressive works similarly in Icelandic as it does in English, there are some cases where it seems as it does not. For instance, the progressive is frequently used with weather verbs, posture verbs and locative verbs in English but not in Icelandic. Also, English has the so-called active be (Partee 1977) where the copula occurs in the progressive (John is being a fool). However, I will argue that this is a difference in how verbs are categorized in these two languages, rather than a difference between progressives.  1.2 Organization of the thesis In the remainder of this chapter I will discuss aspect in general and particularly imperfective aspect. Following that, in 1.4, I will discuss the progressive constructions of both English and Icelandic, including their origin and usage. In 1.5 the focus is on various theories of the progressive, such as those of Bennett and Partee (1972), Dowty (1979), Parsons (1990), Landman (1992) and Portner (1998). I will pay particular attention to Landman’s theory as that is the one I will be using in this dissertation. Finally, in 1.6 I will give a quick overview of Aktionsart.  1  The Canucks refers to the Vancouver hockey team Vancouver Canucks.  2  Chapter 2 is about stative verbs in general and particularly how they differ from event verbs. I will show that many of the tests that have been provided in order to distinguish them are rather testing for features like control. I will argue that this is not completely surprising as control is one of the eventive features not available to stative verbs. Other eventive features are dynamicity and temporariness. However, even though many of the tests fail as stative tests, three of them are reliable as tests for stativity. In chapter 3 I discuss stative verbs in the progressive and argue that in such cases the verb has been coerced to being eventive. Such coercion is usually brought on by the speaker’s intention to convey one of the prototypical eventive features, and the only way to do so is to shift the verb to becoming eventive. When the coercion has taken place, the verb acts like a regular eventive verb, implying at least one of the eventive properties. The focus of chapter 4 is on what I call the progressive habitual, which is when the progressive construction denotes a series of events. The progressive habitual shows some features of habitual aspect. I will argue that in such cases the progressive operator applies to a habitual sentence quantified over by the characterizing operator CHAR, which gives us the progressive of the habitual. This means that operators like PROG and CHAR can be stacked on top of each other, yielding sentences with complex aspectual structures. As habitual sentences are stative they need to be shifted to an event before PROG applies, and so the shift rule that shifts stative verbs to events also applies to states. This shows a clear parallel between states and habits in the progressive. In chapter 5 I discuss the present participle progressive in Icelandic and argue that it is in fact a pluractional progressive that indicates that a series of events is in progressive. The plurality is a presupposition and not a part of the truth conditions of the construction. Adverbs of quantity frequently occur with the present participle progressive, both frequency adverbs and adverbs of quantification, but are not necessary for the grammaticality of the present participle progressive, as long as the iterativity is somehow clear. Chapter 6 concludes the thesis.  1.3 Aspect 1.3.1  A relation between times  Comrie (1976:3) gives the following definition of aspect:  (6)  Aspects are different ways of viewing the internal temporal constituency of a situation.  He gives the following examples:  3  (7)  English: John was reading when I entered. Russian: Ivan ital, kogda ja voel. French: Jean lisait quant j’entrai. Spanish: Juan leía cuando entré. Italian: Gianni leggeva quando entrai.  In each sentence the first verb, read, provides the background to an event provided by the second verb, enter. The second verb shows an event in its totality without referring to the inner structure of that event, whereas the first verb makes explicit reference to an internal portion of the event. The first verb is usually said to be in the imperfective aspect while the second is in the perfective aspect. Comrie describes the difference between the two aspects in such a way that the perfective looks at the situation from outside but the imperfective from inside. Smith says that the “sentential aspect represents the speaker’s choice of perspective on the situation” (Smith 1983:479).2 Klein (1994) has pointed out that these explanations of aspect are not particularly precise and it can be hard to see whether the speaker is referring to a situation from the inside or from the outside. This can be particularly hard with non-dynamic situations. He gives the following pair (1994:29):  (8)  a. He aimed for a better solution. b. He was aiming for a better solution.  How can we establish the difference between these two sentences with the tools provided by Comrie, and even by Smith? In answer to that, Klein provides his own analysis, following Reichenbach (1947), who coined the terms event time, speech time and reference time. Reichenbach described event time as the time at which an event takes place (time of situation in Klein’s model; TSit) and speech time as the time at which the sentence is uttered (time of utterance in Klein’s model; TU). Reichenbach does not define reference time but it can be described as a reference point, which is provided by the context. In a sentence like (9) it is the past perfect that locates the event of Mary being ill before the event of Mary looking pale. Both events take place before the speech time (Klein 1994:25):  (9)  Mary looked pale. She had been very ill.  Using Klein’s analysis we can then order the events in the following way: 2  Notice here that Smith uses the term sentential aspect which includes both ‘situation aspect’ and ‘viewpoint aspect’. This distinction has been discussed by various linguists, such as Brinton (1988), Binnick (1991), Comrie (1995) and Bache (1997), to name a few. Smith’s ‘situation aspect’ is ultimately the same as ‘Aktionsart’ (see for instance Brinton (1988)), which is the term I will be using in this dissertation. It is discussed in section 1.6. 4  (10)  ET (Mary ill) < RT (Mary pale) < ST (sentence uttered)  However, Klein points out that times cannot always be defined in relation to an event, such as in (11), where Mary’s leaving the building cannot be situated in time in reference to another event.  (11)  At nine o’clock, Mary had left the building.  Instead of using reference time Klein therefore suggests calling it topic time (TT) and defines it in the following way: Topic time is “the time for which the particular utterance makes an assertion” (Klein 1994:37). Using this slightly modified version of Reichenbach’s analysis Klein argues that tense relates TT to TU but aspect relates TT and TSit.3 Furthermore, according to Klein’s analysis in the perfective TT includes TSit, the imperfective has TSit including TT and the perfect has TT following TSit.4 As it is the imperfective aspect, and particularly the progressive, that is the topic of this dissertation, we will not further discuss the perfective or the perfect but turn to the imperfective.  1.3.2  Imperfective aspect  Comrie (1976:24) defines imperfective aspect as an “explicit reference to the internal temporal structure of a situation, viewing a situation from within”, but as discussed in the previous section that may be a somewhat simplified description. Following Klein we can say that the imperfective really means that the time of the situation includes the topic time. Some languages have a single category to express imperfectivity, in some languages it is subdivided into distinct categories, whereas in others there are some categories that correspond to only part of the meaning of imperfectivity. Comrie (1976:25) provides the following classification:  3  I will use Klein’s relations of time in this dissertation but will be using the terminology of Reichenbach.  4  Lucko (1995) points out that Klein’s system fails to account for the perfect progressive which Klein claims has TT after TSit. It is hard to explain how that can be the case if the progressive has TT included in TSit and the perfect has it following TSit. As the perfect progressive is not my focus in this dissertation I will overlook this problem.  5  Figure 1-1: Classification of the aspect system  English has no special imperfective form but does mark the progressive, (12a), specifically. It also has a way to mark the habitual aspect, (12b), even though more often habituals are simply represented with the verb in the simple past or present. In the present tense, if the verb is eventive, such sentences are purely habitual, (12c), but in the past tense they are ambiguous between being habitual and episodic, (12d):  (12)  English a. John was working (when I entered).  (Progressive)  b. John used to work here.  (Habitual)  c. John works here.  (Habitual)  d. John worked here.  (Habitual/Episodic)  Just like English, Icelandic marks the progressive and the habitual separately:  (13)  Icelandic a. Jón var a vinna (egar ég kom). Jón was to  (Progressive)  work (when I came)  ‘Jón was working (when I came)’  b. Jón vinnur  hérna.  (Habitual)  Jón works (1sg pres) here ‘Jón works here.’  c. Jón vann  hérna.  (Habitual/Episodic)  Jón worked (1sg past) here ‘Jón worked here.’  Icelandic has another construction it often uses to mark the habitual – the vera vanur a ‘be used to’ construction:  6  (14)  Jón var vanur a bora hafragraut á morgnana. Jón was used to eat  oatmeal  (Habitual)  in mornings.the  ‘Jón used to eat oatmeal in the mornings.’  The vera vanur a construction is not quite equivalent to the English be used to and cannot always be used in the same contexts. I will discuss vera vanur a in chapter 4. Spanish has both an imperfective form and a progressive form, where the progressive is optional, as the imperfective does not exclude the progressive reading.  (15)  Spanish (Comrie 1976:25) a. Juan llegó.  (Simple past)  Juan arrive (past) ‘John arrived.’  b. Juan llegaba.  (Imperfective)  Juan arrive (imperf) ‘John was arriving/John arrives.’  c. Juan estaba  llegando.  (Progressive)  Juan be (imperf) arrive (imperf) ‘John was arriving.’  Sentence (15a) is episodic, (15b) can be either habitual or progressive and (15c) is progressive. Many languages have one special imperfective form that includes both the progressive and the habitual reading (examples from Comrie 1976: 26):  (16)  a. Il lisait Le Monde.  (French)  ‘He was reading/used to read Le Monde.’ b. On ital Pravdu.  (Russian)  ‘He was reading/used to read Pravda.’ c. etee Rabotniesko delo.  (Bulgarian)  ‘He was reading/used to read Rabotniesko delo.’  7  Kranich (2010:30) gives an excellent overview of the difference between general imperfective and progressive markers: The term ‘progressive’ is generally used to refer to a subtype of imperfective. A good definition, which reflects a widely accepted view of the term, comes from Bybee and Dahl who define the progressive as “indicating the situation is in progress at reference time” (Bybee and Dahl 1989:55). Put differently, progressive aspect “refers to the combination of (non-habitual) Imperfective aspect with dynamic (as opposed to stative) semantics” (Comrie 1995:1245). Prototypical progressives thus are used to refer to dynamic situations only. Furthermore, being dynamic, they are generally connected to limited duration (…), not to permanent states of affairs. In fact, Dahl (1985:93) has pointed out that habituals do not in general occur in the progressive and Poutsma (1926), Lakoff (1966, 1970), Visser (1973) and others have argued that neither do statives. Both, however, do occur in the imperfective in languages like Spanish that have a dedicated imperfective form. Yet, as has been discussed by many (such as Hirtle (1967, 2007), Mufwene (1984), Brinton (1988), Hirtle and Bégin (1991), Gachelin (1997) Kakietek (1997), mieciska (2002/2003), ráinsson (2005), Torfadóttir (2004)) stative verbs can be used in the progressive in both English and Icelandic, and in both languages the progressive can be used with certain habitual sentences (see for instance Hirtle (1967), Leech (1971), Palmer (1988)). We might be dealing with a case of language change. Kranich (2010) discusses the question whether English is shifting its progressive construction to becoming a general imperfect and points to Bybee and Dahl (1989) and Heine (1994), who in recent studies have shown that there is a common tendency for the progressive to develop into a more general imperfective. The intermediate stage in the change is that the progressive markers get used with categories that usually are more associated with the imperfective, such as stative verbs. Kranich (2010) has done an extensive corpus-based study of the progressive, looking at the development of the progressive in Modern English. Her research shows that there is an increase in the use of the progressive and that there is a certain rise in what she calls ‘derived meanings’, which are readings of the progressive other than the basic reading that an event is in progress at the reference time. However, she argues one should remain cautious in saying that the construction is developing into a general imperfective marker (Kranich 2010:172). Kranich points out that this does not mean that the progressive is not going through a change, and refers to Goossens (1994:165), who does in fact claim that the English progressive has broadened its functional scope enough to be seen as an imperfective construction even though he recognizes that it is only a partial realization of the imperfective. The fact that the progressive still has strong connections with dynamicity is one of the indicators that the change has not yet been completed. This is something that Comrie (1976) also points out, as lexically stative verbs get a dynamic reading when used in the progressive and Kranich (2010) argues that the fact that the progressive still refers to temporary situations when used with statives and habituals indicates that it is not yet an imperfective construction. However,  8  even if there is an ongoing language change, that does not solve the problem of how to represent this in the mental grammar of an individual speaker. That is what I will attempt to do in this dissertation.  1.4 The progressive 1.4.1  The English progressive  The progressive is formed in English by taking the verb be plus the main verb in the present participle, which means that the suffix –ing is attached to the root of the verb:  (17)  Progressive: be + V-ing  Icelandic and English are both Germanic languages and therefore have the same ancestor. By looking at the origin of the progressive constructions in each language we can see the connection between the English progressive and the Icelandic present participle progressive, the less used progressive construction in Icelandic. I will therefore give a quick overview of the origin of the English progressive and in section 1.4.2 I will give information on the Icelandic progressive constructions.  1.4.1.1 The origin of the English progressive The BE-verbs beon, wesan and weoran were used in Old English with V-ende to indicate ongoing action, as well as to provide a frame of reference for some other activity (Traugott 1992:187).5 This BE + ende construction, Traugott says, was fairly uncommon and largely restricted to event verbs and in that way the construction was similar to that of the Modern English V + ing. 6,7 However, Traugott points out that not all Old English –ende constructions can be translated into Modern English with be + ing giving the following example (Traugott 1992:187):  5  The similarity of the English and Icelandic progressive is particularly clear when one looks at Old English. The verb wesan is cognate with the Icelandic vera (< Old Norse vesa) and the participle ending – ende is of the same origin as the Icelandic –andi. 6  She calls them activity verbs.  7  For discussions of how the ending –ende later became –ing see particularly Traugott (1992:189-190) and Fischer (1992:252-253) as well as the linguists they reference. 9  (18)  Europe hio ongin … of Europe she begins … irnende of  Danai ære ie,  from Don nordæle …  that  seo is  river, that is  &  seo  running from northern-part … and that suryhte  on westhealfe  Alexandres  ea  Danai irn onan  river Don  runs thence  herga  Due-south into western-part Alexander’s  kingdom  ‘Europe begins… at the river Don, which runs from the North…and the river Don runs thence due South into the Western part of Alexander’s kingdom.’  The BE + ende construction was particularly favoured by verbs that denote activities without an inherent beginning or end. Traugott (1992:187) cites verbs like wunian ‘live’, faran ‘go’, cwean ‘speak’, feothan ‘fight’, libban ‘live’, and growan ‘grow’. She points out that, interestingly enough, the construction at that point already had the reading of an action that continued through a limited period of time (see discussion of temporariness in chapter 3, section 3.4.2.3). Fischer (1992:254) points out that unlike in Modern English the use of the progressive in Old and Middle English was optional and that it was not until the modern period that the progressive became grammaticalized and therefore a part of the aspectual system of English. Like Traugott and Fischer, Rissanen points out that the progressive was still unsettled in the sixteenth and seventeenth centuries. He gives the following pair of examples from two Shakespeare plays (Rissanen 1999:216; my italics):  (19)  a. What do you read, my Lord? (Hamlet II.ii)8 b. What are you reading? (Troilus and Cressida III.iii)  However, Rissanen (1999:216) says that the progressive “can be regarded as a grammaticalized aspectual indicator in the verbal system by 1700” and that the “set of the progressive forms in all tenses, active and passive, is fully developed around the end of the eighteenth century”. Denison (2000) puts the full grammaticalization of the progressive even later, in the late eighteenth century because of, among other reasons, the fact that the progressive passive developed only later.9 English and Icelandic are the only two Germanic languages with a grammaticalized progressive so the development of the progressive in these two languages is quite interesting for other languages with a less developed progressive constructions. For instance, when a certain contruction has been grammaticalized as a progressive construction, is it more likely that states and habituals start occurring in 8  In the same article Rissanen gives this example as: What doe you reade my Lord. (Rissanen 1999: 221)  9  Thanks to Laurel Brinton for pointing this out to me. 10  that construction as well? That is not the topic of this dissertation but is, nevertheless, interesting to keep in mind when we examine states and habituals in the progressive in Icelandic and English.  1.4.1.2 The use of the English progressive Before going any further, let us look at how the progressive construction is used by English speakers. I will start by discussing the most straightforward progressive meaning (that an event is in progress) before discussing some uses that seem to be of a different nature. The most common use of the progressive construction is to indicate that a certain action or event is taking place at a certain reference time. In the present progressive that reference time is the time of speech (now) but in the past tense the reference time is either clearly stated in the sentence or is clear from context.  (20)  a. John is singing (at this very moment). b. Peter was eating an apple (when Paul entered the kitchen). c. A: Did you see Anne at all? B: No, she was playing golf.  The past progressive sentences in (20b) and (20c) require the reference time either to be stated in the sentence, (20b), or made clear some other way by context, (20c). So with the reference clause in (20b) the sentence could be uttered out of the blue, but without it the listener needs to have been given the reference time previously or might be prompted to ask: ‘When?’ Partee (1973) claimed that referential past tense sentences are like pronouns and therefore require some way of being specified either by context or explicit marking. However, the ‘when’-effect of the progressive is much stronger than that of the simple past. That might be the result of the imperfective, which requires the reference time to be included in the event time. Without the when-clause in (20b) and the context sentence in (20c) the reference time is not given and the hearer is left with the information that an event was in progress at a particular but unknown time. In the simple past, on the other hand, the event time is inside the reference time and so the hearer understands the sentence to mean that there was an event which took place in the past. Another idea is that this is because the progressive is not a pure temporal construction but one that usually connects the situation referred to by the verb to another situation, whether that is an event or time. So, the simple past, for instance, means that at some point in the past a certain event took place. The progressive, on the other hand (in the past) says that at some point in the past at a particular time or when some event took place, another event was in process.  11  In addition to the ‘in progress’ reading of the progressive it also seems to have what Kranich (2010) calls ‘derived meanings’. Of those, only one will be discussed in detail in this thesis – the habitual reading (chapter 4), where repeated events are occurring over a long period of time: a. The Whitecaps are playing their games at Empire Field.10  (21)  b. Elli is applying for jobs.  There are other readings of the progressive that I will only mention here but not discuss in any detail. Firstly there is the future reading of the progressive where the progressive is used to refer to imminent or future actions:  (22)  a. Scott is moving to the USA. b. Are you going to the game tonight? c. I am coming. Just need to grab my jacket first.  None of the actions are taking place at the speech time but they are expected to happen in the future. Notice that the sentence in (22a) could be uttered even if Scott was not moving to the USA until three months from now, for instance. This becomes particularly clear if the relevant context is also in the distant future:  (23)  A: Is Scott going to attend the 1-year-return party next February? B: No, he will not. He is moving to the USA.  Basically, the progressive can be used if an event is planned, whether it happens in the near future 11  or not. Leech et al. (2009) have pointed out that the progressive often denotes a situation which may not have happened yet, but has been planned and therefore may be seen as being in progress. However, I do not believe that in examples like (23), for instance, we can say that Scott’s move is in progress even if it may be decided. I suggest the future progressive has more to do with planned future than the future being in progress. 10  The Whitecaps is the professional soccer team in Vancouver.  11  The reference time in (23) is next February which could be a month from the speech time or could be eight months from the speech time. All that matters is that Scott will have moved to the USA before that time, and that his move has already been planned. This could be called ‘planned future’ (see for instance Copley (2004, 2005)). Notice that this future meaning can also occur in the past tense: (i) John could not help his mother with the dishes because he was going to the Canucks game.  12  In fact, the progressive cannot be used for future events that cannot be planned:  (24)  *The Canucks are winning tomorrow.  If someone wants to predict that the Canucks will win, or if he can see the future he has to use the future auxiliary will:  (25)  The Canucks will win tomorrow.  Another ‘derived meaning’ are repetitive actions with a hint of disapproval, which we get when adverbs like always, continually, etc. are used with the progressive (e.g. Palmer 1974:64).12  (26)  a. I was continually falling ill. b. They were forever leaving the gate open. c. He is always asking silly questions.  There may be some truth to Palmer’s claims that there is a hint of disapproval in these sentences. However in that case it may be wise not to use predicates that on their own represent something undesirable such as ‘falling ill’, ‘leaving the gate open’ or ‘asking silly questions’. It is better to show examples with more positive predicates because if those still get a negative connotation in the progressive we are in much better position to say that the negative reading has something to do with the progressive. Is there, for instance, the same hint of negativity in (27b) as there is in (27a)?  (27)  a. The boys are always teasing the girls at school. b. The boys are always kissing the girls at school.  On its own, and without context, I do not find (27b) as negative as (27a) but it is true that there could be a negative connotation to (27b), as if the girls did not in fact want the boys to kiss them or if the kissing somehow created problems. This may, however, depend on context. If we imagine a situation where kids in day-care are being discussed and someone asks the day-care worker if there is any problem between the boys and the girls, (27b) would probably be seen as quite positive. If however the same question was asked of a Grade 9 teacher we would be more likely to get that hint of negativity. The same can be said about (28). Coming from a proud parent the sentence is positive, whereas said by a 14 year old of a schoolmate that might not be the case. 12  For a more detailed discussion in this hint of disapproval with progressive sentences, see Kranich (2010). 13  (28)  She is always studying, that girl.  (28), as well as (26) and (27), are examples of the progressive habitual which I will discuss in chapter 3. I do believe that sentences in the progressive do often have a hint of negativity but that it might not necessarily result from the progressive itself but could as well come from other expressions or context. Achievements are generally said not to occur in the progressive (see e.g. Dowty 1979) and examples such as these ones from Rothstein (2004:36) are given to show that:13  (29)  a. #Jane is reaching the summit of the mountain.14 b. #Mary is spotting her friend at the party.  One of the possible explanations for this is that achievements are momentaneous and the reference time, therefore, cannot be included in the event time. However, the fact is that achievement verbs sometimes do occur in the progressive as the following examples from Rothstein (2004:36) show:  (30) a. Fred and Susan are finally leaving. b. The plane is landing.  Rothstein claims that these sentences have “normal” progressive readings, meaning that they do not appear to differ from accomplishments in the progressive. In fact, Rothstein argues that when achievement verbs occur in the progressive they have been shifted to becoming accomplishments. I will discuss Rothstein’s theory in detail in chapter 3, section 3.3.2.  1.4.2  The Icelandic progressive  Although the Icelandic language has more than one way to mark that an event is ongoing at a particular time, the most common one, and the only one that is fully grammaticalized, is what I call the infinitival progressive. It has the verb vera ‘to be’ coupled with an infinitive marker and the main verb in the infinitive. I will, therefore, start by looking at that construction in detail, but I will also give an overview of two other constructions, the present participle progressive and the posture verb progressive.15 13  For discussion of achievements and other Aktionsart see section 1.6.  14  Although Dowty does recognize the fact that there are exceptions to this.  15  ráinsson (1974:23) discusses in his M.A. thesis what constructions really should be considered progressive and names sentences like hann svaf, egar ég kom ‘he slept (simple present), when I came’, hann var sofandi, egar ég kom ‘he was sleeping (present participle), when I came’, hann sat og las, egar ég kom ‘he sat and read, when I came’, hann sat a snæingi, egar ég kom ‘he sat at dinner, when 14  1.4.2.1 The infinitival progressive 1.4.2.1.1  The origin of the infinitival progressive  The historical origin of the infinitival progressive construction, vera a + infinitive, is the construction vera a a + infinitive (Smári (1920), Benediktsson (1976), ráinsson (1974, 1999)) where the first a is a locative preposition (or possibly an adverb), but the second one is the infinitival marker.16 The following examples are from the 13th century text Njáls saga:17  (31)  a. Hann var a a hlaa skútuna He  was at to load vessel.the  ‘He was loading the vessel.’  b. Bar  a saman  og á  var Gunnar a a segja söguna  Happened that together and then was Gunnar at to Kári hlddu  til á  tell  en  eir  story.the but they  mean úti.  Kári listened to on while  outside  ‘At the same time Gunnar was telling the story but Kári and the others listened outside.’  Smári (1920:181) and Benediktsson (1976) have pointed out that this prepositional construction existed alongside the simpler version (our current construction) with only the infinitival marker. As the preposition and the infinitival marker are identical it is not possible to be certain which one was dropped, but Icelandic linguists generally believe that it was the preposition that dropped (see e.g. Smári (1920), Benediktsson (1976), ráinsson (1999)). However, Bertinetto et al. (2000:522) categorize Icelandic with languages such as Breton, Dutch, Frisian, German, Italian and Portuguese in combining the copula with a I came’, au voru í háarifrildi, egar ég kom ‘they were in an argument, when I came’. I discuss two of these constructions in this dissertation, the present participle and the posture verb construction. The last two, however, are quite different as there you cannot substitute different verbs and get progressive reading with them all. Nevertheless, ráinsson is correct in that these sentences really do seem to indicate an event in progress and it would be interesting to see whether there are other ways to indicate an event in progress. 16  Middle English had a similar construction with the preposition on followed by V+ing. Visser (1973:1993) says that there were actually three forms “vying with each other for the hegemony”. These are the old present participle he is huntende, the previously mentioned form with the preposition, he is on (a) hunting and the Modern English progressive construction he is hunting. Rissanen (1999:217) also discusses the construction with the preposition and gives the following example: I am on reading. Thanks to Laurel Brinton for pointing this out to me. 17  The example in (31a) is given in ráinsson (1999:219), but according to ráinsson it was Nygaard (1905:225) who originally referred to it in the discussion of the Icelandic progressive.  15  prepositional phrase, instead of categorizing it with Estonian, Finnish, Karelian, Sami, Livonian and Vepsian which combine a copula with an infinitive or related form. Comrie (1976) also analyses the a as a preposition rather than the infinitival marker.  1.4.2.1.2  The use of the progressive  Just like in English, the most common use of the Icelandic progressive is to indicate that a certain action or event is taking place at a certain reference time.18 In the present progressive that reference time is the time of speech (now) but in the past tense the reference time is either clearly stated in the sentence or is made clear by context. In (32b) the specified reference time is the time when Páll came and at that time Jón was eating an apple.  (32) a. Ég er I  a skrifa.  am to write  ‘I am writing.’  b. Jón var a bora epli Jón was to eat  egar Páll kom.  apple when Páll came  ‘Jón was eating an apple when Páll came.’  Just like in English, the infinitival progressive also allows other meanings. I will give a short overview of these without going into much detail. First, let us look at the progressive habitual. What I call the progressive habitual is a progressive construction used with habitual sentences and the reading is that a habit is in progress. These are sentences like the following:  18  ráinsson (1974:46) points out that many linguists, such as Nordlin (1928:403), Einarsson (1967:145), and Kress (1963:202 ff), look at the construction fara a + V ‘go to + V’, which Einarsson (1967:145) calls ‘beginning action’, as ‘connected’ to the progressive construction vera a + V ‘be at + V’. He therefore suggests that it can be used as a test: If the vera a-sentence can also be used with fara a, then it has a true progressive meaning. He gives examples of hann fer a bora – hann er a bora ‘he goes to eat – he is eating’ and hann fer a tala – hann er a tala ‘he goes to talk – he is talking’ where we have the beginning of the event and then the progressive of the event. He claims that vi ví er a búast ‘with that is to expect’ or hva er a frétta ‘what is to hear’ are not progressive sentences even though they use the vera a construction as they are ungrammatical with fer a: *vi ví fer a búast ‘with that goes to expect’ and *hva fer a frétta ‘what goes to hear’. Interestingly enough we can say á fór ég a elska etta lag ‘then went I to love this song’ and á fórum vi a halda a ú kæmir ekkert ‘then went we to think you were not coming’. So if this test really does show whether a sentence has a progressive reading or not the stative verbs in the vera a construction actually do pass the test. However, I would need a much more thorough examination of the vera a + V construction as well as the fara a + V construction before I can judge whether this really is a valid test or not. 16  (33) a. órsarar eru a spila vel.19 órsarar are to  play well  ‘órsarar are playing well.’  b. Jón er Jón is  a hjóla í  vinnuna essa dagana.  to bike to work.the these days.the  ‘Jón is biking to work these days.’ Here we do not have one ongoing event in progress but a series of events. ráinsson (2001) calls sentences like (33a) íróttamál ‘sport language’ as it has become very common amongst athletes and sports writers. These types of sentences will be discussed in detail in chapter 4. Other readings are also available for the infinitival progressive in Icelandic. First we have the future meaning just like in English. The infinitival progressive construction can be used to indicate that an event will take place at some point in the future:  (34) a. Ég er I  a flytja til Íslands.  am to  move to Iceland  ‘I am moving to Iceland.’  b. Sigur Rós er a spila á morgun. Sigur Rós is to  play on morning  ‘Sigur Rós is playing tomorrow.’ Also, the progressive cannot be used for future events that cannot be planned:20  (35) *órsarar eru a vinna á morgun. órsarar are to  win  on morning  ‘órsarar are winning tomorrow.’  Here we would need the future auxiliary munu ‘will’, although the simple present would also do:  19  ór is a sports club from Akureyri, Iceland. órsarar is a common name used for both fans of the club as well as the athletes that play for it. 20  We also cannot say a er a rigna á morgun ‘it is raining tomorrow’, but as will be discussed in chapter 3, weather verbs never occur in the progressive in Icelandic so we cannot really use this as an example for the future reading of the progressive.  17  (36)  órsarar munu vinna á órsarar will  win  morgun.  on morning  ‘órsarar will win tomorrow.’  (37)  órsarar vinna á morgun. órsarar win  on morning  ‘órsarar will win tomorrow.’  In this way the future progressive marks a planned future, just like in English. The Icelandic infinitival construction not only can mark that an event is about to happen, but also that one has just happened.21  (38)  a. Ég var a bora. I  was to  eat  ‘I just finished eating.’  b. Ég var a hlaupa. I  was to  run  ‘I just finished running.’  Here we obviously do not have a progressive reading even though the progressive construction is being used. In section 1.6.4 I mentioned that achievements are generally said not to occur in the progressive and that one of the possible explanations for this is that they are momentaneous and their reference time, therefore, cannot be included in the event time. However, Rothstein has claimed that achievements in English can be coerced to being accomplishments and as such do occur in the progressive. We also get achievements in the progressive in Icelandic but then they get the reading that an event is about to happen or that it just happened. In this case we can say that the progressive has an immediate future/past meaning. The following sentences are from ráinsson (2001:248).  (39) a. Barni  var (alveg) a detta egar ég greip  Child.the was (just)  to fall  í  a.  (event just about to happen)  when I grabbed at it  ‘The child was about to fall when I grabbed it.’ 21  Interestingly, additional stress on the copula can influence the closeness of the event. Gunnar var a fara ‘Gunnar was to go’, without any additional stress, has the matter-of-fact reading that he just left. However, Gunnar VAR a fara, with additional stress on the copula, focuses on the fact that it just happened moments ago, for instance if the speaker wants to stress that the hearer just missed Gunnar. 18  b. Ég var rétt a loka bókinni egar hún kom.  (event just happened)  I was just to close book.the when she came ‘I had just closed the book when she came.’  It is interesting that we get the about-to reading in (39a) but in (39b) we get the reading that something just happened. Here the adverbs alveg and rétt might influence the reading. Both are translated as ‘just’ in the data but they do have a fine meaning difference – we use alveg when the event is about to happen but rétt when it just did. This is clearer when we use the same verb:  (40) a. Ég var rétt a ná  takmarkinu.  (event just happened)  I was just to reach goal.the ‘I just reached the goal.’  b. Ég var alveg a ná I was just  takmarkinu  (event about to happen)  to reach goal.  ‘I was just about to reach the.’  These adverbs are not necessary and we can get both readings without them:  (41) a. Ég er I  glö núna ví  ég  am happy now because I  var a ná  takmarkinu.  (event just happened)  was to reach goal.the  ‘I am happy now because I just reached the goal.’  b. Ég var a ná I  was to  takmarkinu egar tíminn  reach goal.the  rann út.  when time.the ran  (event about to happen)  out  ‘I was just about to reach the goal when the time ran out.’  The when-clause seems to influence the reading of the sentence somewhat. The sentence in (41a) has no when-clause and it clearly has the reading of the event just having taken place. In (41b) we have a whenclause and here not all Icelandic speakers agree on the meaning of the sentence. Some say that it can only have the ‘about-to’ reading, whereas others accept both the ‘about-to reading’ and the ‘just happened’ reading. Gunnar Ól. Hansson (p.c.) has suggested that the ‘just happened’ reading appears when the speech time is the only possible reference time, and as the sentence is in the past tense it has to get the reading that something just happened. When another clause, such as a when-clause, provides the reference time the natural reading is that the event was about to happen. Obviously more than one factor  19  influences the reading of these achievement verbs in the progressive and further study is needed in order to provide a proper analysis. Just like in English, Icelandic sentences in the progressive sometimes indicate a hint of disapproval. ráinsson (1999:216) calls this háttarmerking ‘manner meaning’ as the focus is not on an event being in progress but rather there is some kind of emotional emphasis (example from ráinsson 2001:248):  (42)  Hann var ekkert a hanga yfir He  was not  to hang  ví.  over it  ‘He did not bother spending time on it.’  Stative verbs like eiga ‘own’ and búa ‘live’ do not in general occur in the progressive in Icelandic. However, interestingly enough, when they do they often get this emotion emphasis: (43) a. *Hann er a eiga hund.22 He  is to own dog  ‘He is owning a dog.’  b. Hva What  er hann líka a eiga hund!23 is he  also to  own dog  ‘What is he thinking anyway, owning a dog!’  (44)  a. *Hann er a búa í He  is to  Reykjavík.  live in Reykjavík  ‘He is living in Reykjavík.’  b. Ég er I  á  ekkert a búa  am then not  to live  lengur í  Reykjavík fyrst ég kemst ekki í  longer in Reykjavík since I get  not  in  borgarstjórn. city.council ‘I am not going to be living in Reykjavík anymore, then, since I cannot be on the city council’  22  A very similar example is originally given by ráinsson (1999:217).  23  This might be a specialized construction in the style of Kay and Fillmore’s (1999) What’s X doing Y. 20  A similar emotional emphasis can also appear with verbs that regularly do occur in the progressive:  (45)  a. María var a hjálpa mér vi María was to help  heimaverkefni  egar ú hringdir.  me with home.assignment.the when you called  ‘María was helping me with the home assignment when you called.’  b. María María  var ekki miki a hjálpa mér vi was not  much to help  heimaverkefni!  me with home.assignment.the  ‘María did not bother helping me with the home assignment!’  However, just as previously mentioned for English, I am not certain that this emotive meaning comes from the progressive. Instead it is quite possible that the use of the adverbs influences that reading. For more detailed discussions of the vera a + V construction in Icelandic see ráinsson (1974, 1999, 2001) where he discusses the question whether this construction should even be given one common name (such as framvinduhorf ‘the progressive’) or whether we should look at it as having many different functions.24  1.4.2.2 Other progressive constructions in Icelandic 1.4.2.2.1  The present participle progressive  Just like the English progressive is built from the verb be plus the present participle ending –ing on the main verb, Icelandic has a similar construction with the copula vera ‘be’ plus the present participle ending –andi attached to the main root. However, its use is quite limited as can be seen in the following examples:  (46)  a. Jón er sitjandi. Jón is sitting (pres part) ‘Jón is sitting.’  b. Ása er sofandi. Ása is sleeping (pres part) ‘Ása is sleeping.’ 24  For a detailed discussion of the construction vera+a see ráinsson (1974), and particularly (1974:2641) for an overview of what linguists said about the construction before 1974. Later discussion can also be found in ráinsson (1999) and ráinsson (2001).  21  (47)  a. *Jón er borandi. Jón is eating (pres part) ‘Jón is eating.’  b. *Ása er horfandi  á sjónvarp.  Ása is watching (pres part) on television ‘Ása is watching television.’  Here sitja ‘sit’ and sofa ‘sleep’ are fine in the present participle, but bora ‘eat’ and horfa ‘watch’ are not. Notice that we get completely opposite results in the infinitival progressive:  (48)  a. *Jón er a sitja. Jón is to sit (inf) ‘John is sitting.’  b. *Ása er a sofa. Ása is to sleep (inf) ‘Ása is sleeping.’  (49)  a.  Jón er a bora. Jón is to eat (inf) ‘Jón is eating.’  b.  Ása er a horfa  á sjónvarp.  Ása is to watch (inf) on television ‘Ása is watching television.’  Based on this limited data the present participle seems to be in complementary distribution with the infinitival progressive. Eventive predicates such as bora ‘eat’, lesa ‘read’ and horfa á sjónvarp ‘watch television’ can only be used in the infinitival progressive and not in the present participle. Positionals, such as sitja ‘sit’, liggja ‘lie’, standa ‘stand’, etc., as well as sofa ‘sleep’ and vaka ‘be awake’, occur in the present participle but not in the infinitial progressive.25 As for stative verbs, they can occur in the infinitival progressive, as will be discussed in chapter 2, but not in the present participle progressive: 25  It is worth pointing out that when vaka ‘be awake’ has an eventive meaning, as in vaka eftir ‘stay awake for’ or vaka yfir ‘stay awake over’, as in (i) and (ii), then the infinitival progressive is used:  22  (50)  a. *Jón Jón  er elskandi  Siggu.  is loving (pres part) Sigga  ‘Jón is loving Sigga.’  b. *Jón er vitandi  svari.  Jón is knowing (pres part) answer.the ‘John is knowing the answer.’  Why is it that the present participle can occur with a small subset of roots and then have a progressive reading? One possible explanation is that in the present participle examples above we do not have the suffix –andi attached to a verb but an adjective. Jón er sofandi ‘Jón is asleep’ is then similar to Jón er reyttur ‘Jón is tired’ but not to the English progressive ‘Jón is sleeping’. If that is true then the examples in (46) should not be said to include a special progressive construction.26 So why a special subsection about the present participle progressive? Look at the following examples:  (51)  a.  Jón er alltaf  borandi.  Jón is always eating (pres part) ‘Jón is always eating.’  b. María er alltaf lærandi. María is always studying (pres part) ‘María is always studying.’  Here we have the event verbs bora ‘eat’ and læra ‘study’ with the present participle suffix, as well as the adverb alltaf ‘always’, and not only are the sentences perfectly fine, but they give us a progressive reading. In fact, the meaning is exactly the same as we would get if we used the infinitival progressive:  au eru a vaka /*vakandi eftir táningsdótturinni. They are to wake (inf) /wake (pres part) after teenage.daughter.the ‘They are staying awake (waiting) for the teenage daughter.’ (ii) au eru a vaka/*vakandi yfir veiku barninu. They are to wake (inf)/wake (pres part) over sick child.the ‘They are staying awake (while watching) over the sick child.’ This indicates that the infinitival progressive is the regular progressive construction with an event whereas the present participle is more commonly used with stative sentences. (i)  26  We might then instead want to ask us why Icelandic uses adjectives to indicate posture but English uses verbs. However, it is not surprising for languages to differ in that way and that question will have to wait future research. 23  (52)  a.  Jón er alltaf  a bora.  Jón is always to eat (inf) ‘Jón is always eating.’  b.  María er alltaf  a læra.  María is always to study (inf) ‘María is always studying.’  It appears that we have here an example of the present participle with event verbs giving us a progressive reading. Notice that here the posture verbs and other stative verbs are ungrammatical, just as they are in the infinitival progressive:  (53)  a. *Jón er alltaf Jón  a sitja.  (positional verb, inf.)  is always to sit (inf)  ‘Jón is always sitting.’  b. *Jón er Jón is  alltaf a elska always to  /elskandi  Siggu.  (stative, inf/pres part)  love (inf) /loving (pres part) Sigga  ‘John is always loving Sigga.’  In chapter 4 I will look closely at the present participle progressive.  1.4.2.2.2  The posture verb progressive  It is quite common that morphemes that carry posture meanings can also be used with grammatical functions. In many cases they mark some kind of aspect, the most common being progressive, continuous, durative, imperfective or continuative/persistive (Lichtenberk 2002:308). The following are a few examples from different languages:  (54)  a. Progressive aspect: Swedish (Platzack 1979:55) Linda sitter och röker Linda sits  på expeditionen.  and smokes on office.the  ‘Linda is smoking in the office.’  24  b. Progressive aspect: Kxoe (Kilian-Hatz 2002:324) Tí  kx’ó-à-nùè.  1SG eat-I-sit ‘I am eating (while sitting).’  c.  Continuative aspect: Manhartha (Austin 1998:24) Ngatha  kumpa-artu tharla-rnu  1SG.NOM sit-USIT  papa-jaka.  feed-IMPF.SS water-COM  ‘I used to feed (him) with water.’  d. Habitual aspect: Yankunytjatjara. (Austin 1998:33. Originally from Goddard 1985:207) Wati-ngku kali man-ERG  atu-ra  nyina-nyi.  boomerang.ACC chop-SERIAL  sit-PRES  ‘The man makes boomerangs.’  In (54a) and (54b), which are taken from Swedish and Kxoe, respectively, we have the verb ‘sit’ functioning as a progressive aspect marker. In (54c) we have an example from the Australian language Manhartha, where the verb kumpa, which generally has the meaning ‘sit, camp, stay, live, be’ is used for non-punctual or continuous aspect (Austin 1998:24) and in (54d), from Yankunytjatjara, another Australian language, the verb ‘sit’ is used for a habitual meaning. Using a posture-verb coordination like this is a very common way to form the progressive in the Scandinavian languages and as Tonne (2001:74) has pointed out it is usually either called “a pseudocoordination” (Teleman (1983), Wiklund (1996), Johannessen (1998), Tonne (2001)) or a “postural verb construction” (e.g. Ebert 2000). Tonne (2001) describes it as a “coordination of two (or more) verbs in the same tense (or lack of tense), where the first verb describes a state or movement and functions in the discourse as a background for the action (or state) described by the next verb.” Now let us look at some examples from Icelandic.  (55)  a.  María les. María reads ‘María reads.’ ’Mary is reading.’  b.  María er a lesa. María is to read (inf) ‘María is reading.’ 25  c.  María situr og les. María sits  and reads  ‘María is (sitting and) reading.’  Here the simple present gives us a habitual reading and the meaning is that Mary generally reads, or that she is a reader, whereas both (55b) and (55c) have the meaning that ‘at this particular moment Mary is reading’.27 (55c) additionally gives us the information that Mary is in a sitting position while doing her reading. It does not mean that the sitting event and the reading event had to start at the same time or that they have to end at the same time, only that they are both taking place at the reference time. Now, this does not mean that the Icelandic simple present verb form situr means ‘is sitting’ or that les means ‘is reading’. However, when put together they form a unit, which yields a progressive construction. In Icelandic a sentence with a posture verb conjunction can only be true if both members of the conjunction are true. Thus the subject absolutely has to be in the posture denoted by the posture verb. 28 This seems to indicate that the construction has not been fully grammaticalized. However, notice that the order cannot be reversed:  (56)  *?María les  og situr.  María reads and sits ‘María is sitting and reads.’  So even though the construction as such has not yet been fully grammaticalized as a progressive construction, this indicates that the conjunction here is not a regular conjunction.29 Even though the posture verb progressive clearly gives progressive meaning, it is not as freely used as the other two progressive constructions, partly because it requires a stative posture (sitting, lying, standing) and can hardly be used when the subject is, for instance, on the move:  27  The only non-habitual context where the a-sentence is natural is in a play-by-play context, like in a play script, where it is not uncommon to read things like: Stofan í íbú Jóns og Maríu. Jón horfir út um gluggann. María les. ‘The living room in John and Mary’s apartment. John looks out the window. Mary reads.’ The same can be heard on the radio: Næst á dagskrá er útvarpssagan: „Heimsljós“ eftir Halldór Laxness. Höfundur les. (Next is a reading from the book “Light of the World” by Halldór Laxness. Author reads). Thanks to Gunnar Ól. Hansson for this example. 28  When asked, some Swedish speakers claim that in Swedish the person has to be in the posture denoted by the sentence but others say that is not the case. This means that the meaning of the posture verb is relevant to at least some speakers of Swedish.  29  I should point out that we get the same pattern in English when we have a posture verb with an event verb, connected by the conjunction and – the posture verb has to precede the event verb. Therefore John sits and reads is fine whereas John reads and sits seems awkward. 26  (57)  #María  gengur og  syngur  María  walks and  sings  ‘María is (walking and) singing.’  These facts, as well as the fact that the posture verb progressive is so rarely used in Icelandic even though it is so commonly used in the other Scandinavian languages, show us that the posture verb progressive needs to be handled separately from the other progressive constructions of Icelandic. I will, therefore, not discuss it further in this dissertation but save it for future research. Until then, the reader is referred to the more detailed discussion in Jóhannsdóttir (2007b, 2007c).  1.4.3  Conclusions  It is clear from the previous discussion that English has one grammaticalized construction that it uses to indicate that an event is in progress. Icelandic also has such a construction, the infinitival progressive, which is used with most eventive verbs for the same purpose. These two constructions, the present participle progressive of English and the infinitival progressive of Icelandic, are the main topic of this dissertation and will be discussed in more detail in chapters 2, 3 and 4. However, Icelandic also has other constructions that it uses for the progressive, namely the present participle progressive and the posture verb progressive. In chapter 5 I will focus on the present participle progressive which is used extensively with iterative events in the progressive. Table 1:1: Progressive constructions of English and Icelandic. Present participle progressive Infinitival progressive Posture verb progressive   * *  English (discussed throughout)  Icelandic  (discussed in ch 5)  (discussed in chs 2,3,4) () (not discussed here)  1.5 Theories of the progressive There exist many formal theories of the progressive but the most influential ones have been those of Bennett and Partee (1972), Dowty (1979), Parsons (1990), Landman (1992) and Portner (1998). Most of the discussion has focused on the so-called imperfective paradox, which arises first and foremost with accomplishment verbs. Consider the following examples:  (58)  Activity a. Mary was pushing a cart. b. Mary pushed a cart.  27  (59)  Accomplishment a. Mary was drawing a circle. b. Mary drew a circle.  The important difference here is that the activity sentence in (58a) entails the activity sentence in (58b) but the accomplishment sentence in (59a) does not entail the accomplishment sentence in (59b). Let us now look at the main theories of the progressive, starting with Bennett and Partee.  1.5.1  Bennett and Partee  Bennett and Partee (1972) introduced the idea of subintervals in their account of the progressive and gave the analysis in (60). Subinterval verb phrases have the property that if they are the main verb phrase of a sentence which is true at some interval of time I, then the sentence is true at every subinterval of I including every moment of time in I.  (60)  [Prog ] is true at interval I iff there exists an interval I’ such that I  I’, I is not a final subinterval of I’, and  is true at I’.  This works well in explaining the entailment in (58): If ‘Mary was pushing a cart’ is true then there is an interval I’ which contains a past interval I, such that ‘Mary push a cart’ is true at I’. Because ‘Mary push a cart’ is an activity, the subinterval property holds of it and therefore ‘Mary push a cart’ is also true of I. From this it follows that I is a past interval at which ‘Mary push a cart’ is true. Bennett and Partee would describe an accomplishment like (59a) as: ‘Mary is drawing a circle’ is true at an interval iff that interval is a subinterval of a larger, later ending interval where ‘Mary draw a circle’ is true. Another way of saying this is: Mary’s drawing a circle is in progress at a certain period iff it will continue beyond this period and eventually be fully realized. Bennett and Partee’s analysis predicts (59b) to be true but as Landman (1992) points out, Mary could have been interrupted and never finished drawing the circle and in that case (59b) is not true. This is where Dowty’s (1979) account comes in, to which I will now turn.  1.5.2  Dowty  Dowty (1979) points out that the problems for Bennett and Partee’s (1972) analysis are due to the requirement that  has to be true at a superinterval in the real world. Dowty, thus, proposes an analysis where  does not have to be true at a superinterval in the real world and instead suggests a set of worlds that he calls inertia worlds:  28  Inertia Worlds - are to be thought of as worlds which are exactly like the given world up to the time in question and in which the future course of events after this time develops in ways most compatible with the past course of events. (Dowty 1979:148) Dowty treats the progressive as a sentence operator for which he gives the following definition, where Inr is an accessibility relation, which picks out the sets of inertia world. Therefore Inr(I,w) stands for the set of inertia worlds for w and interval I.  (61)  [Prog ] is true at <I,w> iff for some interval I’ s.t. I  I’ and I is not a final subinterval for I’, and for all w’ such that w’  Inr(I,w),  is true at < I’,w’>. Dowty discusses the inability of certain stative verbs (which he calls ‘object predicates’) to occur  in the progressive.  The usefulness of such predicates as know, like, believe, intelligent, soluble, fragile ... in language is that they indicate a potential for having stage-properties of a certain kind at some future or hypothetical time. And this potential exists at any one moment during the whole interval of their truth as much as at any other moment. The intervals at which stage predicates are true, by contrast, are shorter, have distinct boundaries, and may have truth conditions that differentiate among parts of the interval, so it is perhaps not surprising that our language has a means for locating the present (or some past or future event) at a time within such an interval for stage-predicates but not for object predicates. (Dowty 1979:179-180) As it is part of the function of the progressive to locate a particular time within a larger interval of time where the corresponding non-progressive sentence would be true, the fact that stative sentences like John is knowing French would automatically be true at both the longer interval and any of its subintervals is the reason why such stative sentences cannot occur in the progressive.  1.5.3  Parsons  Many linguists (e.g. Vlach (1981), ter Meulen (1986), Bach (1986), and Parsons (1990)), have rejected the modal analysis of Dowty and have argued that the relation between the kind of events that are described by the progressive and those that are described by a non-progressive, is a primitive fact about the domain of events and therefore cannot be defined in modal terms. Instead, Parsons (1990) introduces incomplete events as primitives. Parsons assumes that it is not the function of the progressive to relate an event in progress to a complete event but to turn a complete event into an incomplete progressive event (at least in the case of accomplishments – achievements are a bit different but Parsons does not really discuss them). He introduces the terminology hold and culminate which plays an important role in his theory as the difference between the progressive and the non-progressive is really that the former contains the predicate 29  hold whereas the latter contains the predicate culminate. More accurately, the progressive changes the culminate relation into a hold relation and so the truth of the past progressive only postulates an incomplete event.  (62)  Hold: Agatha was crossing the street. t[t< now  e[crossing(e)  Subject(e, Agatha)  Object(e, the street)  Hold(e,t)]]  (63)  Cul (Culminate): Agatha crossed the street. t[t< now  e[crossing(e)  Subject(e, Agatha)  Object(e, the street)  Cul(e,t)]]  The important part here is that just because some event holds, this does not mean it culminates, so even though Agatha was crossing the street this does not entail that Agatha crossed the street. Parsons gives Hold and Cul as relations between events and times and so crossing a street may either culminate at time t, or hold at time t. Parsons does not define Hold and Cul further than that. Parsons’ theory makes it easy to deal with the imperfective paradox. As the progressive does not involve any complete events and since its function is simply to tell of an event that it holds, it does not have to culminate. According to that the sentence in (64) is true even if Mary was interrupted and never finished drawing a circle (Landman 1992:5):  (64)  a. Mary was drawing a circle. b. x[Circle(x)  ei [i < now Drawing(e) Agent(e)=Mary)  Theme(e)=x  Hold(e,i)]]  However, in (64) there is a reference to a circle, and if Mary got interrupted before she finished drawing the circle no circle came into existence. So how can there be a circle in the truth conditions of the sentence? Parsons (1990:178) does not see this as a problem and points out that people refer to unfinished houses as houses all the time. If an unfinished house can be called a house then an unfinished circle could be called a circle. However, if we assert that an unfinished circle is a circle then this amounts to denying the imperfective paradox, since then ‘Mary drew a circle’ is true even if she only drew part of one.30 If we claim that ‘Mary drew a circle’ is true even if she did not finish, then it is not clear what the difference between the progressive and the non-progressive really is. Additionally, Landman asks how we can deal with situations where no incomplete object exists: 30  Many linguists and philosophers reject the idea of talking about non-existing objects in such a way (see, for instance, Quine (1953), Lewis (1990), Priest (2005)). For a more detailed discussion on incomplete events see Zucchi (1999) and Szabó (2008). 30  (65)  God was creating a unicorn, when he changed his mind.  Here there is no such thing as an incomplete unicorn and so Parsons’ explanation would not hold. Let us turn to Landman.  1.5.4  Landman  According to Landman (1992), in a sentence like ‘Mary is building a house’ the progressive presents an internal perspective and so Mary’s building of a house is presented as perceived from the inside as an incomplete event, that is, an event in progress. Therefore ‘Mary is building a house’ is true iff Mary’s building a house is in progress. As I will be using Landman’s theory in this dissertation I will now explain his analysis in somewhat more detail than the previous analyses. In his account, Landman introduces two new terms that are necessary for his analysis of the progressive, stages and continuation branch. Landman discusses both stages of individuals and stages of events. Any spatio-temporal realization of an individual is a stage of that individual (cf. Carlson 1977), but something is a stage of an event only if the bigger event is a more developed version of the first. Landman argues that the stages of e are more like e than any arbitrary parts are and he assumes that often, stages will share certain characteristics with the event they are a stage of:  (66)  Let f be an activity or accomplishment and e a stage of f. e is a process stage of f iff e has the same process characteristics as f.  If f is a process of Mary running, e is also a process of Mary running and if f is an event of Mary running to the store then e is a process of Mary running (not Mary running to the store). Landman builds his stage-of relation on Carlson’s (1977) notion of stages. Carlson assumes an ontology of individuals and stages of individuals, the latter being spatio-temporal realizations of the former. Basically, in any given world an individual can be realized at more than one interval. Stages of individuals are uniquely linked to intervals. A stage of Hanny Schaft in 1940 is a part of the stage of Hanny in 1940-1941. Landman assumes that e is a set of events and that it is ordered by two relations: a relation ‘part-of’ and a relation ‘stage-of’, both of which are partial orders. A stage of an event is then a special sort of part of that event. Landman (1992:23) describes this in the following way:  An event is a stage of another event if the second can be regarded as a more developed version of the first, that is, if we can point at it and say, “It is the same event in a further stage of development.” Thus, not every part of e at an interval is a stage of e; to be a stage, a part has to be big enough and share enough with e so that we can call it a less developed version of e.  31  Landman (1992:23) assumes the following:  An event e can be a stage of two events f and g even where f and g are not stages of a common event. That is, an event can possibly develop into different events (my walking now can possibly develop into a walking to Rome and into a walking to Tietjerksteradeel). But within a given world this is not possible: if e is a stage of f and of g and all of e, f, and g are located (by t) in w, then f and g are stages of a common event h located in w (if my walking actually develops into both a walking to Rome and to Tietjerksteradeel, then those are part of the same walking event, which by the nature of walking means that I pass through one on the way to the other). In Landman’s proposal we look at a certain event in a world and its continuation until it stops in that world. When it stops, “there is no bigger event in the world of which it is a stage” (Landman 1992:23). Let us now look at Landman’s notion of a continuation branch and how it involves stages. As mentioned above it is a problem for Dowty’s analysis that we can see (64) above, Mary was drawing a circle even though she never finished drawing it. Inertia worlds, as per Dowty, do not work because they entail that the event finishes in the most similar worlds, but that is not always true. So Landman tries to solve the problem by using the notion of a continuation branch. He assumes that if event e stops in w, there is a closest world where e does not stop. In the case of Mary drawing a circle and not finishing, the closest world is a world where she does finish. Landman (1992:26-27) introduces the notion of continuation branch, which is defined as in (67). As Landman defines them, thematic roles are partial functions from events to individuals (Landman 1992:19). Here R(e,w) stands for the set of reasonable options for e in w.  (67)  The continuation branch for e in w is the smallest set of pairs of events and worlds such that 1. for every event f in w such that e is a stage of f, f, w  C(e, w); the continuation stretch of e in w; 2. if the continuation stretch of e in w stops in w, it has a maximal element f and f stops in w. Consider the closest world v where f does not stop: — if v is not in R(e, w), the continuation branch stops. — if v is in R(e, w) then f, v  C(e, w). In this case we repeat the construction: 3. for every g in v such that f is a stage of g, g, v  C(g, w), the continuation stretch of e in v; 4. if the continuation stretch of e in v stops, we look at the closest world z where its maximal element g does not stop: — if z is not in R(e, w), the continuation branch stops. — if z is in R(e, w) then g, z  C(e, w) and we continue as above, etc.  What this all means is that we follow e in our world: if its continuation stops, we follow it in the closest world where it does not stop, if that world is a reasonable option for e in w; if the continuation stops in that world, we go to the closest world again, and so on.  32  Figure 1-2: Landman’s (1992:27) continuation branch  Landman argues that the progressive is a relation between an event e and a predicate P; he gives the following semantics:  (68)  [[PROG(e,P)]]w,g = 1 iff fv:f, v  CON(g(e), w) and [[P]]v,g (f) = 1 where CON(g(e), w) is the continuation branch of g(e) in w.  Then PROG(e, P) is true in w relative to g if in some world on the continuation branch of g(e) in w, some event realizes the event type P. Here g is the assignment function, e is an object-language variable over events, and g(e) is the event that the function g assigns to e. Landman points out himself that by the definition of continuation branch, this means that g(e) is a stage of that event. Hence stages are an important part of the continuation branch and the continuation branch is an important part of the definition of the progressive. Even though Landman himself never discusses states it is reasonable to assume from his analysis that states do not have stages as they lack internal structure and are temporally homogeneous. Every moment of loving Paul is is a loving-Paul moment, whereas not every instance of building a house is a building-a-house moment, and not even every instance of running is a runninginstance, as a minimal running is needed in order for it to be running. So states do not have stages and as stages are a requirement for the progressive we can draw the conclusion that we should not get states in the progressive. Based on this, Landman gives the translation of Mary was building a house in the following way. Notice that his t(e’) stands for the time of the event, usually represented with (e’): 31  (69)  e’. t (e’) < now  PROG (e’, e. y[House(y) Build(e)  A(e)=m  T(e)=y])  (69) is true in w iff in some world, an event of building a house by Mary goes on, a stage of which goes on in our world at some past interval, a stage which develops into that event on its continuation branch (Landman 1992:27-28). 31  I will use  throughout this dissertation to represent the run-time of the event.  33  Landman’s idea does well in explaining what it means for an event of certain type to be in the progressive. Portner, however, points out that the fact that Landman’s analysis is quite unique and in no way related to other analyses of intentional operators or other aspectual morphemes like the perfect, weakens that theory considerably and instead suggests that a preferable approach would be to treat the progressive as a more “ordinary, familiar-seeming kind of element” (Portner 1998:11). So let us now look at Portner’s own analysis.  1.5.5  Portner  By using Dowty’s (1979) theory of the progressive and Kratzer’s (1977, 1981, 1991) theory of modality, Portner (1998) introduces his own analysis of the progressive. He integrates the notions of events and inertia worlds into Kratzer’s modal base/ordering source semantics. I will illustrate this with respect to one of his examples (Portner 1998:772):  (70)  At 7 o’clock, Mary was climbing Mount Toby.  The modal base is the set of facts relevant to whether e is completed as an event of Mary climbing the mountain. The identities of the propositions in the modal base are contextually determined – based on actual facts of the matter at hand combined with the knowledge and interest of both the speaker and the hearer. For (70) he gives the modal base shown in (71) and refers to it as Circ(e):  (71)  M(w) = {‘Mary is in good physical condition’, ‘Mary does not give up easily’, ‘It was raining lightly on Mount Toby at 7 o’clock’, ‘Mary was one third of the way up the Mount Toby trail at 7 o’clock’, ‘Mary was headed the right way on the trail at 7 o’clock’, ...}  It is not necessary for Mary to climb Mount Toby in all of the worlds compatible with this modal base, in order for the sentence in (70) to be true. As Portner points out, a bear might come out of the forest and get her, she might hurt her ankle, or the like. However, (70) is true because if she is not interrupted, Mary will climb Mount Toby. The ordering source (abbreviated O) provides the set of all propositions that express that the event is not interrupted. These propositions are contextually determined just like the modal base. Portner suggests the following, where O(w) is seen as the set of outside factors which need to go right for Mary if ‘Mary climbs Mount Toby’ is true. He refers to it as NI(e) (Portner 1998:773:  (72)  O(w) = {‘Mary does not get eaten by a bear’, ‘Mary does not give up easily’, ‘A surprise summer blizzard does not start on Mount Toby’, ‘Mary does not get lost’, ...} 34  Portner assumes that there is a unique best set of worlds and gives the following formal definition (Portner 1998:771):32 (73) Best(M, L, w) = the set of worlds w  in M(w) such that there is no w   in M(w) where w  <L,w W . In the scenario where Mary climbs Mount Toby, the best worlds would of course be ones where she does not get eaten by a bear does not slip and hurt her ankle, etc. Best(M, O, w) corresponds to the inertia worlds in Dowty’s theory. If we combine (71) and (72) we get Best(M,O,w) (Portner 1998:15). Portner suggests that the progressive sentence in (70) is true because all such worlds are ones where Mary climbs Mount Toby. The interpretation of the progressive is then:  (74)  PROG() is true at a pair of an interval and world <i,w> iff there is an event e in w such that T(e)=i and for all worlds w’ in Best(Circ, NI, e), there is an interval i’ which includes i as a nonfinal subinterval, such that  is true at <i’,w’>.33  This can be read in such a way that (70) is true at <i,w> iff there is an event going on during i in w which, if it is not interrupted, will become an event in which Mary climbs Mount Toby. Portner’s analysis also applies to activity sentences where there is no natural endpoint. So sentences like (75) have the characterizing feature of being true of all subintervals of an interval at which they are true, as long as that subinterval is long enough.  (75)  Max is running.  So if Max runs is true between 1pm and 2pm, then it is also true at 1:15pm or from 1:15 to 1:45. Basically, (75) is true because in all inertia worlds there is an appropriate event e´, which makes the tenseless Max runs true. Portner’s analysis is a very appealing approach with the advantage that it analyses the progressive by using independently-motivated tools from modality. In fact, I do believe that Portner’s analysis is a good alternative to Landman’s theory and the analysis I will be presenting in this dissertation could also be re-formulated within a Pornter-style system.  32  L stands here for the ordering source which he also refers to as O and as I have chosen to do in this discussion.  33  Portner uses here T(e) for what is more commonly written as (e). 35  1.6 Aktionsart We can say that the Aktionsart of a verb is a part of the way in which that verb is structured in relation to time. Aktionsart differs considerably from viewpoint aspect in that most verbs can be seen as belonging to a particular Aktionsart, and so it is not the speaker’s choice which Aktionsart he uses each time, as it is with viewpoint aspect. The speaker does, however, have some control over the Aktionsart he uses, as he can coerce verbs from one Aktionsart to another, and as that is exactly what I claim happens when we get stative verbs in the progressive. I will now discuss the major categories of Aktionsarten.  1.6.1  Aristotle and Kenny  The idea of Aktionsart, or lexical aspect, can be traced back to Aristotle who was the first to introduce the relevance of the inherent temporal structure of verbs. He distinguished between kineseis verbs (verbs that need to reach an end, similar to telic verbs) and energeiai verbs (similar to atelic verbs) (see e.g. Verkuyl 1993:43). Kenny (1963) divided situations into states and occurrences, with the latter again splitting into activities and performances. In order to distinguish between activities and performances he proposed an entailment test where the progressive activity sentence in (76a) entails (76b), but the progressive accomplishment sentence in (77a) does not entail (77b).  (76)  Activity a. John is pushing a cart. b. John pushed a cart.  (77)  Performance a. John was drawing a circle. b. John drew a circle.  As we have seen, this entailment is one of the most discussed issues of the progressive as it leads to the well-known imperfective paradox, already discussed in section 1.5.  1.6.2  Vendler’s classification  Vendler (1957) has provided what is probably the best known classification of Aktionsarten. He classified verbs into four Aktionsarten: activities, accomplishments, achievements and states.  36  (78)  Activities a. Scott is playing golf. b. Diane is eating peanuts.  (79)  Accomplishments a. Gina is drawing a picture. b. Bill is carving a bird.  (80)  Achievement a. Clive reaches the summit. b. Hannah won the race.  (81)  States a. John loves his wife b. Mary knows the National Anthem by heart.  Vendler gave the following ‘time schemata’ to characterize these verb classes:  (82)  STATE: A loved somebody from t1 to t2 means that at any instant between t1 and t2 A loved that person. ACTIVITY: A was running at time t means that time instant t is on a time stretch throughout which A was running. ACCOMPLISHMENT: A was drawing a circle at t means that t is on the time stretch in which A drew that circle. ACHIEVEMENT: A won the race between t1 and t2 means that the time instant at which A won the race is between t1 and t2.  The important parameters here are instant and stretch, which Vendler uses to separate states and achievements from activities and accomplishments. Notice also the difference between the indefiniteness of states and activities versus the definiteness of accomplishments and achievements. Essentially, states and activities are non-unique, indefinite temporal entities whereas achievements and accomplishments are unique, definite temporal units.  37  1.6.3  The classification used in this dissertation  Since Vendler, many different variations of this system have been proposed. Mourelatos (1978) unites Vendler’s and Kenny’s analyses, distinguishing four categories like Vendler but using the binary system of Kenny, further splitting processes into developments (accomplishments) and punctual occurrences (achievements). Dowty (1979), Hoeksema (1984) and ter Meulen (1986) keep Vendler’s classes but differ in criteria and terminology. In addition Carlson (1981) and Moens (1987) add to Vendler’s classes, categorizing verbs into six and five classes respectively. The system I use in this dissertation is based on Vendler’s classification into the four main verb types, but the features are more related to Smith (1997), who uses the features ±Telic, ±Durative and ±Static.34 However, as dynamic – the opposite of static – will play an important role in chapter 2, I will use ±Dynamic instead of ±Static. Notice though that which system I use is in no way crucial for my analysis as all the systems capture essentially the same empirical divisions. Table 1:2: The aspect classification used in this dissertation Telic – – + +  States Activities Accomplishments Achievements  1.6.4  Durative + + + –  Dynamic – + + +  Aktionsart and the progressive  Often, only activities and accomplishments, and not statives and achievements, are said to occur in the progressive (see e.g. Poutsma (1926), Lakoff (1966, 1970), Visser (1973), Dowty (1979)):  (83)  a. John is running.  (activity)  b. Mary is drawing a circle.  (accomplishment)  And yet, most achievements do occur in the progressive and many stative verbs as well:  (84)  a. John is reaching the summit. (achievement) b. I am loving this song.  (stative)  There is an obvious difference between these two sentences: The achievement in the progressive gets the ‘about-to’ reading whereas the stative implies a temporary situation. This is not surprising considering the 34  Smith also added the category of semelfactives which are similar to achievements but usually require iteration of the event. As semelfactives are not important in this dissertation I will stick to using Vendler’s classification into four categories. 38  difference between the two kinds of Aktionsart. Achievements are momentaneous and cannot have subintervals so the progressive cannot pick a time within the event. If the achievement is coerced to being an accomplishment, as suggested by Rothstein (2004), the progressive picks a time immediately preceding the culmination of the event, and in Icelandic, sometimes immediately following the culmination of the event. Statives, on the other hand, are durative and so theoretically it should be easy to pick a subinterval for an ongoing state. But as each moment of a state is identical to the next, each subinterval would also be exactly the same as the next and state, therefore, do not have stages. If the stative is coerced to being eventive, however, it behaves like an activity and the progressive becomes available. I will discuss my analysis of statives in the progressive as well as Rothstein’s analysis of achievements in the progressive in chapter 3. But before that I will discuss stative verbs in general in chapter 2.  1.7 Some final notes As previously mentioned, this dissertation focuses on the progressive in Icelandic and English with special attention given to the less commonly studied uses of the progressive, such as stative verbs and habitual sentences in the progressive. In order to come up with an accurate analysis of these progressive uses in the two languages various different methods were used in gathering data. I used The Corpus of Contemporary American English, the Internet, television shows, personal communication and made-up examples, as well as data from the Icelandic ScanDiaSyn project (ráinsson 2010). This is not a corpusbased study, but I hope that it does give a fairly accurate picture of how the progressive is used in these languages. The question of how reliable the Internet is as a language resource has not been answered but it is clear that the Web plays an increased role in linguistic research today. This can, for instance, be seen by the creation of the Linguistic Search Engine, designed to search for linguistic data on the Web (Resnik et al. 2005), as well as the WaCky (Web as Corpus kool ynitiative) project, which is currently working on developing a set of tools that will allow linguists to “crawl a section of the web, process the data, index and search them” (http://wacky.sslmit.unibo.it/). There are, of course, many dangers of the Web, such as how hard it is to weed out non-native speakers of the language being studied, as well as other factors that might skew the research. The linguist using the Internet needs to be careful with the data she gathers and cannot rely too much on the findings if they are not backed up by data gathered elsewhere. But this is not to say that linguists should not use the Internet as a language source and in fact I believe it is a quite valuable source if used right. In my study I use it to gather examples of states and habits in the progressive, as an addition to examples I have found in other written texts, on the TV and in personal conversations. 39  I use standardized linguistic conventions in marking sentences, where ‘*’ stands for an ungrammatical sentence, ‘?’ for a questionable sentence, and ‘#’ represents sentences that are grammatical but only acceptable in a different context than the given one. The glosses of the Icelandic data are simple word glosses where I do not mark linguistic categories or internal morphological structure, except for the relevant verbal form of the sentence. I also want to point out that even though my analysis is event-based, in the style of Davidson (1967), I do adapt non-event-based analyses, such as Van Geenhoven’s (1995, 2005) analysis of frequency adverbs. In these cases, I adapt their analysis to work with events. Lastly I want to explain some of the terminology in the thesis. The term eventuality refers to states and events alike. As English mainly uses only one construction to indicate an event in progressive, I will always refer to it as the progressive. In Icelandic, however, where more than one construction is available to indicate an event in progress I will use the term infinitival progressive or simply the progressive for the most common progressive construction in the language, where the main verb in the infinitive follows the copula. When that construction is used in a habitual context I will call it the progressive habitual. In such cases it still has the same grammatical structure but the meaning differs. This is different from the present participle progressive, which is a separate progressive construction, which uses the copula followed by the main verb with the present participle suffix –andi attached.  40  2 Stative verbs 2.1 Introduction Event verbs in both English and Icelandic are frequently used in the progressive to convey that an event is in progress, but stative verbs generally occur in the simple past or the simple present, and not in the progressive, as has already been mentioned. However, as also has been mentioned already there are cases when verbs that are usually categorized as stative verbs do occur in the progressive, and as I will argue in chapter 3, this usually happens because the speaker wants to convey additional information that is not so easily accessible with regular stative verbs. In order to do so he will then shift the state to becoming an event. In order for us to understand what exactly it means for the verb to be shifted from a state to an event it is important to make a clear distinction between states and events so we know what it is that makes the shift necessary. In this chapter I will, therefore, discuss the category of stative verbs and in what way they differ from event verbs. I will discuss in detail the tests that have been used to distinguish states and events and what exactly these tests tell us. In chapter 3 I will use the results of these tests to argue for the reason behind the shift from a stative predicate to an eventive predicate. This chapter is organized as follows. In section 2.2 I will discuss the category of the stative verb, what constitutes a stative and show some tests that have been used by linguists to distinguish states from events. Section 2.3 will focus on the difference between states and events with special focus on Davidsonian and Neo-Davidsonian analyses of events and states, and in section 2.5 I will summarize the results of the chapter.  2.2 The category of the stative 2.2.1  What constitutes a stative?  Most linguists agree on the existence of stative verbs as a class even though they are not always referred to as ‘statives’. Sweet (1903:98) talks about verbs which “express feelings, physical and mental perception”, and gives examples such as feel, like, think. Kruisinga (1931:361-2) uses similar words and refers to “verbs expressing feelings and mental or physical perception (not sensations)”. He gives examples such as love, have, detest, like, prefer, see, hear, believe, belong to, consist, contain, possess, resemble, suffice. However, not everyone agrees on how the stative should be categorized, nor what constitutes a stative. Brinton (1987:203) describes states in the following way:  States are characterized by the inherent qualities of duration and homogeneity, as well as by the lack of change, limits and agency. States exist or endure for an undefined period of time. They do 41  not change or develop during that period; the temporal phases of a state are undifferentiated. States do not happen and are not done, but simply are. Dowty (1979:71) uses time as an important tool to distinguish stative verbs from other verbs and claims that statives “can be judged true or false of an individual by reference to the state of the world at only a single moment of time (while other classes of verb require “information” about more than one point in time and in some cases, from more than one possible world).” Linguists do not always agree on which verbs should be classified as stative verbs. Mufwene (1984:3), for instance, points out that Katz (1972) classifies suffer and sleep as stative verbs whereas Comrie (1976:36) and Palmer (1974:72) classify suffer as an activity/non-stative verb, and Freed (1979:47) and Palmer (1974:71) classify sleep as an activity verb. Verbs like burn, enjoy, lie, sit, stand, stay and wait seem to be categorized equally often as statives and as eventives. Mufwene (1984:3-4) wonders whether the reason why some linguists categorize those verbs as non-statives is merely because they occur in the progressive in English. In order to help with the categorization of verbs, linguists have come up with various tests that aim at distinguishing statives from events. However, it is not all that clear that some of the tests really are tests for stativity and not for some other property, such as control. This has, for instance, been argued by Levin and Rappaport Hovav (2005). I will discuss control in more detail in section 3.4.2.1, but first I will discuss each of the supposed stativity tests and apply them to various verbs. The goal is twofold: on the one hand I want to see how well these tests really work as tests for stativity, and on the other hand I want to explore how clearly the category of stative verbs can be separated from that of event verbs.  2.2.2  Stative tests  Lakoff (1966) provides a number of tests that are supposed to distinguish the category of statives from events. What do these tests really tell us and do they really distinguish statives from events? If they do not, what does that tell us about the category of the stative verb? I will use Lakoff’s original examples for English but additionally I will show the same constructions in Icelandic. I will also apply a few other tests that have been suggested by other linguists.  2.2.2.1 Progressive As has been previously discussed, stative verbs have generally been considered impossible in the progressive. Examples like the following have been given to show this (see e.g. Poutsma (1926:339), Lakoff (1966), Visser (1973:1969)):  42  (85)  a. *John is knowing that. b. John is slicing the salami.  (86)  a. *Jón er a vita Jón  svari.  is to know answer.the  ‘Jón is knowing the answer.’  b. Jón er a skera pylsuna. Jón is to cut  sausage.the  ‘Jón is cutting the sausage.’  The idea is then that verbs that occur in the progressive are events but those that cannot are stative verbs. However, as has been discussed already, some verbs that are intuitively categorized as stative do indeed occur in the progressive:  (87)  a. Not sure if I mentioned this or not, but I am pretty much loving chia seeds right now. http://www.eatingbirdfood.com/2009/06/what-im-loving-and-quinoa-recipe/  b. I am Liking This Guy But I Do not Know How He Feels About Me Exactly. http://answers.yahoo.com/question/index?qid=20100416101216AAwzATT  (88)  a. Vá  hva ég  Wow what I  er  a hata pabba minn núna...  am to  hate dad  mine now  ‘Wow, how I am hating my dad right now...’ http://spjall.liverpool.is/m519098-p2.aspx  b. Úff, ég er  ekki a skilja  Phew I am not  etta mál.  to understand this  case  ‘Phew, I am not understanding this case.’ http://skessa.blog.is/blog/skessa/entry/928620/  If this is a proper stative test then we have to categorize verbs such as love, hate, like and understand as event verbs and not stative verbs. And as most linguists agree that those are typical stative verbs it would require us to revisit the whole idea of stative verbs. I will get back to this in chapter 3. If we assume, for the moment, that these are indeed stative verbs that nevertheless fail this particular stative test, can we trust that the test might hold in one direction in the sense that if a verb cannot occur in the progressive, it is stative? That is not quite true either, as achievement verbs are  43  believed not to occur in the progressive, as discussed in chapter 1, and in the cases where they do, they have been coerced to being accomplishments (Rothstein 2004):  (89)  a. #Jane is reaching the summit of the mountain. b. #Mary is spotting her friend at the party.  Perhaps we could say that if a verb cannot occur in the progressive at all and it is not an achievement, then it is stative. However, I believe there are only a few stative verbs that cannot be coerced to being eventive, which means that as a test the progressive can only reliably pick out a very small set of verbs.35 That is simply not sufficient to count as a proper stative test.  2.2.2.2 Imperatives Lakoff (1970:121) used the imperative as a stative test, claiming that statives could not occur in the imperative, unlike non-statives.  (90)  a. *Know that I am here!  (stative)  b. Slice the salami!  (non-stative)  This makes sense as the imperative mood expresses direct commands or requests. If Brinton’s definition of a stative verb as something that is, but not something that happens or is done, is on the right track, we can see why it would be difficult to use stative verbs in the imperative; it is usually easy to command someone to do something but not to be something. However, it is not that uncommon to use stative predicates in the imperative:  (91)  Sweet dreams, sleep tight and know that I love you. - TV show Medium.  The same can be seen in Icelandic:  35  Of all the common stative verbs in English, all of them can be found in the progressive on the Internet or in COCA. I assume, however, that some rare stative verbs may not occur in the progressive but this would require a more detailed work. 44  (92)  Og vittu  a  a er líka „delete” og „mark as spam” element í WP  And know that there is also “delete” and “mark as spam” element in WP strikerfinu  ínu.  operating.system.the yours ‘And know that there is also a “delete” and “mark as spam” element in your WP operating system.’ http://blog.eyjan.is/ragganagli/2009/08/30/betur-hugsad-um-bilskrjodinn/  Mufwene (1984:9) provides a number of examples where he claims stative verbs are used in the imperative:  (93)  a. Believe me. (Things are more complex than they look.) b. Trust him. (He is quite dependable.) c. Stay home and relax. d. Assume Dowty’s 1977 interpretation of the progressive with performance verbs is correct.  He also provides the following sentences as examples of non-stative verbs that cannot occur in the imperative (Mufwene 1984:9):  (94)  a. *Misplace your keys. b. *Collapse. c. *Shatter. d. *Die.  With the right context those sentences are actually fine as long as someone has the control to do what is asked. If an actor, for instance, is asked by his director to misplace his keys, (94a) is fine. Also, if we have built a mechanical device that has been designed to self-destruct when given certain preprogrammed spoken commands, (94b) and (94c) are fine. The example in (94d) is a bit different as one does usually not have control over whether one dies or not, and is therefore not easily ordered to do so. However, as a wish, (94d) is perfectly fine, and we do not have to look far to find examples of this.36  (95)  a. Die You Bastard! (Song by Motorhead) b. Die, you Stupid Hurdlers! (Free Arcade game on the Internet) c. Die You Zombie Bastards! (Movie from Zombastic Productions)  36  Laurel Brinton (p.c.) has suggested these might be old subjunctives, or optatives – something like “may you die...” 45  The imperative appears to be problematic as a test in Icelandic, as certain verbs are impossible in the imperative for reasons that do not have anything to do with meaning:  (96)  a. Win the race!  (English)  b. *Vinndu  (Icelandic)  hlaupi!  Win (imp) run.the ‘Win the race!’  Now, it is important to note that the English example in (96a) is only possible in a situation where the person spoken to has it somehow within her power to win the race, say, if she is so much better than everyone else that she will win easily if she just wants to. The Icelandic example in (96b), however, is bad even if we create such a controlled context. In Icelandic many strong verbs with stem-final ll and nn have a gap in their paradigm: they cannot occur in the imperative. This gap is most likely due to competition between the two allomorphs of the imperative singular suffix, -du and -tu, and resulting uncertainty on speakers’ part (for details, see Hansson (1999) and Albright (2009)). And since for morpho-phonological reasons, it is impossible to put verbs like vinna ‘work; win’ and falla ‘fall; fail; flunk’ into the imperative form, they cannot occur in sentences like (97):  (97)  a. *Vinndu/*Vinntu eins og ú Work (imp)  as  getur!  and you can  ‘Work as much as you can!’  b. *Falldu/*Falltu ekki á Flunk (imp)  prófunum!  not on exams.the  ‘Do not flunk the exams!’  Interestingly enough, this only happens in the singular, so by changing the imperative to plural the sentences are perfectly grammatical.  (98)  a. Vinni  eins og  Work (imp.pl) as  i  geti!  and you.pl can  ‘Work as much as you can!’  b. Falli  ekki á  prófunum!  Flunk (imp.pl) not on exams.the ‘Do not flunk the exams!’ 46  The important thing here is that there can be various reasons for why a verb is unable to occur in the imperative form.37 In sum, even though most stative verbs do not occur in the imperative, and most non-stative verbs do, I agree with Mufwene (1984:10) that using the imperative as a test for stativity might not always render the correct outcome. As such it can barely be said to be a useful test for stativity.  2.2.2.3 Do something Another test provided by Lakoff (1970:121) has to do with the verb do and whether it can be substituted for the verb in question. Lakoff claimed that only eventive verbs could occur in that context, and that statives could not. If statives indeed are, and cannot be done, as suggested by Brinton (1987; see section 2.2.1 above), Lakoff should be right, as it should be impossible to use the word do for something that cannot be done:  (99)  a. *What I did was hear the music. b. What I did was slice the salami.  (100)  a. *a sem ég That  that I  geri  var a  heyra tónlistina.  did  was to  hear  music.the  ‘What I did was hear the music.’ b. a  sem ég geri var a skera pylsuna.  That that I did  was to cut  sausage.the  ‘What I did was cut the sausage.’  (101)  a. *I knew the answer, though Bill told me not to do so.38 b. I learned the answer, though Bill told me not to do so.  37  In Icelandic it is not uncommon to put a progressive sentence in the imperative: (i) Vertu ekki a lesa essa bók. Hún er algjör væla. Be (imp) not to read this book. She is total rubbish ‘Do not bother reading this book. It is total rubbish.’ This is not a completely productive construction, however, as it mostly appears in a negative context with a hint of disapproval, similar to what Palmer (1974) suggests about the progressive habitual, see further discussion in section 4.3.1 38  This sentence is a bit problematic for the simple reason that it does not seem natural to tell someone not to know something. If the sentence was turned around and we said ‘I did not know the lyrics even though Bill told me to do so’ the sentence seems a whole lot better. And yet we are referring to a state with the verb do. However, in that case we can assume that what Bill really told me to do was to learn the lyrics, rather than actually to know them. 47  (102)  a. *Ég vissi svari I  ótt  Baldur hafi sagt mér a gera a ekki.  knew answer.the though Baldur have told me to do  that not  ‘I knew the answer even though Baldur had told me not to do so.’  b. Ég læri I  svari  ótt  Baldur hafi sagt mér a gera a ekki.  learned answer.the though Baldur have told me to do  that not  ‘I learned the answer even though Baldur had told me not to do so.’  So far, so good! However, consider the following examples with achievement verbs:  (103)  a. I fell into the well even though John told me not to do so. b. I won the race even though Mary told me not to do so. c. ?John died even though Anne told him not to do so.  In (103a) and (103b) the event in the initial clause can be seen as under the control of the agent although some native speakers of English can also get the accidental reading. The subject of (103c) cannot as easily be seen as in control as the only way for John to control his dying is by committing suicide. In such a case the sentence would have been phrased differently:  (104)  John killed himself even though Anne told him not to do so.  If, on the other hand, John died from an illness, he had no control over the situation and the sentence is bad. It is obvious that control is an important factor in this test and that verbs with non-control subjects will most likely fail the test, whether they are stative or not. It is, however, important to distinguish between situations we can never have control over, such as hearing, and situations we usually do not have control over, such as falling into a well or winning a race. This explains why the sentences in (99a) and (100a) are so much worse than the sentences in (103a-b), where we can under certain circumstances have control. Filip (1999) argues that this test is really testing for agentivity, not stativity, and points out that agentivity entails non-stativity but not vice versa. Control is one of the features of agentivity, as will be discussed further in section 3.4.2.1, so ultimately I agree with Filip. Knowing an answer, hating a dog and believing someone are not seen as ‘doing anything’ as they are not actions of any kind. This fits perfectly with Brinton’s (1987) definition of the progressive.  48  2.2.2.4 Occurrence in the complements of persuade and remind Another test provided by Lakoff has to do with occurrence in the complements of persuade and remind. Lakoff claimed that statives should not be available in those contexts, without explaining further why that is. Let us start by looking at Lakoff’s examples:  (105)  a. *I persuaded John to hear the music. b. I persuaded John to listen to the music.  (106)  a. *Ég fékk Jón til ess a heyra tónlistina. I  got Jón to that to hear  music.the  ‘I got John to hear the music.’ b. Ég fékk Jón til ess a hlusta á tónlistina. I  got  Jón to that to listen to music.’  ‘I got John to listen to the music.’  Just as Lakoff predicts, the stative verbs do not fare well in this context. However, that does not necessarily mean that the test is really testing for stativity. When one persuades someone else, that person usually needs to have control over the situation as it is impossible to persuade someone to do something or be something that is not under his control.  (107)  a. *I persuaded John to fall into the well. b. *I persuaded Mary to win the race. c. *I persuaded Annie to die.  (108)  a. *Ég fékk Jón til ess a detta í brunninn. I  got Jón to that to fall  in well.the  ‘I got Jón to fall into the well.’ b. *Ég fékk Jón til ess a vinna hlaupi.39 I  got Jón to that to win  run.the  ‘I got Jón to win the race.’  39  One Icelandic speaker said that in order for this sentence to work you also had to persuade the other competitors to lose.  49  c. *Ég fékk Önnu til ess a deyja.40 I  got Anna to that to die  ‘I got Anna to die.’  (109)  *I persuaded the sun to shine.  (110)  *Ég fékk sólina til ess a skína. I  got sun.the to that to shine  ‘I got the sun to shine.’  All these sentences are impossible, even though (108b) might work if John was so much better than any of his competitors that he could easily win the race if he so much as tried. This is because in order to persuade someone to do something, it has to be in that person’s power to do so. Otherwise he can only try. Notice that if we tweak the sentences in such a way that the person does indeed have control over the situation they get much better:  (111)  a. I persuaded John not to fall into the well. b. I persuaded Mary not to win the race. c. ?I persuaded Annie not to die.  Even though it is not within one’s control to win a race, it is always in one’s control not to win a race. So even though we cannot persuade Mary to win the race, we can persuade her not to win the race. Similarly one can exhort John not to fall into the well and even not to die, even though the example with die is a bit different because of the nature of death. Also, as soon as the sun in example (110) is anthropomorphized (for example in a fairy-tale context), this made-up sentence automatically becomes acceptable.  (112)  The North Wind persuaded the Sun to shine into the Giant’s garden.  The one being persuaded needs to be in control of the situation.  40  One speaker said that this sentence could work if Anna was on life support and I persuaded her to ask for her machine to be disconnected. In that way Anna has, in some way, the power over her life and death. So basically, this can work if Anna has the control. 50  2.2.2.5 Occurrence with manner adverbs like deliberately, masterfully, etc. Lakoff assumed that stative verbs should not be able to occur with manner adverbs such as deliberately, reluctantly, masterfully, carefully, enthusiastically and well. He calls these adverbs ‘subcategorized for animate sentence subjects’ since they put restrictions on the subject; that is, it has to be animate. However, even though Lakoff pointed to the animacy requirement, the contrast between (113a) and (113b) shows that this is really a test for control rather than animacy as the subject in (113b) is the same as in (113a). Therefore both are animate, and yet (113a) is ungrammatical whereas (113b) is perfectly acceptable. So even though Lakoff is ultimately correct that states do not occur with these adverbs, it is not because of the animacy of the subject but because of control:  (113) a. *John knew the answer brilliantly. b. John sliced the salami carefully.  (114)  a. *Jón vissi svari Jón  snilldarlega.  knew answer.the brilliantly  ‘Jón knew the answer brilliantly.’  b. Jón Jón  skar pylsuna  vandlega niur.  cut sausage.the carefully down  ‘Jón cut the sausage carefully.’  Let us once again look at the non-control event verbs:  (115)  a. ?John fell into the well deliberately. b. ?Mary won the race deliberately. c. ?Anne reluctantly died.  These sentences are strange and call for a very carefully constructed context. John can, for instance, not have jumped into the well, as then it would be impossible to say he fell. But in order to fall deliberately he must have had something to do with the fall, for instance if he deliberately stood too close to the edge when a huge truck drove close by (and swept him in), or something of the kind. In order for the sentences to be grammatical we need to give the subjects some kind of control over the situation. If true, this does indeed indicate that the test really is testing for control and not stativity.  51  2.2.2.6 For phrases According to Lakoff statives cannot occur with a for-phrase such as for his teacher’s sake or for all human kind. In order to do something for someone else one has to have control over the situation and so this test is also testing for control:  (116) a. *John knew that fact for his teacher’s sake. b. John sliced the salami for his teacher’s sake.  (117)  a. *Jón vissi Jón  essa stareynd fyrir kennarann sinn.  knew this  fact  for teacher.the self  ‘John knew this fact for his teacher’s sake.’  b. Jón Jón  skar niur pylsuna  fyrir kennarann sinn.  cut down sausage.the for teacher.the self  ‘Jón cut the sausage for his teacher’s sake.’  Our non-control eventive verbs, as predicted, do not fare so well in this context:  (118)  a. ?John fell into the well for his teacher’s sake. b. ?Mary won the race for her teacher’s sake. c. Jesus died on the cross for all human kind.  For these sentences to be grammatical the subject needs to have control over the situation. If John fell into the well accidentally (118a) is bad, and in (118b) Mary would need to have control over the result of the race. (118c) is better as Christians believe that it was Jesus’ choice to die for us. He must, therefore, have had some control over the event. Thus the test is indeed distinguishing between control and non-control rather than eventive and stative verbs.  2.2.2.7 Occurrence on either side of instead of Lakoff (1966) claimed that stative predicates cannot occur on either side of the construction X instead of Y, whereas non-statives can:  (119) a.*I heard the music instead of looking at the painting. b. I listened to the music instead of looking at the painting. c. *I looked at the painting instead of hearing the music. 52  (120)  a. *Ég heyri tónlistina í sta I  ess a horfa á málverki.  heard music.the instead that to look at painting.the  ‘I heard the music instead of looking at the painting.’  b. Ég hlustai á tónlistina í sta I  ess a horfa á málverki.  listened to music.the instead that  to look at painting.the  ‘I listened to the music instead of looking at the painting.’  c. *Ég horfi á málverki I  í sta  ess a heyra tónlistina.  looked at painting.the instead that  to hear music.the  ‘I looked at the painting instead of hearing the music.’  Let us once again look at the non-control eventive verbs:  (121)  a. ?John fell into the well instead of climbing the mountain. b. ?Mary won the race instead of going to Starbucks. c. ?Anne died instead of going to Paris.  These sentences are oddly comical – perhaps because they compare two completely different things, or maybe because one event is agentive and the other one is not. But that can easily be tested. In (122a) we have two eventive verbs and the events are completely unrelated. The sentence does make one wonder why John would make such an odd choice (at least for those who love Paris) but the sentence is not weird or comical. Example (122b), however, is quite strange as it implies that John could choose between loving his wife and feeling pain. And so is (122c), which is eventive but non-control.  (122)  a. John read a book instead of flying to Paris. b. *John loved his wife instead of feeling pain. c. *John fell into the well instead of dying.  There is a notion of choice involved. In order to do one thing instead of another, one must be able to make some kind of choice, have control over the situation.  2.2.2.8 Habitual meaning One of the distinctions often made between stative verbs and eventive verbs is that the latter get a habitual reading in the simple present whereas the former do not (see for instance Kenny 1963):  53  (123)  a. John plays guitar. b. Steve sings opera. c. Sherlock investigates crimes.  (124)  a. Doug loves his wife. b. Marion likes the Beatles. c. Beth knows the national anthem.  The sentences in (123) are non-episodic. John knows how to play guitar but he may not be playing right now. In fact, he may not play guitar very often at all. Similarly Steve does not have to be singing at the utterance time as long as he is someone who every now and then sings opera. To get the present episodic meaning the progressive is required:  (125)  a. John is playing guitar. b. Steve is singing opera. c. Sherlock is investigating crimes.  The sentences in (124) are stative and if Doug loves his wife then it must mean that he loves her now. Similarly Marion does not just like the Beatles in general, she likes them now as well. The statives clearly include the speech time as a necessary time where the state holds. (124c) is false if Beth does not know the national anthem right now. We get the same distinction in Icelandic. Event verbs in the simple present get a habitual reading:  (126)  a. Jón spilar á  gítar.  Jón plays on guitar ‘Jón plays guitar.’  b. Stefán syngur óperutónlist. Stefán sings  opera  ‘Stefán sings opera.’  c. Sherlock rannsakar  glæpi.  Sherlock investigates crimes ‘Sherlock investigates crimes.’  If we want the episodic reading we use the progressive: 54  (127)  a. Jón er a spila á  gítar.  Jón is to play on guitar ‘Jón is playing guitar.’  b. Stefán er a syngja óperutónlist. Stefán is to sing  opera  ‘Stefán is singing opera.’  c. Sherlock er a rannsaka glæpi. Sherlock is to investigate crimes ‘Sherlock is investigating crimes.’  Stative verbs, however, have a regular stative reading and so the state has to hold at the speech time:  (128)  a. Doddi elskar konuna sína. Doddi loves wife.the self ‘Doddi loves his wife.’  b. María er hrifin  af Bítlunum.  María is enchanted by Beatles.the ‘María likes the Beatles.’ c. Beta kann jósönginn. Beta knows national.anthem.the ‘Beta knows the national anthem.’  Notice what happens when these sentences are in the past tense:  (129)  a. John played guitar. b. Steve sang opera. c. Sherlock investigated crimes.  (130)  a. Doug loved his wife. b. Marion liked the Beatles. c. Beth knew the national anthem.  55  Now the event sentences are ambiguous. The habitual reading is still available but so is the episodic reading; John can either be someone who used to play guitar, or he may have played guitar only once. The stative sentences say that at some point the state held that Doug loved his wife. There is no ambiguity. Again, the same pattern can be seen in Icelandic.  (131)  a. Jón spilai á  gítar.  Jón played on guitar ‘Jón played guitar.’  b. Stefán söng óperutónlist. Stefán sang opera ‘Stefán sang opera.’  c. Sherlock rannsakai glæpi. Sherlock investigated crimes ‘Sherlock investigated crimes.’  (132)  a. Doddi elskai konuna sína. Doddi loved  wife.the self  ‘Doddi loved his wife.’  b. Marion var hrifin  af Bítlunum.  Marion was enchanted of Beatles ‘Marion liked the Beatles.’ c. Beta kunni jósönginn. Beta knew national.anthem.the ‘Beta knew the national anthem.’  Just like in English, the event sentences in the simple past are ambiguous between a habitual reading and an episodic reading. Stefán might have been an opera singer who sang opera on a regular basis, but it is also possible that he sang opera only that one time. The stative verbs in the simple past, however, are not ambiguous: (132c) simply means that at some point in the past Beta knew the national anthem. Non-agentive events have the same pattern as agentive events. They are habitual in the simple present but ambiguous in the simple past. 56  (133)  a. John falls into the well (on a regular basis). b. Mary finds four-leaf clovers (whenever she goes for a walk).  (134)  a. John fell into the well (yesterday/on a regular basis). b. May found four-leaf clovers (yesterday/whenever she went for a walk).  And yet again, we see the same pattern in Icelandic.  (135)  a. Jón dettur (reglulega) í brunninn. Jón falls  (regularly) in well.the  ‘Jón (regularly) falls into the well.’  b. María finnur (reglulega) fjögurra laufa María finds (regularly) four  smára.  leaves clovers  ‘María (regularly) finds four-leaf clovers.’  (136)  a. Jón datt (reglulega) í brunninn (í  gær).  Jón fell (regularly) in well.the (in yesterday) ‘Jón (regularly) fell into the well (yesterday).’  b. María fann  (reglulega) fjögurra laufa smára (í  María found (regularly) four  gær).  leaves clovers (in yesterday)  ‘María (regularly) found four-leaf clovers (yesterday).’  The achievement sentences, just like other events, are habitual in the simple present but ambiguous in the simple past. So all the event verbs behave in the same way in this context and we do not get a division between control and non-control verbs. In sum, the habitual test really does work as a stative test.  2.2.2.9 Sequence of tense Portner (2003:481) shows that states and events differ when we have embedded sentences like the one in (137a), where the simultaneous reading of the embedded past is possible but it is not in (137b) (Portner 2003:481-482):  57  (137)  a. John said that Mary was upset.  (stative complement)  b. John said that Mary read Middlemarch.  (eventive complement)  In (137a), Mary being upset can either precede or overlap with the time of John saying that she is upset, but in (137b) Mary’s reading of Middlemarch must have preceded John’s saying she did. The former is called the simultaneous reading and the latter the shifted reading. Abusch (1988) and Ogihara (1989) attribute this contrast to “whether the embedded past tense contributes to the semantics of its clause” (quoted from Portner (2003:482)). We can use this as a test for stativity as states should get the simultaneous reading whereas eventives should get the shifted reading:41  (138)  a. John said that Mary ate the apple. b. John said that Ben ran 14 kilometers. c. John said that Liz broke the window.  (139)  a. John said that Mary was hungry. b. John said that Ben loved his wife. c. John said that Liz hated school.  As predicted, the events in (138) have to have taken place before John talked about them, whereas the states in (139) could either have taken place before John talked about them or they can overlap with John’s speech.42 Now let us consider Icelandic:  41  Glasbey (1998:106) has given the example in (i) as an example of a sentence where a stative predicate does not have an overlapping reading: (i) Max was happy when I arrived. Here Max might have been happy before I arrived (and continued after I arrived) or he might become happy when I arrived. This would not work in Icelandic where the stative verb vera ‘be’ would be used for the overlap reading but the achivement verb vera ‘become’ would be used for the non-overlap reading.  42  Lisa Matthewson (p.c.) has pointed out to me that she finds it very hard to get anything other than the simultanous reading with the sentence in (139a) and would require a past perfect for the shifted reading. She finds it easier to get the shifted reading with (139b) and (139c) but both would, nevertheless, need an adverbial or a clear discourse context, for instance where the wife is dead and so the loving could not be simultaneous. Thus, obviously not all English speakers agree on the judgement of these sentences. However, they do show a clear difference between states and events and therefore do work well as a stative test.  58  (140)  a. *Jón sagi a María át epli. John said that María ate apple.the ‘John said that María ate the apple.’  b. *Jón sagi a Benni hljóp 14 kílómetra. John said that Benni ran 14 kilometres ‘John said that Benni ran 14 kilometres.’  c. *Jón sagi a Jón  Lísa braut gluggann.  said that Lísa broke window.the  ‘Jón said that Lísa broke the window.’  In Icelandic the simple past tense in the embedded clause is ungrammatical and instead the perfect is required to show the sequence of events.43 In those sentences the event time precedes the time of Jón’s speech, just like in English.  (141)  a. Jón sagi a  María hefi éti epli.  Jón said that María had  eaten apple.the  ‘John said that María had eaten the apple’  b. Jón sagi a Jón said  Benni hefi hlaupi 14 kílómetra.  that Benni had  run  14 kilometres  ‘Jón said that Benni had run 14 kilometres.’  c. Jón sagi a  Lísa hefi broti gluggann.  Jón said that Lísa had  broke window.the  ‘Jón said that Lísa had broken the window.’  Unlike with events, the simple past tense can be used with states, but then we can only have an overlap of events. The sentence is not ambiguous as in English:  (142)  a. Jón sagi a  María væri svöng.  Jón said that María was hungry ‘Jón said that María was hungry.’ 43  If the main clause is in the past tense, the embedded clause is in the past perfect, but if the main clause is in the present tense, the embedded clause is in the present perfect. 59  b. Jón sagi a  Benni elskai konuna sína.  Jón said that Benni loved  wife.the self’s  ‘Jón said that Benni loved his wife.’  c. Jón sagi a  Lísa hatai skólann.  Jón said that Lísa hated school.the ‘Jón said that Lísa hated school.’  For the state to precede the time of John’s speech the embedded sentence has to be in the perfect:  (143)  a. Jón sagi a  María hefi veri svöng.  Jón said that María had  been hungry  ‘Jón said that María had been hungry.’  b. Jón sagi a  Benni hefi elska konuna sína.  Jón said that Benni had  loved wife.the self’s  ‘Jón said that Benni had loved his wife.’  c. Jón sagi a  Lísa hefi hata skólann.  Jón said that Lísa had  hated school.the  ‘Jón said that Lísa had hated school.’  Notice that non-control events behave here like other events so this is not a test for control:  (144)  a. John said that Ben fell into the well. b. John said that Mary won the race. c. John said that Anna died.  (145)  a. Jón sagi a  Benni *féll/hafi falli í  Jón said that Benni fell/had  brunninn.  fallen into well.the  ‘John said that Ben fell into the well.’  b. Jón sagi a  María *vann/hafi unni hlaupi.  Jón said that María won/had  won  run.the  ‘Jón said that María won the race.’  60  c. Jón sagi a  Anna *dó/hafi dái.  Jón said that Anna died/had died ‘Jón said that Anne died.’  All these events precede the time of John’s speech and so non-control events do not differ in any way from other events when it comes to sequence of tense. Even though Icelandic and English do not behave in the same way in these examples, the important thing is that in neither language do states and events behave in the same way. So this test works as a stative test.  2.2.2.10 The perfect Portner (2003:462-463; referring to Bauer 1970) discusses how when we have an event in the present perfect the event time has to precede the speech time, whereas states allows the state to overlap the speech time, resulting in a continuative reading of the sentence. He gives the following examples:  (146) a. Mary has run a mile. Mary has run for twenty minutes. Mary has reached the finish line. Mary has slept today.  b. Mary has been running for twenty minutes. Mary has understood the issue. Mary has been angry all day.  The a-sentences here are eventive and the event time fully precedes the speech time, whereas the bsentences are stative and the event time can either overlap with the speech time or precede it. Portner claims this shows that events and states relate differently to speech time. Notice that Portner includes the progressive here with statives but he excludes all (non-progressive) stage-level predicates, such as sleep. Let us apply this test to a few more verbs:  (147)  a. Mary has eaten for twenty minutes. b. Ben has run 14 kilometres. c. Liz has broken the window.  61  (148)  a. Mary has been hungry (before/since noon). b. Ben has loved women (many times/since he hit puberty). c. Liz has hated school (before/since last year).  As predicted the events in (147) have all taken place before the speech time, whereas the states in (148) could either have taken place before the speech time or they can overlap with it.  (149)  Paul has reached the mountain (many times).  Now let us consider Icelandic:  (150)  a. María hefur éti epli. María has  eaten apple.the  ‘María has eaten the apple.’  b. Benni hefur hlaupi 14 kílómetra. Benni has  run  14 kilometres  ‘Benni has run 14 kilometres.’  c. Lísa hefur broti gluggann. Lísa has  broke window.the  ‘Lísa has broken the window.’  Just like in English, these sentences require the event to have finished before the speech time. In Icelandic the simple past tense in the embedded clause is ungrammatical and instead the perfect is required to show the sequence of events. Also, just as in English, stative sentences can, but do not have to, get the continuative reading of the perfect:  (151)  a. María hefur veri svöng (oft María has  og  mörgum sinnum/sían í gær).  been hungry (often and many  times/since  ‘María has been hungry (many times/since yesterday).’  b. Benni hefur elska konur. Benni has  loved women  ‘Benni has loved women.’ 62  in yesterday)  c. Lísa hefur hata skólann Lísa has  (á stundum/sían í fyrra).  hated school.the (on times/since  in last.year)  ‘Lísa has hated school (on time/since last year).’  As Portner predicts an event in the present perfect has to precede the speech time, whereas a state in the present perfect can either precede or continue into the speech time. Non-control events behave here like other events:  (152)  a. John has fallen into the well. b. Mary has won the race. c. Anna has died.  The events of falling, winning the race or dying all precede the speech time as they would with regular control events. The same applies to Icelandic:44  (153)  a. Jón hefur dotti í Jón has  brunninn.  fallen into well.the  ‘Jón has fallen into the well.’  b. María hefur unni hlaupi María has  won  run.the  ‘María has won the race.’  So this is an example of a proper stative test where control does not matter.  2.2.3  The results of the tests  Let us look at a table with the results of the test.  44  There is no point in giving the Icelandic equivalent to ‘Anna has died’ as it cannot be said with the perfect and instead Icelanders would say Anna er látin ‘Anna is dead’. 63  Table 2:1: Results of the stative tests  Progressive Imperatives Do-something Occurrence in the complements of persuade and remind Occurrence with manner adverbs like deliberately, etc Occurrence with for-phrases Occurrence on either side of instead of Habitual meaning in the simple present Sequence of tense The perfect  Stative verbs */ * * * * * * * * *  Event verbs control non-control  */  *  ?  *  ?  ?  ??        The last three tests, habitual meaning in the simple present, sequence of tense and the perfect, all make a clear distinction between states and events and count as real stative tests. Most of the other tests, however, test for control rather than stativity and there the non-control event verbs pattern more often with states than with events. Sag (1973), Dowty (1975), Boertien (1979) and Mufwene (1984), to name a few, had already argued that most of Lakoff’s tests were more about control or agentivity, rather than stativity, and Lee (1971:8), in fact, used some of these tests specifically to test for agentivity: namely the imperative test, the manner adverbs test (which he calls intentional adverbs), and occurrence in the complements of persuade and remind. Furthermore Ross (1972:105) used the do-something test to test for agentivity and he actually argued that the notion of agent can be replaced by the notion “possible subject of do”. Even though the first seven tests are not sufficient for categorizing stative verbs from event verbs, they do give us some ideas about stative verbs. I have already discussed how the imperative-test, persuade-test and the for-test show that stative verbs do not in general have control. Events, however, can have either control or non-control subjects. Many achievements are non-control and some activities are as well, such as those with inanimate subjects. Therefore, failing the control tests may not make a verb stative, but passing such a test makes it highly unlikely that it is a stative verb. Similarly, failing the do-test may not make a verb stative but passing it clearly shows that it is not. In other words, failing the control tests and the do-test are necessary conditions for being a stative verb but not sufficient ones. States, therefore, cannot be seen as something that is done. None of the tests focuses on how stative verbs have to hold at the speech time if they are in the present tense, but unlike events that get a habitual reading in that construction the states simply hold. So we can say that states have to apply at the speech time if the sentence is in the present tense. Sequence of tense and the pefect both show how states do not move the time forward. The sequence of tense test shows that states do not move time as an imbedded verb in the past gives a simultaneous reading and not a shifted reading and the same can be seen with the perfect where the perfect shows a simultaneous reading unlike events that order events in time.  64  We can draw these conclusions together and form a description of stative verbs: 45  (154)  Stative verbs • Cannot have control subjects (imperative-test, persuade-test, for-test). • Cannot be seen as something that is done (do-test). • Have to apply at the speech time if the sentence is in the present tense (habitual test). • Do not move the time forward (sequence of tense, the perfect).  2.3 The difference between stative verbs and event verbs In this section I will discuss the difference between stative and eventive verbs. In section 2.3.1 I will introduce the Davidsonian and the Neo-Davidsonian accounts. Section 2.3.2 will focus on the discussion of whether states have the Davidsonian argument position or not, mostly referring to arguments from Katz (1997, 2000, 2008), Maienborn (2004) and Mittwoch (2005). Kratzer’s (1995) analysis is introduced in section 2.3.3 and 2.3.4 concludes the section.  2.3.1  The Davidsonian and Neo-Davidsonian accounts  2.3.1.1 The Davidsonian account As I will argue in this dissertation that stative verbs in the progressive have in fact been coerced to being eventive verbs, the difference between states and events is quite important. One of the ideas that have been presented in the literature is that events but not states have an underlying argument position that picks out the event in question (see e.g. Katz (1997, 2000, 2008) and Maienborn (2004)). This event argument was originally proposed by Davidson (1967) but he never discussed states. Parsons (1985, 1989, 1990) and Higginbotham (1985, 1996), however claimed that states do indeed have a Davidsonian argument as well but that the state argument differs somewhat from the event argument. I will follow Parsons and Higginbotham, but before I discuss why, we should look at Davidson’s original proposal.46 Before Davidson’s (1967) analysis, most philosophers saw a sentence like (155) as a five-place predicate with the arguments being ‘Jones’, ‘the toast’, ‘in the bathroom’, ‘with a knife’ and ‘at midnight’: 45  It has been pointed out (for instance by Dowty 1979, but similar things have been said by many others) that stative verbs have neither dynamicity nor agentivity so this is by no means a new description of the stative predicate.  46  An alternative was pointed out to me by Paul Portner (p.c.) and that was to make events a parameter of interpretation. That is an interesting idea but I choose to follow Davidson in this dissertation in making the events an argument of the verb. 65  (155)  Jones buttered the toast in the bathroom with a knife at midnight.  However as Davidson argued (originally pointed out by Kenny (1963)), by analysing the sentence as a five-place predicate we obliterate the logical relation between (155) and the sentences in (156) which are all entailed by (155):  (156)  a. Jones buttered the toast in the bathroom with a knife. b. Jones buttered the toast in the bathroom. c. Jones buttered the toast with a knife. d. Jones buttered the toast.  Furthermore, if the adverbials are arguments of the verb then the verb butter in (156a) is a different verb from the verb butter in (156d) as the former is a four-place predicate whereas the latter is only a twoplace predicate. Davidson argued that instead we have here a three-place relation between nominal arguments and an implicit event argument, see (157).47  (157)  butter yxe.butter (e,x,y)  A sentence like (158a) is assigned the logical form of (158b):  (158)  a. Jones buttered the toast. b. e[butter(e, Jones, toast)]  The so-called Davidson’s entailment is accounted for by this proposal:  (159)  Jones buttered the toast in the bathroom  Jones buttered the toast  According to Davidson the expression in the bathroom is true of the same event that the butter-predicate is true of. So if there was an event of buttering the toast by Jones in the bathroom then there was an event which was a buttering of the toast by Jones. This can be put more symbolically in this way:  (160)  47  e[butter(e, Jones, toast)  In-bathroom(e)]  e[butter(e, Jones, toast)]  Intransitive verbs are thus two-place relations and ditransitive verbs are four-place relations. 66  Notice that this entailment is an instance of the general schema:  (161)  (p  q)  q  Landman (2000:12) has summed up Davidson’s theory beautifully:  (162)  1. Verbs have an implicit argument 2. Modifiers apply to this argument as co-predicates to the verb. 3. The argument is an event argument. 4. It gets existentially closed over.  In this dissertation I follow Davidson’s analysis, as well as the Neo-Davidsonian analysis by Parsons (1985, 1989, 1990) and Higginbotham (1985, 1996), who have further developed Davidson’s original theory. I adopt Davidson’s theory and therefore I use the event argument instead of, for instance, just places and times. I believe the analysis in the thesis does not depend on it and could also be presented within other systems. However, events are more fine-grained than times and space-locations because two distinct events can have the same time and place and therefore, everything you can do with times/places you can also do with events, but not vice versa. Furthermore, the state argument of stative verbs (which I will be arguing for) may not always have a place and time as some states are timeless, such as mathematical states in cases like two plus two equals four. I do believe, however, that the arguments in the thesis could be redone for other systems if independent reasons showed them to be superior to an event-based framework. I will now discuss the main arguments for the Neo-Davidsonian analyses.  2.3.1.2 Neo-Davidsonian analysis Parsons (1985, 1989, 1990) and Higginbotham (1985, 1996) separately extended Davidson’s proposal, and their additions are usually referred to as the Neo-Davidsonian analysis. The most influential addition they made to the analysis was that grammatical relations are no longer treated as arguments of the verb but are related to the event via two-place predicates representing theta-roles. Parsons (1990) explains the simple sentence in (163) with (164):  (163)  Brutus stabbed Caesar.  67  (164)  For some event e, e is a stabbing, and the agent of e is Brutus, and the theme of e is Caesar, and e culminated at some time in the past.  This can be written out in a formula as (165) (ignoring the tense):  (165)  e[Stabbing(e)  Agent(e, Brutus)  Theme(e, Caesar)]  As many of the arguments for the Davidsonian argument position have been provided by the proponents of the Neo-Davidsonian account I will discuss these two theories somewhat simultaneously, leaving out, for now, the discussion of stative verbs.  2.3.1.3 In support of the Davidsonian argument 2.3.1.3.1  The modifier argument  One of the important arguments for the underlying Davidsonian argument is the non-entailment pattern that occurs when modifiers have been dropped. Consider the following example:  (166)  a. Brutus stabbed Caesar in the back with a knife. b. Brutus stabbed Caesar in the back. c. Brutus stabbed Caesar with a knife. d. Brutus stabbed Caesar.  As expected, a entails b and c, but b and c combined do not entail a. Brutus might have stabbed Caesar in the back with an ice pick and in the thigh with a knife. In that case both b and c are true but a is not. All entail d. Before Davidson such modifiers sometimes assumed to apply to times rather than events and (166a) would have said something like: There exists a time t such that Brutus stabbed Caesar in the back at t and Brutus stabbed Caesar with a knife at t. This means that there might actually be two events, one of which Brutus stabbed Ceaser in the back and the other where he stabbed him with a knife. According to Landman (2000), adverbials can be dropped while preserving the truth of the sentence, as long as they are predicates of event variables that are bound by the same existential  68  quantifiers. However, if there are two existential quantifiers, i.e. two events, they cannot be merged into one.48  2.3.1.3.2  Perception verbs  The Davidsonian account has an easy way of handling sentences whose main verb is a perceptual verb taking as its argument a clause having the structure of a simple sentence missing its tense, such as bare infinitives.49 These kinds of verbs had caused linguists some problems before. Parsons (1990) argues that sentences like (167) are telling us that the subject perceives a certain event that is of the sort picked out by the embedded clause.  (167)  Poppaea saw Brutus leave.  Spelled out in detail we get the following (from Landman 2000:25):  (168)  e[See(e)  Experiencer(e)=p  e’[Leave(e’)  Agent(e’)=b  Theme(e)=e’]]  The strength of Davidson’s theory becomes clear when we consider why (167) does not entail (169):  (169)  Poppaea saw Brutus leave and Caesar come in or not come in.  The Neo-Davidsonian answer, according to Landman (1990:26) is that “the reason is the difference of the events seen: the event of Brutus leaving is not the same as the event of Brutus leaving and Caesar coming in or not coming in, which may very well not be an event at all.  2.3.2  The Davidsonian argument: do stative verbs have it?  Another extension of Davidson’s proposal by Parsons (1990), Higginbotham (1985, 1996), Landman (2000) and Mittwoch (2005) concerns stative verbs. Davidson limited his discussion to event verbs but Parsons and Higginbotham proposed that stative verbs also have a Davidsonian argument, and a stative  48  Landman (2000) claims that the strength of the adverbial argument for the Davidsonian theory lies in the analogy with adjectives. Basically, if we accept the argument that adjectives are predicates conjoined with the predicate they modify and predicated of the same argument, the parallels force us to assume the same for adverbial modifiers. That implies that verbs have an implicit argument of which both the verb and the modifiers are predicated. This is the Davidsonian argument.  49  Barwise (1981) has a similar theory to that of Davidson. 69  sentence like (170a), therefore, has the logical form of (170b), instead of the simpler structure shown in (170c):  (170)  a. John loves Mary b. s[Love(s)  Experiencer(s, John)  Theme (s, Mary)] c. Loves(John, Mary)  Parsons has pointed out that finding good arguments is somewhat difficult, as stative verbs do not allow adverbial modifiers as easily as event verbs. For instance, finding stative verbs with more than one modifier is almost impossible. Katz (1995, 1997, 2000, 2003) and Maienborn (2003, 2004, 2007) have both argued against the Davidsonian argument with stative verbs. Maienborn categorizes them as a special kind of stative verbs that do in fact have an underlying state argument, similar to Davidson’s event argument. Katz (2008), however, freely admits that verbs like sleep, hold and wait have a Davidsonian argument (without saying directly that they are not really statives) but claims that posture verbs like sit and lie are lexicalized both as stative verbs and event verbs and therefore the event version has the Davidsonian argument but the stative verb does not. In the following sections I will go through the arguments for Davidson’s theory in order to see if his arguments also apply to stative verbs.  2.3.2.1 Adverbial modification Stative verbs behave differently from eventive verbs with respect to which type of adverbials they allow; some adverbials do not apply to stative verbs at all, but others do, and therefore we can use these adverbs as evidence for an underlying state argument. In this section I will discuss manner adverbs and locatives.  2.3.2.1.1  Manner adverbs  An analysis of manner adverbs as event predicates, as proposed by Davidson (1967), is generally accepted by those that favour the Davidsonian analysis of events (see for instance Parsons 1990, Kamp and Reyle 1993). A sentence like (171a) is then analyzed as in (171b) (Katz 2008):  (171)  a. John kissed Mary passionately. b. e[kiss(e, John, Mary) passionate(e) past(e)]  70  The manner adverbs themselves are taken to be event predicates and an adverb like passionately is assigned the following denotation:  (172)  Pe[P(e) passionate (e)]  If manner adverbs are event predicates they should not be able to occur with stative verbs if stative verbs do not have a Davidsonian argument.50 In fact, the linguists that have participated in the discussion of whether or not stative verbs have a state argument, all agree that if manner adverbs can occur with stative verbs this provides an important argument for there being a state argument. Similarly, if no such cases exist it provides an argument for stative verbs lacking a Davidsonian argument. What they do not, however, agree on, is whether manner adverbs do occur with stative sentences or not. Landman, Parsons and Mittwoch all say they do, but Katz and Maienborn refute these examples for various reasons. I will now provide some details of this debate. Katz (1997, 2000, 2003, 2008) and Maienborn (2003, 2004, 2007) classify slowly, enthusiastically and revoltingly as manner adverbs and provide examples like (173) which they claim show that statives cannot occur with manner adverbs:  (173)  a. *John resembled Sue slowly. b. *She desired a raise enthusiastically. c. *They hate us revoltingly.  Parsons (1990), Landman (2000) and Mittwoch (2005), however, refer to adverbs like well and quietly as examples of manner adverbs occurring with statives, as shown in (174).  50  Kratzer (1994) suggested that the Davidsonian argument be associated with a specific functional projection, EventP. She also suggested that manner adverbs adjoin to that EventP projection: (i) a. John kissed Mary passionately. b. [John1 [TP past [EventP t1 [Event e [VP kiss Mary]] passionately]]] If statives did not have a state argument they should also lack a projection like EventP (possibly called StateP) and manner adverbs like passionately should have nothing to attach to. Propositional operators that adjoin directly to TP, such as probably, will not help determine whether there is an event argument present or not. Similarly, Musan (2002) argues that temporal adverbials in German adjoin to the sentence at three different places; the tense level, the aspect level and the participle level and therefore such adverbials will not tell us anything about whether there is an underlying state argument or not.  71  (174)  a. Peter knew French well. b. John loves Mary quietly. c. I know John well by face from television.  The big question here is: are the adverbs in (174) indeed manner adverbs modifying states and therefore evidence for the underlying Davidsonian argument with statives? Katz (2008) claims they are not, but gives different reasons for the acceptability of the sentences in (174). (174a), he claims, is grammatical because well is not a manner adverb but a degree adverbial: he gives the following examples in support (Katz 2008:229):  (175)  a. I know John well. b. I know John.  (176)  a. I know John slightly. b. I know John.  The sentence in (175a) entails (175b) for if we drop the adverb well the sentence I know John is still true. However, the same entailment does not really hold in (176) for if I only know John slightly I cannot really say that I know John. This, Katz explains, is because the droppability is tied to the lexical semantics of the adverb and we cannot say that well is associated with the event predicate while slightly is not. Instead he claims these are degree adverbials that do not involve Davidsonian predication and the fact that well can occur with stative verbs does, therefore, not provide support for a Davidsonian argument with stative verbs.51 Now let us look at the example in (174b). It does appear that quietly is a manner adverbial modifying a stative sentence, but Katz claims this is not the case either. He gives the following example:  (177)  a. John loves Mary quietly with all his heart. b. John loves Mary quietly. c. John loves Mary with all his heart.  51  Hotze Rullmann (p.c.) has pointed out to me that slightly can be used as a degree modifier with gradable adjectives (e.g. we were slightly late), but well cannot (*we were well late), at least not in general. So there is clearly more to be investigated when it comes to these adverbs but that falls outside the focus of this dissertation. Furthermore, Laurel Brinton (p.c.) pointed out to me that the non-entailment does not hold for her, and other English speakers that I asked agreed with her. Then, at least for these speakers, slightly works exactly like well and therefore no reason to treat the two differently. 72  Obviously (177a) entails (177b) and (177c), but jointly, (177b) and (177c) do in fact appear to entail (177a), and so the examples show drop but not the non-entailment required by manner adverbs in the Davidsonian framework. Katz argues that because of this we can simply treat those adverbs as predicate modifiers:  (178)  [[quietly]] = Px [P(x)  x is quiet about P(x)]  So according to Katz (2003:228) the set of individuals who love Mary quietly is a subset of those who love Mary, and so is the set of individuals who love Mary with all their heart. The set of individuals who love Mary quietly with all their hearts is then the intersection of the two sets. The example in (174c) has the adverb well, like (174a), which Katz claims is not a manner adverb. However, it also has the adverbial by face. It is an important example for Landman who claims that (179c) shows permutation, drop and non-entailment.  (179)  a. I know John by face from television. b. I know John by face. c. I know John.  Katz claims that (179a) and (179b) do not entail (179c) as knowing someone by face from television is not to know that person, and that it would be false for him to claim that he know Peter Jennings or David Letterman even if he knows both well by face from television. Katz claims that by face “is a type of nonintersective modifier which has particular, lexically-specific entailments”, and suggests that the reading is in some sense “collocational”.52 He points out that the droppability is clear with event verbs:  (180)  a. John recognized Bill by face. b. John recognized Bill.  However, I am not certain of the non-droppability of (179c). What exactly is it to know someone? Does not the following conversation sound perfectly normal?  (181)  A: You know that actor from Sense and Sensibility, colonel Brandon….what’s his name...Alan Rickman… B: Yes, I know him.  52  Katz does not define how exactly he uses the term intersective but most likely uses it in the same way as Kamp and Partee (1995), as he takes as an example of that if something is red and it is a book, then it is a red book, making the adjective ‘red’ and example of an intersective adjective. 73  It is the lack of non-entailment enough to classify these adverbials as anything other than manner adverbials? The ultimate difference between statives and events might on its own explain why statives do not show non-entailment the same way as events do. In the case of Brutus stabbing Caesar, it seems natural to assume that there could be two stabbings where one is in the back with a knife and the other one, which takes place at the same time, is in the thigh with an ice pick. So the reason why nonentailment is an argument for a Davidsonian event argument is the fact that we might have two stabbings. Technically we have the same with a stative as we we might have two states of John loving Mary where in one he loves her quietly but only half-heartedly, and in the other state he loves her with all his heart but in a much more obvious (and public) way than the quiet loving. If that were the case, then (174b) and (174c) would not entail (174a). However, because of the nature of love we do not expect there to be two loving states between the same people at the same reference time, and therefore any modifiers are seen as modifying the same state of love. This does not provide evidence against a Davidsonian argument with stative verbs. Mittwoch (2005) has argued strongly in favour of the underlying state argument. She has come up with several examples she claims favour the proposal:  (182)  a. Ann resembles Beth uncannily in facial expressions. b. Ann resembles Beth uncannily. c. Ann resembles Beth in facial expressions. d. Ann resembles Beth.  (183)  a. Ann is related to Beth by blood on her mother’s side. b. Ann is related to Beth by blood. c. Ann is related to Beth on her mother’s side. d. Ann is related to Beth.  Here the a-sentences entail b and c but b and c combined do not entail a, just the pattern expected if statives had the underlying Davidsonian argument. However, Katz does not agree. He points out that in the case of (182), if Ann and Beth have only their facial expressions in common, but otherwise they do not look like each other at all – Ann is tall but Beth is short, Ann is blonde but Beth a brunette – then (182a) would be true but (182b) and (182d) not. So in that case the entailment does not hold either way and this cannot be seen as a support for the underlying state argument. I am not certain that Katz is correct here. We frequently talk about people resembling each other without them sharing most physical features. Even if Ann resembles Beth only in facial expressions I think we can safely say that she does resemble Beth, just not in every way.  74  The example in (183) is a bit different as here the Davidsonian pattern is clearly shown; a entails b, c and d, but b and c do not entail a. However, Katz claims that the adverbials cannot be regarded as manner adverbs in these sentences as they are modifying relations. According to him, “whether two individuals are related is determined by the existence of (chains of) two sorts of events – conceptions and marriages. For x to be related to y, is in general, for there to be a chain of conceptions or marriages that relate x to y” (Katz 2003:233). So Katz claims that modifiers like by blood, by marriage, etc. specify the type of relation explicitly by specifying what properties these relevant events have. They are second-order properties of the personal relation and this is not Davidsonian modification and cannot be seen as an argument for underlying states. If one accepts Katz’s arguments for by blood and by marriage as nonmanner adverbs, then these examples cannot count as arguments for the Davidsonian argument with states. However, it is a question how fine grained the classification of adverbs should be and whether relational adverbs like this cannot be classified as a kind of manner adverbs. Other examples provided by Mittwoch (2005) in favour of the Davidsonian argument with stative verbs are given in (184):  (184)  a. The car is illegally parked in front of my house. b. The carpet was irreparably damaged. c. Dan is in the country illegally.  Mittwoch argues that the state in (184a) results from an event of parking but not necessarily an event of illegal parking; that is to say, the parking job may have taken place during a time when parking was permitted. Therefore there was no event of illegal parking but we now have a state of the car being illegally parked. In (184b), what is irreparable is the state of the carpet, not the event that lead to the carpet being damaged, and in (184c) it is Dan’s being in the country that is illegal. Katz again disagrees with Mittwoch’s analysis and points out that even though (185a) does not entail an illegal parking event, illegal and in front of my house have simple intersective meanings.  (185)  a. The car is illegally parked in front of my house. b. The car is illegally parked. c. The car is parked in front of my house.  Following Parsons (1990) Katz argues that as the three properties that characterize true Davidsonian manner modification are permutation, drop and non-entailment, they cannot be simple intersective modifiers, and therefore, any intersective modifiers that do not show the three Davidsonian properties cannot be seen as arguments for a Davidsonian argument position. Therefore, since (185a) does in fact entail both (185b) and (185c), it should not be seen as an argument for a Davidsonian argument position 75  with states. The problem here is, however, the same as has before been mentioned with states, that it is hard to have two states of the same kind at the same time and the same place. So if the car is parked infront of the house and it is parked illegally, then it is hard to get that as two separate states. So again, the nature of states might be the reason for why the non-entailment arises. Although, (184a) and (185) might not be good examples of a stative as ‘parked’ is similar to posture verbs and locatives which do behave differently from other statives. This was discussed briefly in chapter 1 and will be discussed in more detail in chapter 5. That is not the case with the example in (184b) where the damage of the carpet is irreparable. Let us suppose that the examples in (184) really are cases of manner adverbs with statives. Why are such examples not more common then? Parsons, Landman and Mittwoch have all argued, and I agree with them, that the reason why modification with stative verbs is so uncommon is because of the nature of statives; that is, manner adverbs rarely occur with statives for completely different reasons than that of a lacking Davidsonian argument. Manner adverbs are used to describe the manner of some performance or activity, how something was done, and as such are simply not compatible with states that simply are. However, I do believe that Mittwoch has successfully shown that manner adverbs can sometimes occur with stative verbs, which should be valid evidence for a Davidsonian argument position with stative verbs. This whole discussion depends on manner adverbs being treated as predicates over events. However, not everyone agrees with manner adverbs being treated in such a way. Geuder (2000; quoted from Arregui and Matthewson 2001) for instance has argued that psychological manner adverbs are predicates over individuals (at least in some interpretations) and not over events, and Arregui and Matthewson (2001) agree with him, based on St’át’imcets (Lilloet Salish). In fact, Arregui and Matthewson argue that locative and temporal modifiers in St’át’imcets differ considerably from manner modifiers in being event predicates inside the VP while manner adverbs or at least psychological manner adverbs predicate over ordinary individuals. They go further than Geuder in saying that psychological manner adverbs never predicate over events. Landman and Morzycki (2003) also go a different way. They model manner in terms of an independently motivated notion, namely kind; they were the first to propose event kinds as an ontological category. Just like Arregui and Matthewson they draw a distinct line between manner modification on the one hand and locative and temporal modification on the other. Temporal and locative adverbials, they say, generally restrict a set of events to having taken place at a particular time or place in a given world, whereas manner modifiers are event kind modifiers. These are just two suggestions linguists have made for treating manner adverbs in a different way from locatives and temporal modifiers, which clearly shows that not everyone agrees with manner adverbs being predicates of events. As manner adverbs in general fall outside of the topic of this dissertation I will leave the discussion of the nature of manner adverbs for later studies. 76  2.3.2.1.2  Locatives  Locatives have been used to argue against the existence of a state argument as stative verbs, other than posture verbs and a few other verbs like wait and sleep, do not usually occur with them. Maienborn (2001, 2003) categorizes stative verbs into two groups: she keeps the name statives for predicates like know French and love but calls verbs like sit and stand state verbs. Her division is very related to the division between individual-level states and stage-level states (see further section 2.3.2.4) but is not exactly the same. She uses locatives to distinguish her state verbs from her stative verbs and introduces three kinds of locatives: event-external, event-internal and frame-setting:  (186)  a. Eva signed the contract in Argentina.  (Event-external)  b. Eva signed the contract on the last page.  (Event-internal)  c. In Argentina, Eva still is very popular.  (Frame-setting)  The event-external locative in (186a) relates to the Davidsonian event argument and refers to the place where the signing of the contract took place. The event-internal locative in (186b) is similar as it also relates to the Davidsonian argument but it does not express a location for the whole signing event but only for a part of it: Eva’s signature. The third locative, the frame-setting locative, shown in (186c), differs considerably from the other two, according to Maienborn. It sets a frame for the proposition expressed by the rest of the sentence. Maienborn argues that, in addition to the locative reading, framesetting locatives can have a temporal reading, and event-internal locatives can have an instrument or manner reading. Event-external locatives, however, she claims, can only have a locative reading. According to Maienborn, frame-setting modifiers are not part of what is properly asserted but instead restrict the speaker’s claim. Hence omitting such modifiers may not necessarily preserve the truth of the sentence:  (187)  In Argentina, Eva still is very popular. Eva still is very popular.  If Maienborn is correct it should not be unusual to see frame-setting locatives with stative sentences, but event-external locatives should not occur with statives if statives do not have a Davidsonian argument, which is what Maienborn assumes. Maienborn claims this is in fact the case and that every time we find locatives with stative verbs, they are always of the frame-setting kind. This is why sentences such as in (186c) and (188) are grammatical (Maienborn 2001) but the sentences in (189) are not (Maienborn 2003):  77  (188)  a. In Bolivia, Britta was blond. b. Carol war im  Auto  Carol was in.the car  (Maienborn 2001) müde/hungrig. (Maienborn 2003) tired/hungry  ‘Carol was tired/hungry in the car.’  (189)  a. *Das Kleid is auf der Wäscheleine nass. The dress is on the clothesline  wet  ‘The dress is wet on the clothes line.’  b. *Carol war (die ganze Zeit) vor Carol  dem Spiegel blond/eitel/intelligent.  was (the whol time) in.front.of the  mirror  blond/vain/intelligent  ‘Carol was (all the time) blonde/vain/intelligent in front of the mirror.’  Maienborn claims that the sentence in (188a) has to have a frame-setting modifier as it allows for a temporal interpretation such that at some/every time when Britta was in Bolivia, she was blond. She points out that the most appropriate question for (188a) is actually ‘When was Britta blond?’ rather than ‘Where was Britta blond?’ The same can be said about (188b). The question of Carol’s hunger really is about when she was tired/hungry but not where.53 If that is the case, then this cannot be seen as an example of a locative modificiation with states but a temporal one – and none doubts that temporal adverbials can occur with states. Mittwoch (2005) claims that event-external modifiers can get a temporal reading, just like the frame-setting modifiers, and therefore the temporal reading of the sentence in (188b) should not automatically make it a frame-setting modifier. Instead she claims it is a case of an event-external modifier and therefore an argument for a Davidsonian argument with statives. For one, the locative can be dropped: If Carol was hungry in the car, it entails she was hungry. Additionally, the only reason why the locative in (188b) can be interpreted temporally is because the location seems to be irrelevant for Carol’s being tired or hungry. To further her point Mittwoch (2005:76) gives the following examples:  53  It has been pointed out to me that (189b) with ‘vain’ works in English: (i) Carol was vain in front of the mirror. Not all English speakers I asked agree with that. Most said it was completely ungrammatical whereas some said it was not. Those that accepted the sentence claimed that the sentence could be an answer to both questions ‘where was Carol vain’ and ‘when was Carol vain’. So either ‘in front of the mirror’ should be categorized as a frame-setting locative as well, or we do have an example of a nonframe-setting locative with a stative verb.  78  (190)  a. Carol was hungry in your house; the food is better there than in ours. b. Kate felt nauseous in the car. c. John felt faint in the stuffy room. d. Once Mary was ecstatic on the top of a mountain.  Mittwoch’s answer is that many locatives in clear eventive settings can also be paraphrased with a whenclause, and so that should not count as an argument for such locatives being frame-setting:  (191)  a. Bob solved the equation on the train. b. Max wrote that book in France. c. I thought about the square root of -2 in the waiting room.  Mittwoch suggests that preposed locatives may indeed be of a separate category, perhaps frame-setting locatives as Maienborn suggests, but that locatives that follow the verb and its complements have to be bona fide locatives. I suggest that the reason for why states are generally not modified by locatives is not because they lack a state argument, but because states are not limited to places in the same way as events often are. If John loves Mary, that is a state that holds whether John is in London, Paris or Rome. So saying that John loves Mary in London sounds odd. If we do, however, imagine that John’s love of Mary is dependent on where he is, for instance if he loves Mary only when he is in London, but he loves Monique while in Paris and Sofia while in Rome then ‘John loves Mary in London’ appears to be a perfectly natural sentence. Therefore I suggest that instead of saying that states cannot occur with place adverbs, they rarely do because of the nature of states. When the nature of the state, however, allows for such a locative then locational modification is grammatical, which must indicate the presence of a state argument.  2.3.2.2 Perception verbs Parsons’ second argument for the underlying event argument had to do with event verbs as objects of perception verbs. He had some problems finding similar examples with stative verbs but suggested this may be because putative states are in general not easily observable. He imagines a futuristic situation where people have developed indirect ways to test for such statives. In that world it might be possible to say:  (192)  For two continuous hours we watched the patient hate her mother.  79  The lack of stative verbs with perception verbs may not have a linguistic reason but have more to do with the way the actual world is. Katz (2000:405-406) gives an example where he imagines a situation where it had been established that Mary was drunk all day yesterday and that Bill saw Mary yesterday. In such a case Bill would actually be stating a falsehood were he to say:  (193)  I have not seen Mary drunk.  So even though Bill did not actually realize that Mary was drunk the sentence is still false. Katz’s argument is that with an event one can either refer to seeing the person or seeing the event. However he claims that with a stative verb one cannot distinguish between the person and the state she is in, so even if the speaker does not realize that he saw Mary when she was drunk, he cannot distinguish between Mary and her state, and therefore he is stating a falsehood if he claims he did not see her drunk. Katz argues that the same pattern cannot be observed if instead of being drunk Mary kept blinking her eyes continuously yesterday and Bill saw her yesterday. After all, he may have seen her from behind and so he may be truthful if he said:  (194)  I have not seen Mary blinking her eyes.  However, I am not convinced this difference lies in the state-event distinction. What if we refer to an event that is as hard to observe as Mary’s drunkenness in Katz’s example? Take for instance Derek Zoolander from the movie Zoolander who was famous for his facial expressions. His new, groundbreaking look was called Blue Steel, even though it happened to look exactly like his other facial expressions. Someone not realizing this might then say:  (195)  A: I have not seen Derek do Blue Steel. B: Yes, you have, it is just exactly like La Tigra.  Assuming that putting on a facial expression can be seen as an event rather than a state, we have here an example parallel to the stative example in (193) where the person claims to see the person but not the event. When John fails to realize Mary’s drunkenness he might nevertheless see the state; he just does not realize that it is a state of drunkenness he is seeing. Parsons might be correct in that the reason why we do not usually get stative predicates as complements of perception verbs is because states are not easily perceived.54 What we would perceive 54  Paul Portner (p.c.) has suggested using vaguer predicates like detect and claims that these should be all right according to my reasoning: 80  would be actions or events related to the state. Therefore we cannot use this as a proper test for a state argument, but I also believe that the general lack of such examples cannot either be seen as evidence against a state argument.  2.3.2.3 Anaphora Katz (1997) argued that event sentences lead a semantic double life. On the one hand they behave like indefinite noun phrases as they can introduce entities into the discourse model (Kamp 1981), and on the other hand they can introduce propositions or facts.  (196)  a. Smith stabbed Jones. It happened at noon. b. Smith stabbed Jones. That bothers me.  The pronoun it refers back to the stabbing event itself whereas that refers to the fact that such an event occurred. Katz claims that these anaphors refer to distinct objects. It, for instance, cannot be said to refer to a fact:  (197)  a. *The fact that Smith stabbed Jones happened at noon. b. The event of Smith stabbing Jones happened at noon.  However, predicates such as bother do take facts as subjects:  (198)  The fact that Smith stabbed Jones bothers me.  This difference is easily explained under the Davidsonian approach. The event anaphors refer to the underlying event that is the event argument of an event verb whereas the fact anaphors refer back to the entire proposition. So if stative verbs have a state argument we would expect the same duality with stative verbs, something that Katz claims is not the case. However, the problem here is that it is very hard to test whether we have a state anaphor or not, as we cannot use predicates like happen or occur, as Katz points out. Mittwoch (2005:71), however, provides the following example:  (199)  a. Traffic was at a standstill. This was caused by a power failure at the intersection. b. John and Bill are at loggerheads. This is due to their supporting rival candidates for the chair.  (i) *I detected Mary running. (ii) *I detected Mary drunk. However, the fact that in that context the event verb is equally ungrammatical as the stative verb indicates there must be some independent reason for why the sentences are bad. 81  Vendler (1968) and Davidson (1967) have argued that causal effects are states (or events), as opposed to facts or propositions, and if they are correct the italicised pronouns in (199) must be anaphoric to a state. What is caused by the power failure at the intersection in (199a) is not the fact that traffic was at standstill but the state of traffic being at standstill. Nor is the fact of John and Bill being at loggerheads in (199b) due to their supporting rival candidates but the state of them being at loggerheads. So the anaphors in (199) really do appear to refer to statives. This is evidence for the underlying state argument.  2.3.2.4 Results The arguments for an underlying Davidsonian argument position with stative verbs are not particularly strong, but neither are the arguments against it. There are obviously some parallels between states and events, and there are also some major differences. Table 2:2 draws the results together.  Table 2:2: Arguments for an underlying state argument.  Manner adverbs Locatives Perception verbs State anaphora  Support the underlying state argument? Yes, but very few examples. Yes, but only stage-level states. No, but do not provide arguments against it either. Yes, but very few examples.  If we assume there is an underlying state argument with stative verbs, just like there is an event argument with event verbs, we can explain why there are parallels between the two. However, there is nothing to say that the state argument has to be fundamentally the same as the event argument. It may well exist but nevertheless differ considerably. That would explain why there are some differences between statives and events. As the arguments do not provide strong support either for or against stative verbs having a Davidsonian argument, it is necessary to consider both options in relation to the theory presented in this dissertation. Even though I have chosen to follow the Neo-Davidsonian theory in assuming that states do have a Davidsonian argument, the main ideas of this dissertation, namely the shift from statives to eventives (see chapters 3 and 4), would not be hugely affected even if we came to the conclusion that no Davidsonian argument existed with stative verbs. We would then say that a verb like love was simply a two-place predicate that took a subject and an object and so a sentence like (200a) would be represented by (200b):  (200)  a. John loves Mary. b. Loves (John, Mary)  82  Event verbs would still have a Davidsonian argument and we could either follow the Davidsonian theory, shown in (201b) or the Neo-Davidsonian analysis, shown in (201c):  (201)  a. John eats cake. b. Eat(e, John, cake) c. e[eat(e)  Agent(e, John)  Theme(e, cake)  I will assume a Neo-Davidsonian analysis of states but my main proposals could still work within a different system (although perhaps more complicatedly).  2.3.3  Kratzer’s analysis  In the debate about whether states are Davidsonian or not, Kratzer (1995) takes a position that falls in the middle, in the sense that she argues that some stative verbs have a Davidsonian-style argument, but others do not. In fact, she argues that stage-level predicates have a spatio-temporal argument, similar to that introduced by Davidson, but that individual-predicates do not. It was Carlson (1977) who first introduced the notions of stage-level predicates and individuallevel predicates. Stage-level predicates are true of temporal stages of their subjects so if John is sick it can be seen as lasting a certain length of time but not (necessarily) for his entire life. Individual-level predicates, however, are true throughout the existence of an individual. For example, if John is short he will most likely continue to be short all his life – shortness is a property of John as an individual, and not just of a stage of John. Therefore, whereas to be sick is a stage level state, to be short is an individuallevel state. Kratzer uses the stage-level/individual-level distinction to account for the different behaviour between these two classes of stative predicates in when-cluses, which is illustrated in (202).  (202)  a. *When Mary knows French, she knows it well. b. When Mary speaks French, she speaks it well.  She treats when-clauses in a similar way as Lewis’ (1975) treatment of if-clauses where the operator to be restricted can be a determiner quantifier, an adverb of quantification or a modal operator of any kind (Kratzer 1995:129). By assuming that we have an implicit universal quantifier with the when-clauses in (202) and by assuming that stage-level predicates but not individual-level predicates introduce a variable that can be bound by the universal quantifier, Kratzer can explain why (202a) is ungrammatical but (202b) is not.  83  Following a natural prohibition against vacuous quantification (originally suggested by Chomsky (1982) as Kratzer points out) only sentences with variables over which the quantifier can quantify, are grammatical. In (202a) the main predicates in both the antecedent and the consequent are individual-level predicates and as there is no Davidsonian argument introducing the variable, according to Kratzer, the quantifier has no variable to bind and the sentence is ungrammatical (Kratzer 1995:131). (202d) differs from (202a) only in the fact that speak is a stage-level predicate and not an individual-level predicate like know in (202a). Assuming that Kratzer is right in that stage-level predicates have a free variable and individual-level predicates do not, then only (202a) will violate the prohibition against vacuous quantification and therefore be ungrammatical. (202d) on the other hand satisfies the prohibition with the Davidsonian free variable (Kratzer 1995: 131). Kratzer (1995:127) showed how sentences with stage-level predicates have two readings with locatives whereas sentences with individual-level predicates have only one:  (203)  ...weil  fast  alle Antragsteller in diesem Wartesaal  ...since almost all  petitioners  in this  saßen.  waiting.room sat  a. ‘...since almost all of the petitioners in this waiting room were sitting.’ b. ‘...since almost all the petitioners were sitting in this waiting room.’  (204)  ...weil  fast  alle Schwäne in Australien schwarz sind.  ...since almost all  swans  in Australia  black  are  ‘...since almost all swans in Australia are black.’  According to Kratzer (1995:128) “the spatial or temporal expression modifies the restricting predicate of the quantifier fast alle” (‘almost all’) in (203a) and (204a), but in (203b), however, “the spatial or temporal expression modifies the main predicate of the sentence”. In the Davidsonian account the verb introduces an event variable, and the modifiers that the verb takes, impose restrictions on this variable. By assuming that stage-level predicates have the Davidsonian argument but individual-level predicates do not, Kratzer can explain why spatial expressions can modify stage-level predicates but not individuallevel predicates. There is another possible explanation of this data that does not rely on a difference between stage-level and individual-level predicates and that is to say that the individual-level predicate black does have a Davidsonian argument but the relevant reading would require swans to change their colour based on location: In Australia they are black, if they fly over to New Zealand they turn white. I do not, therefore, believe that this works as an argument for individual-level predicates not having a Davidsonian argument. As Kratzer does not argue for any difference between the Davidsonian argument with states on the one hand and with events on the other, she must assume that all stage-level predicates have the same 84  kind of Davidsonian argument. As I will argue in chapter 3 that stages are needed in order to occur in the progressive, and that only the event argument has stages, we would expect to see all stage-level predicates in the progressive. And yet be hungry and own a dog cannot do that:  (205) a. *John is being hungry. b. *Mary is owning a dog.  Whether states can occur in the progressive or not does not seem to depend on whether they are stagelevel or individual-level predicates, which counts against Kratzer’s analysis that they differ in whether they have a Davidsonian predicate or not. I will argue that both stage-level and individual-level predicates have a Davidsonian argument but that we have more than one kind of this argument. Events have an eventive Davidsonian argument whereas states have a stative Davidsonian argument. The nature of these arguments differs somewhat as will be discussed in chapter 3. There is also a possibility that a further distinction can be made, based on whether we have a stage-level or individual-level predicate but I will leave further speculations about that for later studies. However, if all eventualities do indeed have a Davidsonian argument then the data given by Kratzer needs to be explained differently. I have already suggested a possible explanation for the locative data but do not, at this moment, have an answer to the data with the when-clauses. Until there is one, Kratzer’s data is a problem for my analysis.  2.4 Conclusion In this chapter I have focused on stative verbs in both English and Icelandic and shown how difficult it is to define the category of the stative. Many of the tests that have been provided in order to separate statives from eventives rather focus on features like control, but other tests really do work in separating the category of statives from events. In chapter 3 I will argue that when we use stative verbs in the progressive the state is shifted to an event, which involves shifting the state argument to being an event argument. This would mean a relatively simple shift. If we followed the Davidsonian theory where stative verbs do not have a Davidsonian argument then the shift has to introduce that argument and we have to assume a considerably more elaborate shift rule.  85  3 States in the progressive 3.1 Introduction As discussed in chapter 2, linguists such as Poutsma (1926), Lakoff (1966, 1970), Visser (1973) and many others have claimed that states cannot occur in the progressive in English, and the same has been argued for Icelandic (e.g. Einarsson (1967), Comrie (1976), and ráinsson (1974, 1999)).  (206) a. *John is knowing the answer. b. *Jón er a elska konuna sína. Jón is to love wife.the self ‘Jón is loving his wife.’  The explanations given for why stative verbs do not occur in the progressive vary considerably. A good overview can be found in Brinton (1987:207):  For example, it is said that states are already durative and hence the progressive is superfluous (e.g. Palmer 1974:71-74) or that since states lack endpoints, they cannot be temporally limited by the progressive (e.g. Joos 1964:107-108, 113, 119). Most convincing is the explanation that states cannot occur in the progressive because they lack dynamicity (e.g. Lyons 1977: 707-708; Comrie 1976: 51; Hirtle 1967:69ff). See also the discussion in Hirtle (2007). Furthermore, Katz (1995, 1997, 2000, 2003) has argued that stative verbs are predicates of times (unlike events which he claims are predicates of events) and tense can, therefore, apply directly to them instead of aspect applying first. The progressive, therefore, cannot apply to stative verbs. However, even though most speakers of English and Icelandic agree that the sentences in (206) are bad, Hirtle (1967, 2007), Mufwene (1984), Brinton (1988), Hirtle and Bégin (1991), Gachelin (1997) Kakietek (1997), mieciska (2002/2003), ráinsson (2005), Torfadóttir (2004) and many others have pointed out that it is in fact not so hard to find stative verbs in the progressive in Icelandic (ráinsson, Torfadóttir) and English (Hirtle, Mufwene, Brinton, Hirtle and Bégin, Kakietek, mieciska). ráinsson gives the following examples (2005:489) from Icelandic:  (207)  a. Vi vorum a halda a We were  to think  ú kæmir ekkert.  that you come  not  ‘We were thinking you weren’t coming.’  86  b. ú  ert bara ekki a skilja  You are just  not  etta.  to understand this  ‘You’re just not understanding this.’  Similar examples can easily be found in English:  (208)  a. Jen Pearson is liking the newest Stars album. (Facebook) b. Now he is understanding that there are things in the world that exist even though they are not in his present environment. http://www.becomingtheparent.com/subsections1/question21.html  I will propose that “true” stative verbs do indeed not occur in the progressive. However, we can shift states to events and as such, these shifted states take on many of the prototypical features of events; they are then also required to occur in the progressive when in the present tense, in order to achieve a non-habitual reading. I will lay this out in detail in the following sections as well as discuss the prerequisites for this shift. In this chapter I will also discuss some cases where Icelandic and English differ; namely when it comes to weather verbs, locatives, posture verbs and the active be. All these verbs can occur in the progressive in English but not in Icelandic. The chapter is organized as follows: In section 3.2 I discuss the use of states in the progressive and in section 3.3 I focus on the semantics of these states in the progressive, including the coercion of states to events. In section 3.4 I will discuss the prototypical eventive properties and implicatures, as well as how stative verbs can be shifted to being eventive in order to convey these different properties. 55 Section 3.5 focuses on states in the progressive in Icelandic, and particularly the ScanDiaSyn project, and in section 3.6 I discuss the difference between Icelandic and English. Results of the chapter are presented in 3.7.  3.2 The use of states in the progressive Even though stative verbs occur in the progressive in English they are not particularly common in the progressive compared to how often they occur in the simple present/past. I did a search through the Corpus of Contemporary American English (Davies 2008-) where I looked for 14 stative verbs in the first  55  I assume Grice’s (1975) theory of conversational implicatures.  87  person singular, present and past tense, both in the simple present/past and in the progressive. The results are presented in Table 3:1.56  Table 3:1: Stative verbs in the first-person singular in COCA  Lie Sit Stand Live Hear Hope Feel Imagine See Remember Love Hate Understand Like Want Know Believe  simple # 542 2174 1751 3886 7597 19261 23714 2496 22136 17112 23982 6255 12170 19549 63697 78481 26939  Present tense simple % prog # 91.3 52 91.6 198 92.1 151 97.9 84 99.1 70 99.4 118 99.5 109 99.6 9 99.7 69 99.93 12 99.95 12 99.97 2 99.98 3 99.98 4 99.92 5 99.95 4 100.0 0  prog % 8.7 8.4 7.9 2.1 0.9 0.6 0.5 0.4 0.3 0.07 0.05 0.03 0.02 0.02 0.008 0.005 0.000  simple # 1452 5673 3986 2388 15251 1392 22302 1601 27431 2731 5176 1517 1911 4188 26507 30322 1884  Past tense simple % prog # 82.2 315 82.8 1174 84.2 746 79.5 614 98.8 187 58.7 979 96.9 707 97.7 37 98.6 376 97.8 61 99.8 12 99.7 5 99.6 7 99.91 4 99.92 21 99.97 1 99.8 4  prog % 17.8 17.2 15.8 20.5 1.2 41.3 3.1 2.3 1.4 2.2 0.2 0.3 0.4 0.09 0.08 0.003 0.2  Most of the stative verbs in the present tense occurred well under one percent in the progressive, except for the posture verbs sit, stand and lie which occurred in the progressive ranging from just under 8% to just under 18%.57 The verbs live and hope which occurred in just over 2% and just under 1% of the time in the present tense, respectively, but 20% and 41% of the time in the past tense. This is quite a difference between the posture verbs and live and hope on the one hand and the other stative verbs on the other hand. There is also a difference between the present and the past, particularly with respect to live and hope, but in general statives are much more common in the progressive in the past tense than in the present tense.58 I have no idea what could explain this difference and am not aware of any studies that 56  When choosing the verbs I made sure to choose a mixture of posture verbs, locative verbs, perception verbs, intellectual verbs and affective verbs. In order to make sure that we only had the verb in the first person I only used the instances of the verbs that occurred with the first person pronoun I.  57  It is worth noting that even though sentences like these are found in English and are not that uncommon, they can also hardly be said to be very common and in no way do all speakers of English use such constructions nor does everyone who does use it, use it with every verb. A detailed survey would be in order. Until then the table in 3:1 will have to do.  58  I also did a small COCA study in the present tense to see if states in the progressive were more common in the first person than in third person, and found out that they were more common in the 3rd person in all cases except for lie and sit. In the third person the verb hope occurred in the progressive in 4.57% of cases, and the verbs feel, imagine and understand occured in over 2% of the times, whereas none of these verbs reached 1% in the 1st person singular, present tense. I would be interested to do a more detailed study of this and then include the second person as well.  88  have ever looked into it. It would certainly be an interesting task to explore further the relationship between the prorgressive and the tenses. In order to evaluate what these low numbers of states in the progressive really mean, it is important to point out that the progressive in general is not as common as one might think, even with eventive verbs. I also did a second search through The Corpus of Contemporary American English (COCA) (Davies 2008-) where I looked at how often a few event verbs occurred in the progressive in comparison to the simple present and past. The results are shown in table 3:2: Table 3:2 Event verbs in the first-person singular in COCA  Run Paint Walk Eat Ask Begin Sell Change Explain  simple # 1274 471 2145 1142 7596 1163 327 325 512  Present tense simple % prog # 94.4 75 92.7 37 97.4 57 97.9 25 97.7 178 92.4 95 95.6 15 94.8 18 99.6 2  prog % 5.6 7.3 2.6 2.1 2.3 7.6 4.4 5.2 0.4  simple # 3530 354 5470 1122 19127 6086 592 689 1121  Past tense simple % prog # 88.6 452 84.5 65 89.4 650 87.0 167 98.6 271 93.3 435 91.8 53 95.6 32 97.4 30  prog % 11.4 15.5 10.6 13.0 1.4 6.7 8.2 4.4 2.6  As table 3:2 shows, in the present tense, first person, event verbs are used in the progressive only 0.37.5% of the time, although in the past tense the range is from 1.4-15.5% of the time. The verbs paint, run, walk and eat all occurred in the progressive just over 10% of the times they occurred in the first person, past tense, but less than 8% of the time in present tense, and, in fact, walk and eat occurred less than 3% of the time in the progressive. The event verb explain occurred only 0.4% of the times in the progressive in the present and 2.6% of the time in the past. These numbers are quite low and it is important to keep that in mind when looking at stative verbs in the progressive. When we compare these numbers to states in the progressive we see that the verb sit is more commonly used in the progressive than any of the event verbs that I counted, and the verbs stand and live are well within the average range of event verbs in the progressive in the present tense and above it in the past tense.59 The fact that the posture verbs pattern so clearly with the event verbs in this respect, and not with state verbs might be a clear indication that, at least in English, posture verbs really are eventive and not stative. I will discuss posture verbs further in section 3.6.1. Other stative verbs (with the exception of hope in the past tense) are much less common in the progressive than event verbs. From this we learn that stative verbs in the progressive are by no means  59  It is important to keep in mynd that I only counted the nine event verbs shown in table 3:2 and so there might be other events that are much more common in the progressive. It would be an interesting project to compare different event verbs in the progressive in order to see whether certain classes of events verbs are more common in the progressive than others. 89  common but they do occur and they occur often enough for us not to be able to ignore them. This is supported by mieciska’s (2002/2003) survey of the progressive. mieciska (2002/2003) did a survey amongst young, educated, native speakers of American English in one particular university in Pennsylvania where she either presented the students with a sentence involving a state in the progressive, or an incomplete sentence where the students chose whether to fill in the missing stative with a verb in the simple present or in the progressive. Her results were that a verb like hear in the progressive was considered acceptable by 93% of the speakers, intending was accepted by 71%, doubting by 64%, understanding by 61%, appearing by 57% and wanting by 54%. Other stative verbs tested ranged from 24% (looking) to 48% (costing), with the exception of liking which was only considered acceptable by 7% of the speakers and knowing which was not accepted by anyone. The difference between these numbers and the ones I received in the simple counting task of COCA most likely lies in the fact that even though a speaker might accept a verb in the progressive as acceptable he might not always choose to use it. We see a clear example of that in the verb like which appeared three times in mieciska’s survey. The first time it appeared it got accepted by 7%, in the second sentence by 36% and in the third sentence by 30%. In the first two sentences it was given as the only option but in the third sentence the subjects could choose between using the verb in the progressive or in the simple present. No corpus-study has been done examining the progressive in Icelandic, so we do not know how common the progressive is in general and how common states are in the progressive. We also do not have any hard numbers of whether states are now being used more frequently in the progressive than before. However ráinsson (2010) has shown that younger people are more likely to accept sentences with states in the progressive than the older generation, whether that is a reliable indication for increase of states in the progressive or not. In mieciska’s (2002/2003) study all the students were a similar age and so the study did not compare the use of the progressive based on age, as was done in ráinsson (2010). However, the question whether the use of states in the progressive has increased over time or not has been discussed by many linguists who do not agree on the answer. Potter (1975:120) claims that the use of states in the progressive has increased, although he does not do a survey or a corpus study. Kranich (2010:172) agrees with Potter, unlike Leech et al. (2009:130) who argue that at least in printed English, “use of the progressive with stative verbs did not contribute substantially to the growing use of the construction between the 1960s and 1990s, in either regional variety.” Mair (2006:92) argues that the reason why it appears that a higher percentage of stative verbs appear in the progressive nowadays than in the past, is because the type of informal context in which it happens is less likely to be preserved in the past. This is actually a fairly believable explanation with modern media such as the Internet spreading informal discourse in ways that never existed before, and making that text available to broader readership.  90  Linguists often correctly point out that the frequency of stative verbs in the progressive is quite low but from this section we have learned that even though this is in fact the case it must be kept in mind that event verbs are not that common in the progressive either, compared to their use in the simple tense. Whether the occurrence of states in the progressive is on the rise, is still unsettled. The important thing for this dissertation, however, is not how frequently states are used in the progressive but the fact that they can occur in that construction, so let us now turn to the semantics of states in the progressive.  3.3 The semantics of states in the progressive 3.3.1  From lexical entry to time of evaluation  In chapter 2 I argued that states really are a separate category from events and that three tests can be used to clearly show that. Furthermore, I concluded in favour of the Davidsonian argument occurring with stative verbs just as with event verbs, but that argument is ontologically distinct from the event argument and should not be seen as identical to the event argument. Basically, following Davidson (1967), an event sentence like (209a) has the simplified logical form of (209b). A stative sentence like the one in (210a) gets the representation in (210b):60  (209)  a. Brutus stabbed Caesar b. e[Stabbing(e)  Agent(e, Brutus)  Theme(e, Caesar)]  (210)  a. John loves Mary b. s[Love(s)  Experiencer(s, John)  Theme(s, Mary)]  I argued in chapter 2 that eventive verbs and stative verbs are different sorts of entities that belong to non-overlapping subsets of the domain D. I use e as a variable over events and s as a variable over states and so the lexical entries of eventive and stative verbs can be written as in (211) and (212):  (211)  stab: yxe[Stab(e)  Agent (e,x)  Theme(e,y)]  (212)  love: yxs[Love(s)  Experiencer (s,x)  Theme(s,y)]  Following Kratzer (1998), Iatridou et al. (2001), von Stechow (2001) and Katz (2003), I assume that in English and Icelandic, the aspectual operators are part of the non-lexical vocabulary of the 60  Simplified in the sense that no tense or aspect is shown in this formula. 91  language and apply to event properties to yield temporal properties. In the syntactic structure operators may be assumed to head the Aspect Phrase. As I mentioned in chapter 1, I will be using Landman’s analysis of the progressive (Landman 2000), so let us repeat here his definition of the progressive operator:  (213)  [[PROG(e,P)]]w,g = 1 iff fv:f, v  CON(g(e), w) and [[P]]v,g (f) = 1 where CON(g(e), w) is the continuation branch of g(e) in w.  Landman ignores tense in his formula but we need to bring it in here. I will, therefore, assume that once PROG applies to a VP we do not yet have a truth condition (as is implied in (213)); rather, once PROG applies, we have a predicate of times. I therefore present an amended progressive formula: 61  (214)  [[PROG(VP)]]w,g,c = tee’w’ [<e’,w’>  CON(e,w)  [[ VP]]w’,g,c (e’) = 1  t  (e)], where CON(e,w) is the continuation branch of e in w.  When we apply this new version of PROG to (215a) we get (215b):  (215)  a. Brutus is stabbing Caesar. b. [[PROG (Brutus stab Caesar)]]w,g,c = tee’w’ [<e’,w’>  CON(e,w)  Stab(e’)  Agent(e’, Brutus)  Theme(e’, Caesar)t  (e)]  Tense, unlike Aspect, is an anaphoric element, not unlike pronouns (see Partee 1973). I use the following definitions from Heim (1994):  (216)  a. [[PASTi]]w,g,c is defined only if g(i) < tc, in which case [[PASTi]]w,g,c = g(i) b. [[PRESi]]w,g,c is defined only if g(i) o tc, in which case [[PRESi]]w,g,c = g(i)  With the tense added, we can give the LF of (215a) as in (217). (218) represents the resulting truth conditions when (217) is evaluated with respect to an assignment g and a world w:  (217)  [TPBrutus1[TPRES2[AspPPROG[VPt1stab Caesar]]]]  61  I have changed Landman’s e to e to follow Kratzer who introduced existential closure over events in the Aspect head. Furthermore, unlike Landman I have PROG be the denotation of –ing directly. I have also added the c parameter representing context, as I will need it in later formulas. 92  (218)  ee’w’ [<e’,w’>  CON(e, w)  Stab(e’)  Agent(e’, Brutus)  Theme(e’, Caesar)  g(2)  (e)], where g(2) o tc  As already mentioned in chapter 1, having stages is necessary in order for a predicate to occur in the progressive according to Landman’s (1992) theory. However, Landman never discusses stative verbs in his progressive theory and therefore never raises the question whether states have stages or not. I argued, nevertheless, that based on his definition of stages, states cannot have stages as they do not develop and each instant is the same as the next. And if states do not have stages, they do not have a continuation branch and cannot occur in the progressive. Yet, we have plenty of examples of states in the progressive, as shown in (207) and (208). So we could either conclude that Landman’s theory of the progressive is wrong, or that states do indeed have stages. But another option, and the one that I will pursue here, is that when we have states in the progressive they have been coerced to being events, and as such do have stages. The rest of this chapter is an exploration of the semantics and pragmatics of the shift from state verbs to event verbs. There is in fact a precedent for this kind of shift, namely that of Rothstein’s achievement shift, which I will take as a model in proposing my own coercion of states to events.  3.3.2  Rothstein’s shift theory  Rothstein (2004) bases her account of achievement verbs in the progressive on Landman’s progressive theory. She points out that she needs a theory that explains why progressive achievements introduce the imperfective paradox but at the same time why they behave differently from accomplishments. She gives the following example:  (219)  Mary is arriving at the station.  According to Landman’s analysis of the progressive (given in (213)), (219) says there is an event e going on in world w which is a stage of an event e’ in the continuation branch, which is in the denotation of ‘arrive at the station’. However, as Rothstein points out, this makes (219) false as the VP is an achievement. Rothstein claims that achievements do not have stages. In order for (219) to be true we need to assert that there is an event going on which will culminate in Mary’s arrival at the station if it is not interrupted. However, the process stage, which warrants this assertion, cannot be a stage of an event in the denotation of ‘arrive at the station’, because that is an achievement, and as such it is instantaneous and does not have stages. To solve this problem Rothstein proposes a type shifting operation, which shifts the achievement meaning to a derived accomplishment meaning. She assumes a set of events e, which are the sum of an activity e1 whose particular properties are lexically unspecified and a culmination event e2 93  which is in the denotation of the lexical achievement. This is the shift rule that Rothstein (2004:48-49) gives:62  (220)  SHIFT(VPpunctual): e.(BECOME)(e)  e.e1e2[e=s(e1e2)  (DO())(e1)  (BECOME(P))(e2)  Cul(e)=e2]  Here the denotation of the VP is changed into the structure of an accomplishment whose culmination is given by the lexical VP. When we apply (220) to ‘arrive at the station’ we get:  (221)  SHIFT(e.ARRIVE AT THE STATION(e)  Th(e)= x) =e.e1e2[e=s(e1e2)  (DO())(e1)  ARRIVE AT THE STATION(e2)  Th(e2)=x  Cul(e)=e2]  When the output of the shift rule in (220) is the argument of PROG, We get the following:  (222)  PROG(e,SHIFT(ARRIVE AT THE STATION)) = PROG(e, e’.e1e2[e=s(e’1e2)  (DO())(e1)  ARRIVE AT THE STATION(e2)  Th(e2)=x  Cul(e’)=e2])  The denotation of (219) is then given in (223):  (223)  a. e[PROG (e, e’.e1e2[e=s(e’1e2)  (DO())(e1 )  ARRIVE AT THE STATION(e2)  Th(e2)=MARY  Cul(e’)=e2]) b. e’w’:<e’,w’>  CON(e,w)  [e’1e’2[e’=s(e’1e’2)  (DO())(e’1)  ARRIVE AT THE STATION(e’2)  Th(e’2)=MARY  Cul(e’)=e’2]  This can be read in the following way: ‘Mary is arriving at the station’ is true in a world w iff there is an event e going on in that world which has on its continuation branch an event which culminates in an event in the denotation of ‘Mary arrive at the station’. It is crucial to note, according to Rothstein, that e is not a stage of the arrive-at-the-station event. So let us now look at how we can use Rothstein’s achievement shift as a model for our coercion of states to events.  62  Here e1e2 is the sum of e1 and e2 and s is the operation forming a singular entity out of this sum. 94  3.3.3  Coercion of states to events  Many linguists have expressed the idea of states in the progressive being interpreted as events even though the details are not always the same. Smith (1983:497), for instance, claimed that progressive statives present a state as if it is an event and as such endows the state with the properties of events, which is quite similar to what I will argue for in section 3.4. Brinton (1987:208ff) has similar ideas as she points out that states, which are not dynamic, are presented as dynamic when occurring in the progressive, and Zucchi (1998) even uses the words ‘aspect shift’ when he claims that a stative verb in the progressive behaves like a non-stative verb. One of the explanations given (e.g. Dowty (1979) is that some verbs are lexically ambiguous between two kinds of readings and what appears to be a stative verb in the progressive is not a stative verb at all but an event verb, homonous to the stative verb but with a slightly different reading. A slightly different analysis has been argued for by Moens and Steedman (1988), who claim that only processes (what we call activities) can occur in the progressive and that other aspectual classes need to be coerced to being a process in order to occur in the progressive. They give examples of culminated processes (accomplishments) and atomic events (achievements) but do not mention statives. However, no one has so far formulized this so that is what I will try to do now. For coercion of states to events I propose a simple shift in the style of Rothstein, where a stative verb is shifted to being an event verb with the denotation of an activity. Let us say, for instance, that we have a state of loving. It takes with it two arguments, the one who loves and the one that is loved, and it is a typical state in that it cannot be seen as something that is done and it does not move the reference time forward, nor does the subject have control over the state. However, the speaker expressing the state can choose to look at it as an event instead, for instance in order to convey some of the prototypical properties associated with events that will be discussed further in the next section. To do so he shifts the state to an event with a special shift rule. For the purpose of the shift rule I introduce the function EV which maps a state onto the corresponding event. The shift-operator changes a set of states into a set of events. If VP is a stative predicate then SHIFT(VP) is the corresponding eventive predicate. I am assuming that for every state there is a unique corresponding event, or at least for the states that can be shifted.  (224)  [[SHIFT(VP)]]w,g,c= es[s  [[VP]]w’,g,c e=EV(s)]  The rule says that SHIFT(VP) denotes the set of all those events which are the result of applying EV to some member of the denotation of VP. As argued in chapter 2 there is an ontological difference between states and events and the Davidsonian arguments that occur with them. The event argument, unlike the state argument, includes stages. When a stative verb has been shifted to an event it appears in a sentence as an eventive verb, and as such it has an event argument which – like any event – comes with stages. When SHIFT is applied to ‘love this song’ we get:  95  (225)  es[s  [[love this song]]w’,g,c  e=EV(s)]  We can then give the semantics of (226a) in (226b):  (226)  a. Mary is loving this song. b. [[Mary PRES2 PROG love this song]]w,g,c = ee’w’s [g(2)  (e) <e’,w’>  CON(e, w)  e’=EV(s) s  [[love]]w’,g,c  Exp(s, Mary) Theme(s, this song)], where g(2) o tc  ‘Mary is loving this song’ is true in a world w relative to an assignment g and a context c iff there is an event e whose runtime includes the time of c (i.e., the utterance time) and which in w has a continuation branch <e’,w’>, where e’ is the event-correlate of a state s (i.e., EV maps s onto e’) such that s is a state of loving in w’ and the experiencer of s is Mary and the theme of s is this song. I would have liked to think of the coerced event as the state viewed in a different way but chose to treat it as a separate entity as it is easier to formulate such a shift. However, now that we have established what kind of shift takes place when states are coerced to being events, we should take a look at what might be the motivation for the shift. I will argue for the existence of prototypical properties associated with events and that they play an important role in understanding the pragmatics of the shift. This will be the topic of section 3.4.  3.4 Eventive implicatures In sections 2.2 and 2.3 I discussed the difference between stative verbs and event verbs and argued that even though both have the underlying argument proposed by Davidson, it is not the same kind of argument for states as for events. In fact, states and events are fundamentally different from each other. In this section I am going to argue that there exist certain prototypical eventive properties. When stative verbs are coerced to being eventive it is usually done in order to convey a meaning associated with these eventive properties. I will now explain this idea in more detail.  3.4.1  Proto-Events  As previously argued, when we have stative verbs in the progressive the verb has been coerced to being an event verb. I will argue in this section that all event verbs have some prototypical eventive properties to some extent. They do not have to have all of these properties, but the more of them they have the more typical an event verb they are. These eventive properties are normally not associated with stative verbs. However, when a speaker wants to convey one or more of these prototypical eventive properties with a  96  stative verb, he can shift the stative verb to being an event verb and by doing so can implicate these eventive properties. Dowty (1987, 1989, 1991) gives an analysis for theta roles that is similar to the kind of account I will give for states vs. events in the sense that he treats theta roles as prototypical concepts, so I will start by discussing his analysis before I present mine. Dowty treats semantic roles as cluster concepts, which are determined for each choice of predicate with respect to a given set of semantic properties. He makes a distinction between Proto-Agents and Proto-Patients (similar to Foley and Van Valin’s (1984) Actor and Undergoer).  • •  Proto-Agents have the properties of volition, causation, sentience and movement. Proto-Patients undergo change of state, are incremental themes, are causally affected by another participant and are stationary relative to movement of another participant.  Dowty’s analysis is formulated in order to decide which of two possible participants is the subject of a verb phrase and which is the object. He argues that the participant with more agentive properties is the subject. If both participants have the same number of agentive properties then a hierarchy of the properties decides which one is the winner. In this way, for instance, a participant with a causer property would beat out a participant with a sentience property. According to Dowty, Agents and Patients do not have to possess all those properties but the more of the properties they have, the more typical Agents or Patients they are. I am not concerned with theta selection or prototypical Agents and Patients, but I do think some aspects of Dowty’s analysis can be useful when looking at verb classes. In the style of Dowty’s Proto-Agents and Proto-Patients, I am going to argue that there exist certain prototypical eventive properties and that all eventive verbs have at least one of these properties. The more of the properties that verbs have the more prototypical events they are. The speaker can shift a stative verb to an event in order to convey these properties with the stative verb. In order to support this theory we need to first establish what those eventive properties are and then show that stative verbs in the progressive do indeed have one or more of these properties. We can start with our description of stative verbs given in chapter 2, repeated here:  (227)  Stative verbs • Cannot have control subjects. • Cannot be seen as something that is done. • Have to apply at the speech time if the sentence is in the present tense. • Do not move the time forward.  97  Since this description was given after comparing states to events it is not unrealistic to assume that event verbs are basically the opposite:  (228)  Events • Can, and most commonly do, have control subjects. • Can be seen as dynamic. • Get a habitual meaning in the simple present. • Move the time forward.  As states and events have different properties it is not unrealistic to assume that when we coerce a state to an event, it is done in order to convey one of the eventive properties.  3.4.2  Eventive properties  When comparing states and events, the most obvious differences are, as outlined in section 3.4.1, those of dynamicity and control – that is, events prototypically have both whereas states have neither. These two properties are directly related to the descriptive points of events here above. Control connects to the first point, can, and most commonly do, have control subjects. Dynamicity connects to the second point, can be seen as dynamic. I will, therefore, propose that both are prototypical eventive properties. It is, however, harder to connect the last two points to a property: events get a habitual meaning in the simple present, and events move the time forward. However, states do not get a habitual meaning in the simple present and they do not move the time forward, and therefore, states appear to be long-lived, and if we look at events as the opposite we can say they are short-lived, or in fact, temporary. I will, therefore, also propose temporariness as an eventive property. I will now show that this line of thought is correct and that states in the progressive do, in fact, show one or more of these prototypical eventive properties.  3.4.2.1 Control As previously mentioned, control is one of the features Dowty proposed for agentivity. Ever since the work of Gruber (1965) and Fillmore (1968) the definition of the agentive role has been discussed but it is not always clear what linguists mean by the notion of agentivity. However, many will understand it to mean something like volitional control over an eventuality (see e.g. Stirling 1993).63 The agentive  63  Some linguists use the terms volition and control as it seems interchangeably. I generally use the term control but when referring to Partee’s (1977) analysis will use her term of volition. It is important to point out that this notion of control has nothing to do with “control” in GB-syntax (where it has to do with the interpretation of the empty subject (PRO) of infinitival clauses), but is more akin to the concept of „out98  participant, therefore, tends to possess or exercise control over the situation and because of this is usually animate. It is, however, not necessary for an agent to have control over the situation. John can, for instance, break a vase by accident, in which case he is the agent of the event but can hardly be said to have been in control of the situation. If he broke the vase on purpose, however, he is not only the agent but also was in control. Basically, control entails agentivity but agentivity does not entail control. This distinction is not always made clear so there may be cases where a linguist refers to tests for agentivity but in fact it is control that matters, as I argued at length in section 2.2.2. When stative verbs occur in the progressive the thematic role of the subject often seems to change. This is seen very clearly with the verb think, which seems to have both a stative and an eventive reading, where in the former case the subject is an Experiencer but in the latter an Agent:64  (229)  a. I think that some books are more successful than others to certain readers. (COCA:MAG)65 b. I’m thinking about going to medical school to become a trauma surgeon. (COCA:MAG)  The subject in (229a) is an Experiencer who holds a certain belief, whereas in (229b) the subject is an Agent, and in this case also a controller, who is actively thinking about what to become. It is an activity. These different usages of the verb think are quite common but the same difference can be seen with other verbs where the distinction is maybe not as common nor as clear. Consider the differences between the two sentences in (230):  (230)  a. I remember the soldiers who died during my tour in May 2002. b. Today I am remembering the soldiers who died during my tour in May 2002. I came home; they did not. I thought I would forget their faces by now. But I haven’t. I know their names. – KF http://www.cbc.ca/news/yourview/2006/11/remembrance_day.html  Example (230a) is a typical stative sentence where I is the Experiencer.66 There is no agentivity, and therefore no control, and the speaker might even be trying to forget. It does not mean he had forgotten the of-control“ in the Salishanist literature (see for instance Thompson 1979, Davis and Demirdache 2000, Jacobs 2011). 64  In fact, when translated into Icelandic, two different Icelandic verbs are usually used. For think that a verb like trúa or halda is used, whereas for think about Icelanders use the verb hugsa. This might indicate that think that and think about in English are simply homonyms but it is more likely that it is the same verb used in two different meanings.  65  All acronyms that follow reference to COCA, such as MAG, FIC, SPOK, refer to the texts within COCA that the example comes from. For information on what each acronym stands see Davies (2008-).  99  soldiers and now he remembers them; instead it means that he is either doing something in order to remember them, or he is in some way voluntarily calling out the memory of those soldiers. We have here the property of control. Recall that the Experiencer theta role does not have either of those properties, but an Agent can have both. This is very clear in the following debate that took place on the Internet, where the discussion is about whom to pay tribute to on Remembrance Day. The discussion is not about whether people remember that soldiers died on both sides during the war, as it is obviously hard to forget that fact. The discussion is about whether the dead soldiers on the enemy side should be shown any kind of respect by keeping their memory alive in some way.  (231)  A: You realise that the nazi soldiers were just following orders? and that only the generals and officers knew about concentration camps and the other bad stuff? The point i’m trying to make is that we should be remembering those who died on both sides.  B. ...  C: Yah that’s retarded, remembering the German regulars. We kind of killed them. No matter how PC you wanna be, that’s just one of the stupidest things I’ve heard in a long time. I’m remembering the soldiers that died for OUR country. http://forums.massassi.net/vb3/archive/index.php/t-3836.html  The verb remember in (231) is not being used as a simple stative verb, referring to one holding a memory of those soldiers. It is being used as an agentive verb – an event verb. In this meaning remember can occur with the adverb deliberately (see the stative tests in chapter 2) and as a complement of try:  (232)  a. Today I am deliberately remembering the soldiers who died during my tour in May 2002. b. Try to remember the soldiers who died during your tour in May 2002.  In sum, if the speaker wants to indicate that a state has control, he can coerce the stative verb to being eventive and as eventive the former state now occurs in the progressive. In fact, Ljung (1980) has pointed out that “whenever the progressive is used with a predicate normally denoting a state, the goingson expressed by the progressive predicate are always interpreted agentively” (Ljung 1980: 29). I think it is too strong of a claim that these predicates are always agentive, but it is certainly one of the motivations for the coercion. Let us now look at the next property, which is dynamicity. 66  Notice that remember can also be an achievement verb: (i) I suddenly remembered the soldier who died during my tour in May 2002. 100  3.4.2.2 Dynamicity Even though I talk here about event verbs versus stative verbs, it is quite common in the literature to use the term dynamic verbs or action verbs rather than event verbs. Dynamic verbs are said to describe activities or events that in general require movement or motion of some kind. This can easily be seen in all the different kinds of event verbs:  (233)  a. John is running.  (activity)  b. Mary is painting a picture.  (accomplishment)  c. Clive has reached the summit.  (achievement)  I use the term dynamicity fairly broadly in this section to refer to physical movement, change or evolution of an event. It is, however, clearly a prototypical eventive property as none of these features are commonly seen with stative verbs. However, a speaker who wants to imply dynamicity with stative verbs can coerce the stative verb to being eventive and then uses the verb in the progressive. It is also important to keep in mind that from Landman’s discussion of stages we can infer that they are closely tied to dynamicity. That is, having stages means being dynamic, although being dynamic does not necessarily mean having stages, as we can see with achievements. We can coerce states to events in order to show a change within a state, such as something that is increasing or decreasing. That kind of dynamicity can be made clear with adverbials such as ‘more and more’ or ‘increasingly’. See this example from Comrie (1976:36):  (234)  I’m understanding more and more about quantum mechanics as each day goes by.  There is an obvious change in the understanding as it is gradually increasing. Similar gradual changes can be seen with other stative verbs:  (235)  What had begun as a one-mile dash to the U.S. Embassy had now become a full-fledged evacuation from Paris. Langdon was liking this idea less and less. (COCA: FIC)  In Icelandic we see:  (236)  Ó  Vín  er svo falleg!  Oh Vienna is so  Er a elska borgina meir og meir me hverjum degi.  beautiful! Am to love city.the more and more with each  ‘Oh, Vienna is so beautiful! I am loving the city more and more every day.’ http://blog.central.is/bleikirfilar  101  day  Sometimes the dynamicity of the state is hard to grasp and comes about more as a feeling than a clear dynamic reading. Look at the following quote from the Internet, where a person describes his/her feelings for the McDonald’s slogan i’m (sic) lovin’ it: 67  (237) In my opinion, McDonald’s takes liberty with English with this slogan for the sake of effect: With  the  continous  form  there  is  some  dynamic  flavour  to  the  saying.  http://www.usingenglish.com/forum/ask-teacher/89801-im-lovin-3.html  Under dynamicity we could also possibly include cases where the use of the progressive with stative verbs seems to indicate more than a feeling, but rather actions that are brought about because of the feeling, or possibly how the actions show what the subject’s feeling is.  (238)  They are loving the gear right now. (Shaw channel)  In (238) we do not have a simple statement of love but in this case the speaker is indicating successful merchandise being offered that results in the consumers buying more of it. The quote is from a story on the Shaw channel where the host was interviewing a salesperson at the Canucks store, shortly after a new line of clothing was introduced. The salesperson is pointing out how well the clothing is selling and attributes this to the customers loving the gear. Here we do have an obvious change of state in that they now love a gear that they did not love before, or might not have known about before, but we can also say that the love manifests itself in a change in related events, such as them buying more of the merchandise. We have to distinguish between the state of loving and the associated event of loving. The state of loving is having a certain feeling. The event of loving, however, includes either some control of that feeling, dynamicity or temporariness. So by coercing the state to an event we showcase it in a different way, as an event, and thereby implicate that it is accompanied by eventive features. In addition there are events that accompany the loving, such as buying more of the gear, but these events and the now lovingevent are connected but separate events. Hirtle and Bégin (1991:102) argue that change versus no change is the important factor when it comes to states in the progressive. A state involves no change but any non-stative event involves change or the possibility thereof. Basically they argue that a “developing activity is essentially an event involving “successive stages” or phases, an event, therefore, whose lexical elements arise successively in time”. They give the following examples:  67  In the McDonald’s slogan, I is written in lower case. 102  (239) a. I’m hating this house party. (Kruisinga and Erades 1953:260) b. Nan wondered how Simon’s family were liking her and sensed a certain reservation about his mother. (Buyssens 1968:33) c. That’s because you do not know any, said Christopher, who wasn’t liking Lewes at that moment. (Kruisinga 1931:362)  Hirtle and Bégin explain (239a) by claiming that it involves a change of some kind, like the subject forming an attitude. Sentence (239b), they say, refers to a situation of reacting to something in a new way, and (239c) is explained as an incomplete process. I think the intuitions here are in general relatively natural but I also believe we do not have to attempt to explain the slight differences between each example.68 I agree that change is a factor when we coerce states to events, but it is just one of several factors and therefore it is unnecessary to try to explain every state in the progressive in terms of change. The notion of change is closely tied to that of temporariness in the sense that a temporary event does require change – it needs to begin or end. So let us now look at temporariness in more detail.  3.4.2.3 Temporariness Dowty (1979) did not overlook the fact that some states do occur in the progressive in English although he only discussed a very limited subclass of stative verbs – posture verbs and locatives. He gives the following examples:  (240)  a. The socks are lying under the bed. b. Your glass is sitting near the edge of the table. c. The long box is standing on end. d. One corner of the piano is resting on the bottom step.  None of these verbs involve control, as they have inanimate subjects, nor is there any apparent movement or change of state. But even though these verbs are acceptable in the progressive with certain subjects, there do seem to be some semantic restrictions (Dowty 1979:174):  68  This matter of intuition does not just apply to Hirtle and Bégin’s analysis but also to mine. As Paul Portner (p.c.) has pointed out to me, the claims about what each progressive sentence derived from a stative means are pretty subjective. It is a question how we can get beyond opinions and find some evidence about the meaning shifts. At this time I do not have an answer to that but this is a criticism that applies to other common proposals, for instance Dowty’s proto-thematic roles. There is really no proof that the subject is in control of a certain sentence, beyond our intuitions that that is what the verb means. 103  (241)  a. New Orleans lies at the mouth of the Mississippi River. b. ??New Orleans is lying at the mouth of the Mississippi River.  (242)  a. John’s house sits at the top of a hill. b. ??John’s house is sitting at the top of a hill.  (243)  a. The new building stands at the corner of First Avenue and Main Street. b. ??The new building is standing at the corner of First Avenue and Main Street.  (244)  a. That argument rests on an invalid assumption. b. ??That argument is resting on an invalid assumption.  From these and other similar examples, Dowty draws the conclusion that the progressive is only acceptable with these verbs if the subject denotes a moveable object – “an object that has recently moved, might be expected to move in the near future, or might possibly have moved in a slightly different situation” (Dowty 1979:175). Because of this, the context can actually play a role:  (245)  a. ??Two trees were standing in the field. b. After the forest fire, only two trees were still standing.69  The a-sentence is inappropriate because without any context, nothing indicates that it is temporary, whereas in the b-sentence we get the information that the forest fire has burned down all the trees but two.70 Therefore, an obvious change has happened and now, unlike what was there before, there are only two trees standing in the forest. Trees in a forest are usually not seen as being temporarily standing there but the forest fire has changed that. By burning down the trees it shows them in a new light as temporary individuals and the b-sentence makes sense. Dowty also points out that in narrative contexts, these verbs can be used in the progressive if they describe stationary objects that momentarily come into the observer’s view. He gives the following example:  69  It could be questioned whether standing is here still a verb in the progressive or whether we have a deverbal adjective instead, similar to upright. 70  Both Dowty and I predict that the a-sentence becomes acceptable in a context where we consider the possibility that the trees might soon be cut down. This is correct, as shown in (i): (i) Everything was changing. Houses were torn down and built up, trees were cut down, roses torn out of the ground. When I arrived there, two trees were standing in the field, but I knew it would not last for long.  104  (246)  When you enter the gate to the park there will be a statue standing on your right, and a small pond will be lying directly in front of you.  Dowty suggests that the position of the moving observer is taken as the “fixed” point of orientation of the narrative. So with respect to this moving point of orientation the stationary object becomes “temporary”.71 Leech and Svartvik (1975:65) seem to more or less agree with Dowty as they claim that “with states the effect of the progressive is to put emphasis on the limited duration of the state of affairs.” They claim that verbs like live, stay, enjoy etc. in the simple present/past describe a fairly permanent state of affairs whereas in the progressive they describe states of affairs that are rather temporary. However, Mufwene (1984) points out that this leaves many states that cannot occur in the progressive unaccounted for. He gives the sentences in (247) as examples of temporary states and yet the sentences are bad (Mufwene 1984:15).  (247)  a. *Mary is having the flu.72 b. *We are needing him badly. c. *Your whiskey is containing too much alcohol. d. *Your degree is not counting for the competition.  Mufwene argues that if being temporary was enough for states to occur in the progressive these sentences should all be grammatical. And yet that does not seem to be the case. However, I propose that of these four examples, only (247a) and (247b) can be seen as truly temporary. (247c) and (247d) are different. The sentence in (247c) can hardly be seen as temporary as whiskey tends to have the same amount of alcohol the whole time it exists. Thus it is not a plausible meaning that the whiskey has too much alcohol for part of its existence and just the right amount for part of it. If that is the case the sentence should be somewhat better if we can imagine a situation where the alocohol content can be seen as changing, and in fact it is:  (248)  71  This solution is still containing too much alcohol. (You need to dilute it some more.)  It is not clear how Dowty’s proposal could be formalized and further research is required.  72  When I asked native English speakers whether they found this sentence acceptable they all denied it and suggested alternatives like ‘Mary’s got the flu’ or ‘Mary has the flu’. Interestingly, one speaker said: “Mary is having the flu would mean for me you can hear her barfing in the next room.” This fits exactly with the claim that states in the progressive get eventive properties – in this case dynamicity.  105  We can find many similar examples on the Internet where the verb containing is frequently used in the progressive:  (249)  Looks like your controller class is containing to much business logics, so you now have a problem in sharing this logics with the page and a third party caller. http://forum.springsource.org/showthread.php?t=33354  Similarly in (247d) we cannot say that the degree does not count for the competition right now but that it will tomorrow. Either it counts or it does not. So it does not appear that we have a temporary situation here and therefore we cannot use the ungrammaticality of the setences to argue that temporariness is not sufficient to license the progressive. In addition to not being temporary (247c) and (247d) cannot have control or be dynamic, and since none of the eventive properties that drive the coercion of states to events is possible with these verbs, the proposal I am arguing for correctly predicts them to be ungrammatical. Sentences (247a) and (247b), however, do indicate a temporary situation and so should be grammatical in the progressive, as Mufwene correctly points out. In fact, I do not think it is uncommon to hear sentences like (247b). I found several similar examples in COCA, and hundreds of such examples on the Internet.  (250)  Some of my staff are giving each other intravenous nutrition because they are needing to be rehydrated. It has to happen today. (COCA:SPOK)  That leaves only (247a) unexplained, but as Mufwene rightly points out, it should be grammatical if being temporary is enough to occur in the progressive. The temporariness of the progressive states becomes clear when the progressive is contrasted with a state in the simple present:  (251)  a. Pete is living/staying at his parents’ house. b. Pete lives/stays at his parents’ house.  106  Most English speakers agree that the situations described by the progressive are more or less temporary whereas the ones described by the simple present are rather permanent, or at least expected to last for a long time. This is, however, not a necessary reading and the progressive could be used for a permanent situation and the simple present for a temporary one. So obviously the temporariness is not part of the truth conditions of the progressive nor is longetivity part of the truth conditions of the simple present.73 The claim that something is not temporary does not entail that it lasts for all eternity, as obviously that would make most situations temporary. Houses and trees, for instance, do not last for eternity but we can rightly assume that in their “lifetime” each of them stays pretty much in the same place. Of course houses and trees can be moved and rivers can be rerouted, and in such cases their location can become temporary. The examples given above all include locatives. However, the same pattern can be seen with other stative verbs in the simple present and in the progressive:  (252)  a. Kristín loves the Beatles. b. Tim is loving the Beatles.  Most English speakers I have asked agree that the a-sentence gives us a more permanent situation: Kristín’s love for the Beatles is not a fad or a temporary thing. Instead she is likely to have been a fan for a long time. Tim, however, is more likely to be going through a period of loving the Beatles, whereas in general he might be a bigger fan of Pink Floyd. This is the same as with the locative sentences: nothing in the truth conditions says that using the simple present gives us a longer time period than when we use the progressive, and yet, that is how most people understand the sentences. So if nothing in the truth conditions says that the situation is temporary, this must be an implicature. I need to point out that even though temporariness is a property of events and events are in general temporary, this does not automatically make all states permanent. Being in love is a state and yet people fall in and out of love all the time. Being poor is a state and fortunately we do not have to assume that the poverty will necessarily last forever. The difference between states and events in that respect, however, is that all events are predicted to end whereas states are not. States can, in principle, go on forever even if they do not have to. Shifting a state to an event, therefore, does implicate that the state will end, just like an event. I will work this idea out in more detail in the next chapter. It is clear that temporariness is an eventive property, just like control and dynamicity. One of the reasons why speakers coerce a state to an event is to convey this temporariness property.  73  Mufwene (1984:20) points out that the same difference can be seen with non-stative verbs: (i) a. John is working/studying/teaching at UGA. b. John works/studies/teaches at UGA.  107  3.4.2.4 The implicatures I have argued in this section that all event verbs have certain eventive properties such as control, dynamicity and temporariness. However, it is not necessary to have all of these properties in order to be seen as an event verb. Verbs like fall and die, for instance, are dynamic but non-control and yet there is no question that they are eventive. These prototypical eventive properties distinguish events from states, so if the speaker chooses to present a situation as an event, rather than a state (as indicated by the use of the progressive) the only reason she can have for doing so is that she wants to emphasize the presence of one or more of these prototypical eventive properties. So from the fact that the speaker employs the shift from states to events, the hearer concludes that one or more of these properties need to be present. For instance, when a speaker says ‘we are remembering the dead soldiers’ he might be implying not only control and dynamicity but also temporariness: he does not just remember the dead soldiers – he is actually performing some action in honour of that memory, and that action is temporary. The eventive properties do not become part of the truth conditions of the sentence, but are mere implicatures. As states do not have dynamicity, control or temporariness, the verb is reanalyzed as an event verb. But why argue for these prototypical properties being implicatures? If there is a shift, are not the eventive properties then a part of the new event and therefore a part of the truth conditions rather than an implicature? I choose to answer that negatively and here is why. If the eventive properties automatically became a part of the truth condition of the sentence we would expect all eventive sentences to include all eventive properties. We have already seen that this is not the case. However, if it is not necessary to have all the eventive properties to be considered an event we can hardly say that the eventive properties are a part of the truth conditions of the sentence. Instead they are properties closely associated with events but never necessary. Let us take an example. John is asked about the living situation of his friend Peter and he answers by saying ‘Peter is living with his parents’. If all the eventive properties mentioned earlier applied, we would assume this was a temporary, dynamic and agentive situation. That would indicate that Peter was in some way actively and temporarily living with his parents. This does not, however, have to be the case. Peter might in fact always have lived with his parents and it is hard to see what exactly is agentive or dymamic about the situation. However, the hearer will assume that the speaker chose to use the progressive for a reason. At this point it is more or less up to the hearer to read into what the speaker intends, unless the context shows that clearly. In this case the speaker would probably assume that the situation is temporary, or possibly that it involves a change of some kind. Either Peter might just have moved in with this parents or he is about to move out. Without any further context the hearer might in fact not know which of these possibilities is intended by the speaker. The reason that John chooses to  108  shift the state to an event, and therefore uses it in the progressive, is without much doubt exactly because he is implicating one of the eventive properties but only further context will make it clear which one.74  3.4.2.5 Results In this section I have argued that there are prototypical eventive properties and that all eventive verbs have one or more of these properties. The more of the properties the verb has the more prototypical event verb it is. The properties are dynamicity, control and temporariness. When states are shifted to events it is usually done because the speaker wants to implicate one or more of these prototypical eventive properties with the verb. The ScanDiaSyn project in Iceland (ráinsson 2010) shows examples that nicely illustrate the shift and how it can be motivated by the wish (on the part of the speaker) to convey the presence of some of the eventive properties.  3.5 States in the progressive in Icelandic 3.5.1  The project  The ScanDiaSyn project is a joint Nordic project documenting syntactic variation in the Scandinavian languages (ráinsson 2010). One of the constructions examined in the Icelandic project was the progressive as there is a general belief in Iceland that there is an ongoing change in the use of the progressive construction in Icelandic (see e.g. Frijónsson 2003, 2004, 2005, 2006, 2007). The various categories studied were split into four: sports language and iterative/habitual sentences, which I will discuss in chapter 4, and stative verbs and weather verbs, which are discussed in this chapter. 187 native speakers were asked to evaluate various sentences in the progressive. The subjects were given three 74  Is there any chance that the fact that the hearer cannot always be sure which eventive property is expressed by the utterance is due to simple vagueness and is not an implicature at all? Obviously these progressive sentences are vague as saying ‘John is staying with his parents’ is less direct than saying ‘John will only stay with his parents for a short time’, if the speaker wants to convey the message that John’s stay with his parents is temporary. One method to test for implicatures is the cancelability test. Consider (i) (i) Luis is working in the mine. If the temporary meaning is not essential, it can hardly be a part of the truth conditions. This is supported by the fact that we can cancel the implicature: (ii) Luis is working in the mine these days. In fact, he has always worked in the mine, and he probably always will. The implicature of (i) is cancelled in ii) by adding the information that Luis’ job in the mine is not a temporary situation. However, further research is required to rule out beyond doubt the possibility that this is vagueness and not an implicature.  109  response choices: ‘yes’, ‘no’ and ‘unsure’. The subjects were furthermore broken down by gender, age and education. This thorough study gives us a great opportunity to see if my analysis of states in the progressive is correct with respect to the eventive properties.  3.5.2  Acceptability based on different categories  If it is correct that stative verbs are systematically coerced to being eventive in order to convey temporariness, control or dynamicity we would expect higher acceptance rate with sentences that have the chance of showing these properties. I split the data into four categories: simple present tense sentences, used for comparison; verbs that are common in the progressive in English; stative verbs with a clear temporary or inceptive interpretation; and finally verbs that clearly refer to a fairly permanent situation.75 There were two sentences with a stative verb in the simple present, intended as a comparison to the progressive sentences. The sentences in this section are all from ráinsson (2010) and the first sentence within a and b gives the context for the second one.  (253) a. Nönnu finnst leiinlegt í Nanna finds boring  Frakklandi.  in France  ‘Nanna finds it boring in France.’  Hún skilur She  ekki neitt  understands not  í  frönskunni.  anything in French.the  ‘She does not understand French.’  b. Sigrún hefur aldrei átt Sigrún has  bíl.  never owned car  ‘Sigrún has never owned a car.’  Hún kann She  ekki einu sinni a keyra.  knows not  one  time to drive  ‘She does not even know how to drive.’  75  The reason why I divide the sentences based on temporariness vs. permanency, rather than control or dynamicity, is simply because only one sentence in the data could be seen as implying either control or dynamicity. It would certainly be interesting to re-do this experiment where more sentences could be seen to imply the other eventive properties than temporariness. 110  70.9% of the subjects accepted the a-sentence and 88.7% accepted the b-sentence. Both of these sentences are perfectly grammatical simple present sentences so it is a bit of a surprise that the acceptance rate was not closer to 100%. In such cases something other than the choice of aspect might make the subjects rule out a sentence. This is good to keep in mind when we look at the stative sentences in the progressive. In the next group we have sentences where the context makes it natural to assume a temporary situation. We expect the acceptance rate for this group to be fairly high as the temporariness would be the motivation for the shift to an event verb. All the sentences have the verb skilja ‘understand’.  (254) a.  Kristinn er búinn  a vera í  Kristinn is finished to be  einkatímum  upp á síkasti.  in private.lessons up on lately  ‘Kristinn has been taking private lessons lately.’  Hann er vonandi He  a skilja  skuna  betur núna.  is hopefully to understand German.the better now  ‘He is hopefully understanding German better now.’  b.  Jón hefur átt í Jón has  erfileikum me námi.  had in difficulties with study.the  ‘Jón has had difficulties with his studying.’  Hann er bara ekki a skilja He  is just  not  stærfræina.  to understand math.the  ‘He is just not understanding (the) math.’  c.  John flutti  hinga fyrir hálfu ári.  John moved here  for  half  year  ‘John moved here six months ago.’  Hann er a skilja He  íslensku  ágætlega.  is to understand Icelandic excellently  ‘He is understanding Icelandic excellently.’  The most acceptable sentences were (254a) and (254b) with 54% and 46.6% acceptance rates respectively. Both had over 20% unsure. In (254a) we have a case where someone has been getting tutoring and as a result of that is (hopefully) understanding German better than before. The time-frame is, therefore, quite narrow and the sentence refers to Kristinn’s knowledge right now, as in contrast to what it 111  was before he got tutored. In (254b) the reference time is not as clear but it can be understood from the definite article on the noun stærfræi ‘math’ that we are not talking about math in general but the math John has been taking. The context is therefore again limited to a shorter time, such as one term.76 (254c) has no such time limitations and simply refers to a gradable increase of knowledge. So it is natural that it is considered quite worse, with only 30.5% acceptance and 43.1% of the subjects who found it unacceptable. Ten sentences indicated a more permanent state and all received very low acceptance rating, ranging from 6.4% to 24.2%, with one exception. Their unacceptability ranged from 80.6% to 53.7%. The sentences with the lowest acceptability rate are shown in (255): (255) a. Ég arf I  smávegis asto  need a.little  fyrir næsta enskupróf.  assistance for  next  English.exam  ‘I need a bit of help for the next English exam.’ Ert ú a kunna eitthva Are you to know  í  ensku?  something in English  ‘Are you knowing any English?’  b. Stína og Gummi eru löngu  hætt  saman.  Stína and Gummi are long.ago stopped together ‘Stína and Gummi broke up a long time ago.’  Hann er samt He  enn a elska hana.  is nevertheless still to love her  ‘He is nevertheless still loving her.’  It is not surprising that these sentences received such low acceptability as they refer to a fairly permanent situation. In (255a) the question is about the hearer’s knowledge of English, with no indication that it is changing in any way. In (255b) we have a change that took place a long time ago, but the state of Gummi loving Stína has not changed during that time and nothing indicates that it will change. In fact, it refers to the total opposite of temporariness as it shows a state that has existed for a long time and still does, even 76  Notice that in a different sentence from the project where the context sentence says: ‘Ásgeir has been having a bad time with his studying this winter’, and the progressive sentence continues Hann er ekki a kunna neitt í efnafræinni ‘He’s not knowing anything in chemistry’, the acceptability is much lower, or 15%, against 63.7% that reject the sentence. This probably has to do with the difference between the verbs skilja ‘understand’ and kunna ‘know’. To know something can be seen as a more permanent situation than to understand.  112  though the two have broken up. There is neither any reason to assume control nor dynamicity in these sentences, and with none of the motivations for the shift the progressive should not be acceptable. Notice also the difference between the fairly temporary situation of (254b) and the much more permanent situation of (256):  (256)  Aron  kann  engin erlend mál.  Aaron knows no  foreign languages  ‘Aaron does not know any foreign languages.’  Hann er ekki einu sinni a skilja He  is not  one  time to  ensku.  understand English  ‘He is not even understanding English.’  Only 24.5% of the subjects accepted this sentence (compared to 46.6% that accepted (254b)). This is not surprising for as with (255) we have no control, no dynamicity and no implied temporariness. There is, therefore, no reason to shift the stative to an event and put the sentence in the progressive. In general the results of this section support my proposal about coercion of stative verbs to events and the reasons for why the verbs are coerced.  3.6 Where English and Icelandic differ English and Icelandic are related languages with quite similar aspectual and progressive systems. However, they do not always correlate when it comes to which verbs can occur in the progressive and which verbs cannot. The most noticeable difference has to do with locative verbs and posture verbs, which occur freely in the progressive in English but not in Icelandic. Additionally, weather verbs occur in the progressive in English but not in Icelandic, and English has an active be construction that is lacking in Icelandic. I will now look more closely at these different situations between the two languages.  3.6.1  Locatives and posture verbs  Let us look back at the examples in (240) from section 3.4.2.3.  113  (240)  a. The socks are lying under the bed. b. Your glass is sitting near the edge of the table. c. The long box is standing on end. d. One corner of the piano is resting on the bottom step.  Interestingly enough, none of these posture verbs can occur in the progressive in Icelandic:  (257) a. *Sokkarnir eru a liggja undir Socks.the  are to  lie  rúminu.  under bed.the  ‘The socks are lying under the bed.’  b. *Glasi  itt  er a standa nálægt borbrúninni.  Glass.the your is to stand  close  table.edge.the  ‘Your glass is standing near the edge of the table.’  c. *Langi kassinn er a standa upp á Long  box.the is to stand  endann.  up on end.the  ‘The long box is standing on end.’  Instead, Icelandic would use the posture verb in the simple present:  (258) a. Sokkarnir liggja undir Socks.the lie  rúminu.  under bed.the  ‘The socks are lying under the bed.’  b. Glasi  itt  stendur nærri borbrúninni.77  Glass.the yours stands  close table-edge.the  ‘Your glass is sitting near the edge of the table.’  c. Langi kassinn stendur upp á Long  box.the stand  endann.  up on end.the  ‘The long box is standing on end.’  77  Icelandic uses standa ‘stand’ rather than sitja ‘sit’ in this context. Laurel Brinton (p.c.) has pointed out to me that the verb ‘stand’ in English would also be natural in that context. 114  Only one sentence with a locative and one with a posture verb were included in the ScanDiaSyn project (ráinsson 2010):  (259) a. Biggi fær ekki íbúina Biggi gets not  fyrr  eftir rjá  en  mánui.  apartment.the before than after three months  ‘Biggi will not get the apartment for at least three months.’  Hann er a búa  hjá mömmu sinni og pabba anga  til.  He  at  to  is to live  mom  self’s and dad  there  ‘He is living with his mom and dad until then.’  c. Gunna er komin. Gunna is come ‘Gunna has arrived  Hún er a standa  arna vi hliina  She  there at  is to stand  á Jóni.  side.the to Jón  ‘She is standing there next to Jón.’  80.9% of all the subjects rejected the locative sentence and 83.2% rejected the posture verb sentence. The acceptance rate was well under 10% for both sentences. So why is it that posture verbs and locatives can occur in the progressive in English but not in Icelandic? One option is that there is simply a fundamental difference between these languages, when it comes to these particular verbs: in English they are event verbs but in Icelandic they are stative verbs. In fact, the COCA results I discussed in section 3.2, indicated that posture verbs seemed to behave like event verbs, rather than stative verbs, as they occurred in the progressive similarly often as event verbs (and even slightly more often), and much more often than stative verbs. If posture verbs really are event verbs in English but stative verbs in Icelandic, we expect them to behave differently in other respects as well, not just in the progressive. Three stative tests from chapter 2 turned out to be reliable stative tests so let us now apply them to posture verbs. In English, posture verbs get a habitual reading in the simple present, just like event verbs. They require the progressive for an episodic reading.  (260)  John sits on a chair *now/in the mornings.  (261)  John is sitting on a chair.  115  This is not so in Icelandic where the posture verbs in the simple present are ambiguous between an episodic reading and a habitual one:  (262) Jón situr á stól Jón sits  núna/á  morgnana.  on chair now/on mornings.the  ‘Jón is sitting on a chair now/Jón sits on a chair in the mornings.’  When we embed a posture verb the simultaneous reading is possible in Icelandic but not in English:  (263)  John said that Mary sat on a chair.  (264) Jón sagi a  María sæti á stól.  Jón said that María sat on chair. ‘John said that María was sitting on a chair.’  When we have a posture verb in the present perfect in English the event time has to precede the speech time. (265) Mary has sat on a chair (*ever since yesterday).78  In Icelandic it can either precede or overlap the speech time:  (266) María hefur seti á stól (sían í María has  sat  on chair since in  gær). yesterday  ‘Mary has sat on a chair/Mary has been sitting on a chair since yesterday.’  When we draw the conclusions together, including the progressive we see that in all these tests the Icelandic posture verbs behave like states and the English posture verbs behave like events.  78  Of seven English speakers I asked to judge this sentence, four said it was impossible in the simple past and that it had to be in the progressive to work. One said she didn’t mind it in the simple past perfect but preferred the progressive, and two said it was perfectly fine as long it had a habitual reading: “Mary usually likes to sit on pillows or cross-legged on the floor or whatever, but, since she hurt her tailbone yesterday skiing, she can’t sit on the floor anymore.”  116  Table 3:3: Do posture verbs pattern with states or events? Icelandic Habitual meaning Sequence of tense The perfect  States     English Events * * *  States * * *  Events     These tests show that posture verbs are like events in English but states in Icelandic. This would explain why events occur in the progressive in English and not in Icelandic.  3.6.2  Weather verbs  Weather verbs are usually seen as dynamic. In English they occur in the progressive like other event verbs:  (267)  a. It is raining. b. It is snowing.  These verbs are non-agentive and take a dummy subject. However, ráinsson (1999:216) has pointed out that weather verbs do not generally occur in the progressive in Icelandic. Instead they are used in the simple present: 79  (268) a. *a er a rigna. It  is to rain  ‘It is raining.’  b. *a er a snjóa. It  is to snow  ‘It is snowing.’ (269) a. a rignir. It  rains  ‘It rains.’  79  I am ignoring weather verbs that mark a change of weather such as hvessa ‘get windy’ and hlna ‘get warmer’.  117  b. a snjóar. It  snows  ‘It snows.’  Just as stative verbs are being used in the progressive, we do find occasional examples of weather verbs in the progressive in Icelandic. Although I have found only one example on the Internet with rigna ‘rain’, there are quite a few examples with snjóa ‘snow’.80  (270) a. Nú  er snjórinn loksins farinn en  Now is snow.the finally gone  a er a rigna enná.  but it  is to rain  still  ‘The snow is finally gone but it is still raining.’ http://gudbrandur.blogcentral.is/eldra/2007/1/  b. a er a snjóa It  is to  hér  fyrir noran.  snow here for  north  ‘It’s snowing here up north.’ http://www.isalp.is/umraedur/9-skiei-og-bretti/6526-tae-er-ae-snjoa-her-fyrir-norean.html  Additionally, in the ScanDiaSyn project, the following sentences were tested (ráinsson 2010): (271) a. a er a rigna mest í It  is to  rain  október.  (Progressive habitual)  most in October  ‘It is raining most in October.’ b. a er enná a rigna. It  is still  (Episodic, regular progressive)  to rain  ‘It is still raining.’ c. a rignir enná. It  rains  (Episodic, simple present)  still  ‘It still rains.’  The most acceptable of these three sentences was the simple present in (271c), as 92.2% found it grammatical and only 2.9% found it unacceptable. The acceptability for the episodic progressive sentence 80  I found a few examples where rigna was used in the progressive as a metaphor where other things than rain are said to be raining over people: ...og a er a rigna yfir mann commentum. ‘...and comments are raining over me’. 118  in (271b) was about half that, 46.9%, with 29.3% deeming it unacceptable and 23.8% being unsure. That is a relatively high acceptability rate for a sentence which usually has been considered ungrammatical by linguists. The habitual sentence in (271a) had a fairly low acceptability rate, as 59.5% found it unacceptable, 22.7% were not sure, and 17.8% found it acceptable. As habitual sentences are commonly expressed by the progressive, as further discussed in the next chapter, we would not expect the sentence to get this low acceptance rate, but let us not forget, that even though habitual sentences often occur in the progressive they are much more commonly expressed with the simple present. When searching COCA, I found 123 examples of it snows and 26 examples of it is snowing. The progressive, therefore, occurs in 17.5% of the times the verb occurs, and the simple present 82.5% of the times, which is much more on par with event verbs than stative verbs. Of the 112 examples I found of it snows only one might possibly refer to an ongoing situation. However, this example appears to be the narrative present and so does not really count as a typical ongoing situation.  (272)  It snows. Halfbaby wakes up to snow. That can not be right. There are still flowers snaggling the gardens at the marsh farm, but there they are, heads poking up through the snow. So it snows. (COCA:FIC)  All the other examples are either habitual/generic or they refer to a future situation:  (273)  a. When it snows, it pours. (COCA:NEWS) b. Do they not know they live in Colorado where it snows like every other day? (COCA:SPOK) c. “...But when it snows, all of that has to be cleared,” said Snyder. (COCA:NEWS) d. I want to get us settled before it snows. (COCA:FIC)  Just like the progressive, this test categorizes weather verbs in Icelandic with states but with events in English. There may be a fundamental difference between English and Icelandic to the effect that weather verbs are event verbs in English but stative in Icelandic. However, just like other stative verbs in Icelandic, they can be coerced to being eventive and that is why we see cases of snjóa ‘snow’ and rigna ‘rain’ in the progressive. Let us now look at the three tests deemed reliable stative tests in chapter 2 and see how the weather verbs in English and Icelandic do when these tests are applied to them. In English, weather verbs get a habitual reading in the simple present, just like event verbs. They require the progressive for an episodic reading.  119  (274)  It snows *now/often.  (275)  It is snowing.  This is not so in Icelandic where weather verbs in the simple present are ambiguous between an episodic reading and a habitual one: (276) a. a snjóar núna. It  snows now  ‘It is snowing now.’ b. a snjóar á veturna. It  snows on winters  ‘It snows in the winters.’  When we embed a weather verb the simultaneous reading is possible in Icelandic but not in English:  (277) John said that it snowed.  (278) Jón sagi a  a snjóai.  Jón said that it  snowed  ‘John said that it was snowing.’  When we have a weather verb in the present perfect the event time has to overlap the speech time in English but in Icelandic it can either precede or overlap the speech time:  (279) It has snowed *yesterday/since yesterday. (280) a hefur snjóa It  has  í  gær/sían  í  gær.  snowed in yesterday/since in yesterday  ‘It has snowed yesterday/it has been snowing since yesterday.’  When drawing together the conclusions of the three tests, Icelandic weather verbs clearly pattern with stative verbs whereas English weather verbs rather pattern with event verbs.  120  Table 3:4: Do weather verbs pattern with states or events? Icelandic States     Habitual meaning Sequence of tense The perfect  English Events * * *  States * * *  Events     We must, therefore, conclude that weather verbs are eventive in English but stative in Icelandic. If English weather verbs are eventive it is not surprising that they occur in the progressive like other events. And if they are stative in Icelandic then it is also not surprising that they do not generally occur in the progressive. But it is then equally unsurprising that they can be coerced to events just like other stative verbs. My analysis does, in fact, predict that when they can be seen as dynamic or temporary, they are more likely to get coerced. It is not a problem for my analysis that English and Icelandic categorize these verbs in a different way as different languages make different choices. This is true particularly when it comes to verbs that could theoretically be categorized in more than one class, such as the posture verbs, locatives and weather verbs. What class a certain verb belongs to is to some extent arbitrary information and therefore must be marked in the lexicon and it is not a choice that the speaker makes on the fly. If it were we would expect much more variation between speakers within the same language.  3.6.3  Be in the progressive  Another difference between Icelandic and English has to do with sentences like (281) where the verb be in English can occur in the progressive but the Icelandic vera ‘be’ cannot:81  (281) John is being a fool.  (282) *Jón  er a vera fífl.  John is to be  fool  ‘John is being a fool.’  In the ScanDiaSyn project (ráinsson 2010) there was only one sentence with be:  (283) a. Hlustau Listen  ekki á not  Halla.  to Halli  ‘Do not listen to Halli.  81  For Icelanders to indicate that John is acting like a fool they would need a verb like hega sér ‘behave’. 121  b. Hann er bara a vera kjánalegur. He  is just  to be  silly  ‘He is just being silly.’  Although the acceptability rate for this verb was not as low as that of the locatives and the posture verbs, only 24.1% of the subjects accepted it, whereas 52.5% rejected it. Compare the sentences in (281) and (282) to the simple present:  (284) John is a fool.  (285) Jón er fífl. Jón is fool ‘Jón is a fool.’  There is an obvious difference between the English progressive sentence in (281) and the simple present sentence in (284): the progressive sentence describes an event of John acting like a fool, whereas the simple present sentence describes a state of John being a fool. As the VP be a fool is usually seen as stative, this would be another case of stative verbs being coerced to event verbs in order to convey some eventive property. So why is it that this coercion is perfectly grammatical in English but not in Icelandic? Partee (1977) has proposed that there exists an eventive verb “active be” which is separate from but homophonous with the more common stative verb be. Active be occurs with adjectives and can be used in the progressive, unlike stative be.82 She showed that with active be animate subjects differ from inanimate subjects, and similarly there is an element of agentivity involved in the interpretations of subjects that occur with active be. Let us start by looking at the following example from Partee, where active be is coupled with the adjective noisy. In the simple present the type of subject does not make any difference, but in the progressive the inanimate NP the river is bad.  (286) a. John is noisy. b. John is being noisy.  (287) a. The river is noisy. b. #The river is being noisy.  82  Partee talks about two stative be-verbs, be1 and be2. Her be1 is Montague’s transitive be which occurs in sentences like Bill is a man and Bill is a tall entity. Be2 on the other hand is inserted by the rule which combines predicate adjectives with their subjects, such as Bill is awake. 122  This cannot be taken as a general prohibition on verbs in the progressive taking inanimate subjects as seen in the following example from Partee, where the inanimate subject of the verb phrase making a lot of noise is fine:  (288) The river is making a lot of noise.  Partee (1977:306-307) claims that the property of volition is quite important with the combination of active be in the progressive and an adjective. This can clearly be shown in (289) where are being quiet can only occur with a volitional subject:  (289) a. ?The children are being quiet right now because they’re asleep. b. The children are being quiet right now because they want a story.  Partee claims that active be requires volition but such a constraint does not apply to event verbs in general, although, as we have seen, control (a close analogue of volition) is in fact one of the prototypical properties associated with event verbs:  (290) a. The children are making so little noise right now because they’re asleep. b. The children are making so little noise right now because they want a story.  Basically, an eventive verb phrase like making noise can occur in the progressive regardless of the animacy or volition of the subject but the seemingly stative being noisy, where we have active be and an adjective, can only occur with animate and volitional subjects. Also, active be can only occur with certain adjectives. Partee mentions noisy and foolish as possible adjectives with active be but impossible are adjectives like awake and healthy. This is not surprising as being awake and being healthy are usually not things one does volitionally.  (291) a. The teenagers are being noisy on purpose. b. John is being foolish. c. *Catherine is being awake. d. *The children are being healthy.83  83  Interestingly enough, we can use be healthy in the progressive if it indicates dynamicity and control, such as in: Wow, you are sure being healthy these days, running every day, eating vegetarian food, going to bed early every night... That fits perfectly with my analysis.  123  Lakoff (1970) had proposed to treat adjectives as verbs and classify them as active or stative. An adjective like noisy, for instance, would be active whereas healthy would be stative. This would then explain why (292a) is ungrammatical but (292b) is not. Just like active verbs, the active noisy occurs in the progressive and just like stative verbs, the stative adjective healthy usually does not:  (292) a. *John is being healthy. b. John is being noisy.  However, as Partee points out, this proposal does not distinguish between (292b) and (292c): (292) c. *The river is being noisy.84 If (292b) were good because noisy is an active adjective, one would expect (292c) to be good as well. As we saw in (287) that noisy does allow inanimate subjects in general, we cannot easily explain the difference between (292b) and (292c) in such a way that (292c) is ruled out because of the inanimacy of the subject. Partee gives the following properties for active be (which she starts out by calling be3): (293) Partee’s properties for the active be (Partee 1977:308-309)85 (i) Be3 is marked +Active with respect to the active/stative feature (which I claim should be a feature only on verbs). (ii) Be3 combines with those adjectives that allow animate subjects. ... The adjective need not be one which requires an animate subject; noisy, for instance. (iii) The resulting verb phrase, consisting of Be3 plus an adjective phrase, requires an animate subject. (iv) Even more informally, I will note that its meaning is a little like the meaning of act in John was acting foolish and which also is restricted to animate subjects. (The river is acting turbulent is odd). The same sort of notion of agentivity which I mentioned in connection with examples [(289a)] and [(289b)] ought to be able to be used to account for the anomaly of be3 with some kinds of adjectives like unhealthy. John was being unhealthy and John was being right are both a little odd because there is not much one can do to be unhealthy or to be right. This is of course quite vague and informal. Active be, Partee claims, can take the progressive, takes manner adverbs like intentionally, can occur in imperatives and in complements of try.86  84  Laurel Brinton (p.c.) has pointed out that the sentence might be possible if one is talking about a temporary state of the river, e.g. when the ice is breaking up. That fits with my discussion of temporariness from section 3.4.2.3.  85  Italics and comments within parenthesis are all those of Partee.  86  Interestingly, the Icelandic verb vera ‘be’ can occur in the imperative (Vertu ekki leiinlegur ‘Don’t be annoying’) and as a complement of reyna ‘try’ (Hann reyndi a vera ægur/fyndinn ‘He tried to be good/funny’), so it is not the case that it cannot be used with an agentive subject. And yet it cannot occur in the progressive. 124  In light of the theory developed in this chapter, I propose that in sentences with active be there is an implicature that the subject has control and/or the event is presented as dynamic and/or temporary. This is obvious in the following pair:  (294) a. John is a fool. b. John is being a fool.  In (294a) we have a stative sentence. The subject is a Theme and there is no dynamicity in the sentence. The state of John being a fool does not require him to necessarily act in a certain way, and in fact, Laurel Brinton (p.c.) has pointed out to me that we could add to the sentence ‘but he behaved very wisely in that situation’, where the speaker indicates that John is a fool in general but that at a certain time he acted quite differently. The sentence in (294b), however, has both control and dynamicity. John is behaving in a certain foolish way. Notice that by saying that John is being a fool, the speaker is not making a commitment as to whether John is a fool in general. The speaker’s opinion might be that John is in fact quite an intelligent person with good common sense, who for the time being is acting like a fool. As I have already argued that control and dynamicity are two of the three prototypical eventive properties that tend to show up with states in the progressive, the parallel is obvious between states and active be in the progressive. Additionally, the sentence in (294b) has the implicature of being temporary – John is not a fool in general, he is just temporarily behaving like a fool – another implicature common with states in the progressive. Recall, however, that we do not have an Icelandic equivalent of active be. Here are some further data showing that:  (295) a. *Jón  er a vera hávær.  John is to  be  noisy  ‘John is being noisy.’  b. *Börnin Children  eru a vera hljólát essa stundina af ví a au are to be  quiet  this  vilja heyra sögu.  hour.the because they want hear story  ‘The children are being quiet right now because they want to hear a story.’  No active be exists with adjectives in Icelandic. In fact, I have not found any examples in Icelandic where the verb vera ‘be’ is used in a similar way as the English active be.87 At this point I am not sure why it is 87  Actually, I did find one case on the Internet – a caption to a picture – that read: (i) Pabbi a vera stoltur maur. Dad to be proud man 125  that in English we can use the verb be in the progressive but in Icelandic we cannot. There is one possible explanation, though, that comes to mind. Warner (1995, 1997) suggests that the reason why the passive progressive (e.g. is being done) became available in English was because of the lexicalization of being as a separate entry in the lexicon of English speakers. Learners classified finite forms of be (and its forms, am, are, is, etc) as an auxiliary but they lexically stored the participle being as a separate verb. He points out that there is some evidence that being should be analysed as a non-auxiliary in contemporary English (Warner 1997:166-167). He argues that this reanalysis made passivization of the progressive available as the VP is being became possible in English. Additionally, Warner argues that this reanalysis gave rise to forms like is being NP. Icelandic is an inflectional language that has not gone through the same changes in inflection as English has. So there is no reason to believe that the infinitive of the verb vera ‘be’ got reanalysed, like being did in English, and got lexicalized separately. And if it was the lexicalization of being that made the passive progressive and the active be available in English, and if such lexicalization did not take place in Icelandic, then the lack of those constructions in Icelandic is exactly what we expect. However, whether this is the explanation or not would require much more research.  3.7 Conclusions I have argued that true stative verbs do not occur in the progressive as they lack stages, which are necessary for the progressive construction. This is because stative verbs have the state argument but not the event argument that events do, and having stages is a property of events and not of states. However, stative verbs can be coerced to events and then tend to have a slightly different meaning. The reason for this is that all event verbs have certain prototypical eventive properties to some extent. They do not have to have all of these properties but the more of them they have the more prototypical events they are. If a speaker wants to convey any of these prototypical eventive properties, such as control, dynamicity or temporariness, with a state, she can shift the state to an event and by doing so implicate one or more of these eventive properties. As eventive verbs need to be in the progressive in order to convey the meaning that the event is in progress, the former-stative verb now appears in the progressive. Control, dynamicity or temporariness are mere implicatures that are not a part of the lexical entry of the verb or of the truth conditions of the progressive itself. Notice that the speaker might also shift the state to an event simply to give the in-progress-reading of the state. There seems to be a difference in the progressive use between Icelandic and English as posture verbs, weather verbs and the verb be can occur in the progressive in English but not in Icelandic. ‘Dad, being a proud man.’ The fact that this is in a photo caption is important because such texts, as well as texts on social networking websites like Facebook and Twitter, appear to follow their own grammatical rules, just like newspaper headlines (see e.g. Jóhannsdóttir 2006). This is an interesting topic for future research.  126  However, I have shown that this is not a difference between the use of the progressives in the two languages but the result of these verbs being eventive in English but stative in Icelandic.  127  4 The progressive habitual 4.1 Introduction As discussed in chapter 1, the progressive construction is usually used to show that an event is ongoing at a particular time:  (296) John was playing guitar when Paul entered the room.  Here we see that John is playing guitar at the point in time when Paul entered the room, and it is quite possible that his guitar-playing continued for some time after that. So we are referring to one playing event, which may or may not have been interrupted by Paul’s entrance or by something else, but is nevertheless seen as one event. In both English and Icelandic, however, the progressive construction can also be used to refer to multiple events of the same kind that happen one after another with possible interruptions.  (297) a. “What people are seeing on TV, that to me, that’s really Taylor”, she said. (Pat Hummel, ctv.ca) b. Kesler’s desire to show his offensive side has been well documented, but what he hasn’t talked about much, up to now, is how well he’s seeing the ice. (Botchford, The Province, Nov.7, 2007, p A51) c. It is great news that when we test our water we are not finding any bacteria. (CTV news)  The sentence in (297a) is ambiguous without the context. It could mean that during one particular TV performance of Taylor (Hicks), he is really showing who he is, or it could refer to multiple performances on multiple occasions. In this case the latter is true as the sentence did indeed refer to several episodes of the show American Idol, where the speaker believed Taylor was showing his true self. Example (297b) is similar to (297a) as we are not talking about one event where Kesler sees the ice well, but repeated events of Kesler seeing the ice – many different games.88 Sentence (297c) is a bit different  88  One might be tempted to categorize seeing here as a stative verb and therefore refer back to chapter 3 where I discuss statives in the progressive. However, as argued there, stative verbs used in the progressive are actually eventive so we should safely be able to refer to this as an event. Also, the meaning of seeing the ice is more similar to reading the ice/situation than actually being the state of seeing something. It is a question whether it is important here that seeing the ice has become a fairly fixed idiom.  128  from (297a) and (297b) as it refers to events that do NOT take place. However, instead of looking at it as the lack of events, it can be seen as a series of events of not finding any bacteria. The scientists test the water; they do not find any bacteria. The scientists test the water for the second time; they do not find any bacteria, etc. We see similar examples in Icelandic.  (298) a. United virist vera a komast í United seems be næla í  to get  stigin  sinn venjulega gír - spila kannski ekki vel en  to self usual  egar á arf  gear play maybe  a halda en  not well but  eir eru ekki a spila  catch in points.the when on needs to hold but they are not to play skemmtilegan bolta fyrir okkur sem fylgjumst me sem hlutlausir ailar. entertaining  ball  for  us  that follow  with that neutral  parties  ‘United seems to be getting into their usual gear – they may not play well but get the points when needed, but they are not playing entertaining soccer for those of us who are watching as neutral parties.’ http://fosterinn.blog.is/blog/fosterinn/entry/312701/  b. eir eru a spila tónlist sem var uppi á They are to play music that was up  9 áratuginum.  on 9th decade.  ‘They are playing music that was popular in the eighties.’ http://www.hugi.is/tonlist/articles.php?page=view&contentId=1049279  In (298a) we have a case where Manchester United plays many games and according to the speaker the soccer they are playing in those games is not entertaining. Therefore, multiple events of Manchester playing non-entertaining soccer. In sentence (298b) each concert is an event of playing music from the eighties.89 When we look at the sentences in (297) and (298) we notice that they all have approximately the following meaning:  (299) (During time period t) when events of kind e take place, they are x.  So if we have a sentence like (300a) we get the translation in (300b):  89  Without context the sentence could be seen as a single occurrence of playing music from the ninth decade. However, the sentence was in fact taken from a context where the speaker was referring to the music the band played in general.  129  (300) a. The Canucks are playing well. b. (During time period t) when the Canucks play [hockey], they play well.  The reference time is not overtly given in the sentence. This description is very reminiscent of habitual sentences as described by, for instance, Krifka et al. (1995):  (301)  a. Mary smokes when she comes home. b. GEN[s,x;](x = Mary  x comes home in s; x smokes in s)  This can be read as: In general, if there is a situation of Mary coming home, she will smoke in that situation. The main difference between the habitual sentence in (301) and the progressive sentence in (300) is that unlike the progressive sentence, the habitual sentence represents a general habit without any implicit event time implied. Just like the habitual sentence, the progressive sentence in (300) indicates a habit. In (301) the when-clause is the restrictor. However, not all habitual sentences have an overt restrictor, and in such cases it is not necessarily obvious what should be considered the restrictor. Look at the following sentence: 90  (302)  Yrsa is eating broccoli (these days).  A possible reading of the sentence would be that ‘whenever Yrsa eats, she eats broccoli’. However, I propose a more common understanding of the sentence is that ‘for natural situations where she is offered broccoli, Yrsa eats broccoli’. This is a similar reading as we get with the simple present, such as in (303):  (303)  Yrsa eats broccoli.  The use of the progressive, however, appears to contrast Yrsa’s habit of eating broccoli against a time or situation where she did not eat broccoli, making the habit either temporary or newly acquired. The similarity between habituality and these progressive sentences is striking. I will argue in this chapter that this similarity is due to the fact that these progressive sentences are in fact habitual and what we have here is the progressive of the habitual. That is, the sentences indicate that a habit is in progress. Contrast that to the opposite where we have the habitual of a progressive:  (304)  90  Every morning when I come to work, John is reading the paper.  Thanks to Gunnar Ól. Hansson for this example and to Laurel Brinton for clarifying it. 130  Here we have the progressive sentence, ‘John is reading the paper’, and a reference time with a universal quantifier meaning that John’s reading takes place every morning (around the time the speaker comes to work). So we can say that the habitual operator quantifies over the progressive, whereas in the sentences I will be discussing in this chapter, the progressive operator quantifies over the habitual sentence. I will call the latter kind of sentences progressive habitual. Table 4:1: The co-occurrence of the progressive and the habitual Construction Progressive habitual Habitual of the progressive  Description The progressive operator applies to a habitual sentence. The habitual operator applies to a progressive sentence.  Example The Canucks are playing well Whenever I come home, John is eating.  The breakdown of this chapter is as follows. In section 4.2 I will discuss habitual aspect in general. I will first explain the difference between habitual aspect and generics, as well as between habitual aspect and iteratives. Then I will discuss habitual constructions in both English and Icelandic and finally the semantics of the habitual, particularly as formulated by Krifka et al. (1995). The main focus of section 4.3 will be on the progressive habitual, including previous treatments of the construction, as well as the characteristics of the progressive habitual. I will show that it is in fact habitual and that what we have is really the progressive of habitual sentences. Furthermore I will discuss cases where we have a habitual sentence co-occurring with a progressive habitual, such as in ‘The Canucks usually play well, but these days they are playing badly.’ I will also discuss the event time of habitual sentences, which appears to differ between regular habitual sentences and the progressive habitual. In section 4.4 I will discuss the semantics of the progressive habitual, including stages, and the shift of habitual sentences to events. Section 4.6 summarizes the conclusions of the chapter.  4.2 Habitual aspect In this section I am going to look at habitual constructions in both English and Icelandic, and explore some characteristics of habituals that have been suggested by linguists, in order to identify the properties of habituals. I will then look at the semantics of habituals, mostly as discussed by Krifka et al. (1995). But before we go further, let us discuss the difference between habituals and other generics, as well as iteratives.  4.2.1  Habitual aspect vs. generics and iteratives  It is not always easy to distinguish between habituals and generics. Sentences such as the one in (305) have been called both (Carlson 2005:2): 131  (305)  Mary eats oatmeal for breakfast.  According to Carlson the term generic is more commonly used in the formal semantics literature whereas habitual is more common in the descriptive literature. He points out that some linguists use the term generic for habitual sentences that have a generic subject noun phrase, rather than a specific reference. So ‘birds fly’ would be generic, whereas ‘John drives to school’ is habitual. Krifka et al. (1995) have a different categorization than Carlson as they use the cover term generic sentences and contrast it with episodic sentences. So episodic sentences usually refer to a single event of some kind, as in (306), whereas generic sentences can refer to habits or generalizations, as in (307).  (306)  John is smoking a cigar.  (episodic)  (307)  John smokes a cigar after dinner.  (generic)  Generic sentences are further divided into characterizing sentences and reference to kinds. Krifka et al.’s characterizing sentences report a habit, or generalizations over events, as shown in (307), whereas in sentences that involve reference to kinds, the NP does not refer to an individual or object but to a kind:  (308)  a. The potato was first cultivated in South America. b. Potatoes were introduced into Ireland by the end of the 17th century.  Krifka et al. point out that characterizing sentences and reference to kinds have something in common: “with kinds we abstract away from particular objects whereas with characterizing sentences we abstract away from particular events and facts” (Krifka et al 1995:4). However, it is quite important to keep the two separated. Krifka et al. further divide characterizing sentences into the categories of habitual sentences and lexical characterizing sentences. According to them, the “verbal predicate of a habitual sentence is morphologically related to an episodic predicate that is commonly used to form episodic sentences, whereas lexical characterizing sentences lack such an episodic counterpart” (Krifka et al. 1995:17). So the sentences in (309) are habitual as they contain the episodic verb smoke:  (309)  a. John smokes. b. Italians smoke.  132  Habitual sentences generalize over patterns of events as a component of their meaning so here the generalization is over instances of John or Italians engaging in smoking. Lexical characterizing verbs, however, “do not have such morphologically related episodic predicates, and consequently there is no semantic generalization over events; rather the generalization would appear to be over characterizing properties of individuals” (Krifka et al. 1995:17).  (310)  a. Italians know French. b. John is intelligent.  Basically, these sentences lack an episodic counterpart. Notice that they are individual-level statives and cannot be seen as being habitual. Krifka et al. also stress the importance of keeping habitual sentences separate from universally quantified sentences, as the former allow exceptions that are not allowed with the latter. They give the following example: if from time to time John does not smoke after dinner, the characterizing sentence in (307) can still be true, whereas that is not the case with the universally quantified sentence in (311) (Krifka et al. 1995:4):  (311)  Always, after dinner/After each dinner, John smokes a cigar.  As habituals tend to denote multiple events, they can easily be confused with iteratives, which Carlson (2005:2) describes as “a subclass of aspectual operators, which do not produce generic or habitual sentences but rather are episodic in nature”. Comrie (1976:27) gives the following example of an iterative event that is not habitual:  (312)  The lecturer stood up, coughed five times, and said...  What we have here in italics is a repeated action that can be seen as one coughing event with iterative components. Common iteratives are verb phrases such as knocking and flapping its wings. Let us now look at how habitual aspect is represented in English and Icelandic.  4.2.2  Habitual aspect in English and Icelandic  The most common way to express habituality in English is by using the simple present, as seen in the following examples taken from Kearns (2000:151):  133  (313)  a. Heath bikes to work. b. Barry feeds the dogs. c. She writes with a fountain pen.  The meaning of the sentences can be made slightly more explicit in the following way:  (314)  a. Whenever Heath goes to work, he bikes. b. Whenever the dogs are fed, it is Barry who feeds them. c. Whenever she writes, she uses a fountain pen.  However, English has another way to express habituality, although only for past tense sentences, involving the construction used to (Comrie 1976:25):  (315)  John used to work here.  The simple form can also be used in the past tense but then the habitual aspect is optional:  (316)  John worked here. =John worked here that one time. =John used to work here.  The habitual aspect in Icelandic is in many ways similar to that of English. Just as in English, the simple present is most commonly used to present habitual aspect. The Icelandic equivalents of the examples in (313) are the following:  (317) a. Hinrik hjólar í Hinrik bikes  vinnuna.  into work.the  ‘Hinrik bikes to work.’  b. Björn gefur hundunum. Björn feeds dogs.the ‘Björn feeds the dogs.’  c. Hún skrifar me blekpenna She  writes with fountain.pen  ‘She writes with a fountain pen.’ 134  And just like English has the used to construction in addition to the simple past, Icelandic has the vera vanur a construction, which means something like ‘be in the habit of’.  (318) Jón er vanur a skrifa me blekpenna. Jón is used  to write with fountain.pen  ‘Jón writes with a fountain pen.’  However, it should be made clear that the Icelandic vera vanur a construction is not a direct equivalent of the English used to construction. However, as it is not essential for my analysis of the progressive habitual I will not discuss it further in this dissertation. In the next section we will look at the characteristics of the habitual aspect.  4.2.3  Characteristics of the habitual aspect  4.2.3.1 Characteristic of an extended period of time Comrie (1976:27-28) defines habitual readings in such a way that “they describe a situation which is characteristic of an extended period of time, so extended in fact that the situation referred to is viewed not as an incidental property of the moment but precisely, as a characteristic feature of a whole period”. His definition is differently worded from that of Krifka et al. (1995) but nevertheless works well with it.  (319) a. Henry bikes to work. b. Henry used to bike to work.  And the Icelandic equivalents:  (320) a. Hinrik hjólar í Hinrik bikes  vinnuna.  to work.the  ‘Hinrik bikes to work.’  b. Hinrik var/er vanur a hjóla í Hinrik was/is used  vinnuna.  to bike to work.the  ‘Hinrik used to bike/usually bikes to work.’  The only way to understand the sentences in (319) and (320) is that it is or was a general habit of Hinrik/Henry to bike to work over an extended period of time. In general, for these sentences to be  135  habitual, we assume there is a certain length of applicable time involved, but how long that time has to be depends on the nature of the event and context. If Hinrik bikes to work three days in a row, but then starts taking the car again, it is doubtful whether anyone would use the habitual to describe the act. However, notice that Hinrik does not always have to bike to work for the sentences to be true, which leads us to the next characteristic.  4.2.3.2 General habit but not universal quantification One can have a habit of doing something without it being necessary that one does it all the time (see e.g. Carlson 2005). So if Henry generally bikes to work, the sentence in (319) is true even though he takes his car every now and then, for instance if he is in a hurry.  (321) a. Henry bikes to work, except when he is in a hurry; then he takes the car. b. Henry used to bike to work, except when he was in a hurry; then he took the car.  And the Icelandic examples:  (322) a. Hinrik hjólar/er vanur a hjóla í Hinrik bikes/is used Fer  vinnuna, nema  to bike to work.the except when he  hann/er hann vanur a fara á  goes he/is  egar hann er a flta sér, á  he  is to hurry him then  bílnum.  used to go on car.the  ‘Hinrik bikes to work, except when he is in a hurry; then he takes the car.’  b. Hinrik hjólai/var vanur a hjóla í Hinrik biked/was á  fer  vinnuna, nema  egar hann var a flta sér,  used to bike to work.the except when he  was to hurry him  hann/er hann vanur a fara á bílnum.  then went he/was  he  used to go on car.the  ‘Hinrik used to bike to work, except when he was in a hurry; then he took the car.’  Notice that biking to work has to be the general method of going to work for the sentence to be true. It cannot be the case that he sometimes bikes but just as often walks or drives. It is, however, not clear how often he has to bike for it to be seen as a habit. I will discuss this issue in more detail in section 4.2.5. It is also worth pointing out that in these examples, most Icelandic speakers would prefer the vera vanur a construction over the simple past/present.  136  4.2.3.3 Event does not have to be ongoing at the reference time Kearns (2000) points out that the present habitual does not necessarily describe an event occurring at the time of speaking. Even though Henry usually bikes to work, (319) does not mean that he has to be biking to work right now.  (323) Henry bikes to work. That is why he is in the shower now. The biking makes him sweat.  The same can be seen in Icelandic:  (324) Hinrik hjólar/er vanur a hjóla í Hinrik bikes/is used  vinnuna. ess vegna er hann í sturtu núna.  to bike to work.the that  why  is he  in shower now  Hann svitnar vi a a hjóla. He  sweats at  that to bike  ‘Hinrik bikes to work. That is why he is in the shower now. The biking makes him sweat.’  4.2.3.4 Conclusion When we draw these results together we can describe habitual sentences in the following way:  (325)  4.2.4  Habituals •  Are characteristic for an extended period of time.  •  Can have exceptions.  •  The eventuality does not have to be taking place at the speech time.  Habituals as statives  In this thesis I will follow linguists such as Leech (1971), Newton (1979), Mufwene (1984), Partee (1984), Chung and Timberlake (1985), Krifka et al. (1995) and Carlson (2005) in claiming that habitual sentences are aspectually stative or at least share major properties with statives. Carlson has mentioned examples where in narrative discourse, a habitual sentence does not move the time forward as events generally do, but instead appears to provide background/setting information, like statives. He also pointed out that habituals observe the subinterval property just like statives do. So, if a habitual sentence is true for a certain period of time then it is also true for any subinterval within that period. That cannot, however, be completely true as there has to be a certain length for a habitual to hold, so the habitual sentence can hardly be true for the smallest stretches of each interval. Krifka et al. (1995) claim that there are two cases of stativity:  137  […] either the verbal predicate is inherently independent of situations, as is the case with lexical statives such as be infertile and know French, or else the situation variable is bound by some operator other than existential closure, such as usually. In either case, the statements do not report on specific situations; and it seems natural to see this independence from specific situations as the core of stativity (though the existence of episodic statives such as progressives and such predicates as be available may demand a reformulation of the roots of stativity). It then follows that habitual sentences, which generalize over situations, are stative. (Krifka et al. 1995:36) It is important to keep in mind here that when we talk about habituals being aspectually stative we do not mean being stative in exactly the same way as stative verbs are stative – that is, as stative verbs are defined in chapter 2. For instance, one of the criteria used in categorizing stative verbs from eventive verbs was that events and not statives got habitual reading in the simple present. That, on its own, shows a difference, otherwise we were saying that habituals did not get habitual meaning in the simple present. So when talking about aspectually stative (as opposed to lexically stative) the term refers to the verb’s relation to time. If we look back at the definition of stative verbs, lexically stative predicates, I argued that they cannot have control subjects, that they cannot be seen as something that is done, and that they apply at the speech time if the sentence is in the present tense. Obviously habitual sentences can have control subjects (walking to school requires such a subject) and they refer to things that can be done, so neither of these properties of lexical stative predicates applies to habituals. However, just like lexical statives, habitual sentences do have to apply at the speech time if the sentence is in the present tense – that is, the habit has to apply even though the individual event does not have to. Furthermore, if we apply stative tests to habituals we get similar results with habituals as we got with statives. In chapter 2, section 2.2, we found that out that the three proper tests for stativity were the habitual reading of events in the present tense, sequence of tense and the perfect. We can not use the first test on habitual sentences, as previously mentioned, but the other two can give us a good support for the stativity of habitual sentences, so let us now apply them. States and events differ when we have embedded sentences, where the simultaneous reading of the embedded past is possible with states but not with events. Here habits behave like stative predicates as Mary’s playing the piano can either precede or overlap the time of John saying that she played piano and similarly Anne walking to school could either precede or overlap the time John said she walked to school.  (326) The sequence of tense phenomenon a. John said that Mary played piano well. b. John said that Anne walked to school.  As discussed in chapter two, events in the present perfect have to precede the speech time, whereas states can allow a continuous reading where the state overlaps the reference time. Here habituals behave like states as they can get the continuative reading in the perfect:  138  (327)  The perfect a. Mary has played the piano for days. b. Anne has walked to school all this week.  From this it is clear that habitual sentences behave like stative verbs. Both refer to an ongoing situation during a certain time, they do not move the reference time forward and the reference time can be seen to overlap with the speech time. We can, therefore, say that habitual sentences are aspectually stative but not lexically stative. The stativity of habitual sentences becomes quite important in the habitual progressive as statives cannot occur in the progressive unless they have first been shifted to events, as discussed in chapter 3. Since habitual sentences are aspectually stative they too need to be shifted to events before they can occur in the progressive. This will be discussed further in sections 4.3.3 and particularly 4.4.1.  4.2.5  The semantics of habitual sentences  Following Carlson (2005) and Krifka et al. (1995) I will assume that habitual sentences have the same tripartite structure as sentences with conditionals and when-clauses. Like such sentences, they have a dyadic operator, such as an adverb of quantification, which relates one set of conditions to another set. Krifka et al. introduce the operator GEN, a generic quantifier underlying all characterizing sentences that lack an overt quantificational adverb.91 GEN can be interpreted in such a way that it only takes into account the situations that are relevant for the generalization at hand, but Krifka et al. do not attempt to explain exactly how that works. Krifka et al. give the following definition for habitual sentences:  (328)  A sentence is habitual if and only if its semantic representation is of the form GEN[...s...;...] (Restrictor [...s...]; Matrix [...s...]) Where s is a situation variable.  The problem with adapting this analysis for my work in this dissertation is that I have been following Davidson (1967) in using event semantics but Krifka et al. use situation semantics. Using situations with habituals is simpler than using events as we can talk about the situation of Mary coming home and Mary smoking as one situation, but these are without a doubt two events. We would have to 91  Krifka et al. (1995) were not the first to propose such an operator for characterizing sentences, as they point out themselves, as Lewis (1975), Kamp (1981), Kratzer (1981), Heim (1982), Farkas and Sugioka (1983), Rooth (1985), ter Meulen (1986), Schubert and Pelletier (1987, 1989), and Declerck (1988), had all proposed such an operator for characterizing sentences which contain a conditional or a when-clause.  139  explain (329) in such a way that if e is an event of Mary coming home then e is usually accompanied by e’ where e’ is an event of Mary smoking.  (329) Mary smokes when she comes home.  What ‘accompanied’ means exactly is then left to context. This is in fact the approach taken by Rothstein (1995). Rothstein (1995) follows a Neo-Davidsonian theory of events and shows that a simple compositional analysis of adverbs is available within that framework. She suggests a matching effect that follows from a functional relation between the adverb and the event argument of the main verb. She argues that the event argument of the main verb is the argument of the adverbial and that the adverbs themselves must quantify over the domain of events.  (330)  a. Mary opens the door every time the bell rings. b. e[[RING(e)  Th(e)=THE BELL]  e’ [OPEN(e’)  Ag(e’)=Mary  Th(e’)=THE DOOR]]  However, the problem with this analysis is that it does not match the events of ringing to the events of opening the door. It simply says that for each event in the set of bell-ringing events there was an event in the set of door-opening events. It does not, however, specify that it was a different event or that the dooropening events were at least as many as the bell-ringing events. Also, the universal quantifier does not bind a variable in the nuclear scope of the sentence, which then violates the prohibition against vacuous quantification. What Rothstein wants is to represent two sets of events that are related in such a way that each event in the first set matches up with an event in the second set. So she suggests the function M, which is a matching function, matching one event with another. M maps events of door-openings onto events of bell-ringings. It takes the e’ argument of ‘open the door’ and maps it onto an event of bell-ringing, and therefore each bell-ringing event is the value of M applied to an opening event. The universal quantifier then defines the range of M as including the entire set of events of bell-ringings. So the sentence from (330a) can now be represented as (422) (Rothstein 1995:16):  (331)  e[[RING(e)  Th(e)=THE BELL]  e’ [OPEN(e’)  Ag(e’)=Mary  Th(e’)=THE DOOR  M(e’)=e]]  We can combine Rothstein’s analysis with Krifka et al.’s analysis of generics writing (329a) as (332):  140  (329a) Mary smokes when she comes home.  GEN[e] ([Come home(e)  Agent(e, Mary)]; e’[Smoke(e’)  Agent(e’, Mary)  M(e’)=e])  (332)  (332) maps events of Mary smoking to events of Mary coming home. It can be read in the following way: Generally, for each event of Mary coming home, it is matched with an event of Mary smoking. Now that we have discussed the semantics of habituals, let us see what the progressive of such habituals looks like.  4.3 The progressive habitual 4.3.1  Previous discussions of the progressive habitual  Hirtle (1967:49ff) is one of the first to have discussed the progressive habitual in detail (although not calling it that). He gives sentences like the one in (333b) where he contrasts the progressive habitual with the habitual simple present:  (333) a. I get up at six in the morning. b. I am getting up at six in the morning.  He claims that the progressive gives a more temporary feeling because it involves the possibility of change:  With the progressive the feeling of something temporary arises from the fact that the represented portion of the event — the accomplished part — is declared not to be typical because the present participle expresses, not a homogenous event, but one involving a possibility of change. Since it is unsafe to extrapolate beyond the accomplished portion of the event, the progressive suggests that the further accomplishment of the event is merely a possibility: I may go on getting up at six, but then again, I may not. An event whose continued existence in time is felt to be doubtful does not give an impression of permanence. (Hirtle 1967:49) Leech (1971:33) said of such sentences that the “progressive concept of ‘temporariness’ applies not to the individual events that make up the series, but to the series as a whole.” He calls it ‘habit in existence over a limited period’. He also discusses a second progressive habitual, which he calls ‘repetition of events of limited duration’. For this he gives the examples in (334), which are instances of what I call the habitual progressive.  141  (334) a. Whenever I pass that house the dog’s barking. b. Do not call on them at 7.30 – they’re normally having dinner. c. Usually the cramp starts just as I’m going to sleep. d. You only seem to come alive when you’re discussing your work.  He claims that the difference between the two kinds of habitual is that in this latter one “the notion of limited duration applies not to the habit as a whole but to the individual events of which the habit is composed” (Leech 1971:33). The reason why the progressive is used rather than the simple present is to “stretch the time-span of the event so that it forms a frame around the recurrent event or time-point”. These sentences are in fact quite different from what I call progressive habitual. The progressive habitual discussed in this chapter is really a habitual construction that occurs in the progressive, whereas the sentences in (334) are pure progressive sentences with an iterative or repeated reference time: ‘Whenever I pass that house’ provides the repeated restrictors and ‘the dog is barking’ is the progressive. So we have repeated barking events and included in each barking event is an event of me passing the house. Palmer’s (1988) suggestions are along similar lines as those of Leech, as he discusses the progressive as indicating habitual activity in a limited period of time (Palmer 1988:62):  (335) a. He’s going to work by bus. b. We’re eating a lot more meat now.  In addition to being used for limited duration, Palmer argues that the progressive can also be used to indicate habitual activity that is repeated and sporadic (Palmer 1988:63):  (336) a. She’s always breaking things. b. The car’s always breaking down.  In such sentences, what is happening happens very often but not at set times. Palmer contrasts the sporadic use with the simple present, arguing that “if there is reference to repeated points of time, indicating regularity, the non-progressive is used” (Palmer 1988:63):  (337) The car always breaks down when I start for home.  He gives another set of examples as an argument for the difference between the progressive and the simple present (Palmer 1988:64):  142  (338) a. I always break the eggs first. b. I’m always breaking the crockery.  The simple present sentence in (338a) shows regularity in the sense that every time I am baking I start by breaking the eggs. So the sentence would be true even if I bake only sporadically. The progressive sentence in (338b), however, has no such regularity. If it is a habit it is a purely accidental one. It might appear that the difference in (338) is one of intention vs. non-intention. And yet, that cannot be the general difference between the simple tense habitual and the progressive habitual as sentences like ‘John is biking to work these days’ are obviously intentional in some way – even if John’s biking is the result of his car being in the garage. He could after all have walked to work, or taken public transport, or not gone to work at all.  4.3.2  Characteristics of the progressive habitual  In order to determine whether the progressive habitual really is a kind of a habitual construction, we need to see how well it fits the criteria of habitual aspect that I identified above. So let us look back at those characteristics discussed in section 4.2.3.  4.3.2.1 Characteristic of an extended period of time The first characteristic is that habituals refer to something that is typical for an extended period of time. The same can be seen with progressive habituals:  (339) The Canucks are playing well. (340) órsarar eru a leika virkilega vel. órsarar are to play very  well  ‘órsarar are playing really well.’  The meaning of (339) is that during a certain time period which includes ‘now’, when the Canucks play (hockey), they play well.92 In this case we do have an extended period of time in which ‘playing well’ is a characteristic of the Canucks’ play. So in that case, the sentence behaves just like habitual sentences. And yet, the meaning is not quite the same as if it were in the simple present:  92  Notice that the same sentence can also be a regular progressive sentence meaning that at this particular moment the Canucks are playing well. 143  (341) The Canucks play well.  The sentence in (341) makes a claim about the Canucks in general; it is a characteristic of the Canucks team that they play well. It does not mean they are necessarily playing right now and even if they were, it does not mean they are necessarily playing well at the moment. In fact, we can have a regular habitual sentence like The Canucks usually play well co-occurring with a progressive habitual sentence like These days the Canucks are playing horribly:  (342) The Canucks usually play well but these days they are playing horribly.  Here the habitual sentence picks out a general habit whereas the progressive sentence seems to pick out a more temporary, or limited situation. This is along the same lines as the claims made by Leech (1971) and Palmer (1988). I will discuss this difference in more detail in section 4.3.3.  4.3.2.2 General habit but not universal quantification As mentioned earlier, habitual sentences denote a general habit but not a universal quantification, so even though (341) claims that in general the Canucks play well it does allow for exceptions – a game here and there where they play fairly badly. The same can be said about the sentence in (339) – even though they have an occasional bad game they can still be said to be playing well. This can be seen in (343):  (343) A: The Canucks really sucked in the game today. B: I know, but they’re nevertheless playing really well this year. (344) A: órsarar spiluu virkilega illa órsarar played really  í  dag.  badly in today  ‘órsarar played really badly today.’  B: Ég veit I  a. En eir  know that  eru nú  samt a spila mjög vel  but they are now still  í  ár.  to play very well in year  ‘I know, but they are nevertheless playing really well this year.’  So just like ordinary habituals, the progressive habitual can be used even though there are exceptions to the habit, as long as the events are frequent enough to be seen as habits.93 93  The adverbials this year in English and í ár ‘this year’ in Icelandic are not compatible with the simple present. The following B-sentences are both ungrammatical: 144  4.3.2.3 Event does not have to be ongoing at the moment The third characteristic of habituals is that the event does not have to be ongoing at the speech time. The same can be said about the progressive habitual. (339) can be true even if the Canucks are not playing at the moment.94 (339) The Canucks are playing well.  The same can be seen in Icelandic. Imagine a situation where Jón and Páll are sitting at a bar in Reykjavík after having just watched ór play on television. The game was well played and ór won their third game in a row. In such a case it would not be unreasonable for Jón to say to Páll: (345) órsarar eru a spila virkilega vel. órsarar are to play really  well  ‘órsarar are playing really well.’ When Jón says this to Páll the game is already over so ór is not playing at the moment. But as Jón is not referring to that particular game alone but ór’s general play this month/season/etc. the sentence is perfectly felicitous. The sentence can even be true if the Canucks are in fact playing badly at the moment. Let us say John and Paul are watching the Canucks playing a really bad game where they are down by three goals. Then John could well say to Paul:  (346) Man, the Canucks are playing horribly today. It is amazing considering that in general they are playing really well (this year).  Now look at the same sentence in the simple present.  (i)  A: The Canucks really sucked in the game today. B: *I know, but they nevertheless play really well this year. (ii) A: órsarar spiluu virkilega illa í dag. órsarar played really badly in today ‘órsarar played really badly today.’ B: *Ég veit a. En eir spila samt mjög vel í ár. I know that but they play still very well in year ‘I know, but they nevertheless play really well this year.’ At this moment I do not have an answer as to why this is. 94  Notice though that this can hardly be said over the summer, as that is a time when the National Hockey League is not in session and no games take place. This seems to indicate that the pauses allowed between games have some limits. 145  (347) Man, the Canucks are playing horribly today. It is amazing considering that in general they play really well.  There is a subtle difference between the two sentences, whether or not we have the temporal adverbial this year with the progressive sentence. With the simple present we get a general habit that may be applicable to a very long period of time and may in fact refer to the Canucks from the time they were established as a team. That reading is much harder to get with the progressive as the progressive seems to pick out a subinterval of that general time, which is ‘this year’ if we have the adverbial but is unspecified without it. Even though this year is unnecessary for the sentence to be grammatical, we do need the adverbial in general for the sentence to work. I will discuss that in more detail in section 4.3.1.  4.3.2.4 Conclusions When we revisit the characteristics we gave habitual sentences in section 4.2.3 we can see that all also apply to the progressive habitual:  (325)  The progressive habitual •  is characteristic for an extended (but usually temporary) period of time.  •  can have exceptions.  •  does not require the event to be taking place at the speech time.  The only difference is in the first point where habituals are said to convey characteristics for an extended period of time, whereas the progressive habitual says that it is characteristic for an extended but temporary period of time. What this means is that in both cases we have an extended period but in the case of the progressive it is expected to be temporary but with habituals no such expectation is present. Even with this slight difference the similarities are strong enough to say that the progressive habitual really is habitual. The differences between the two are the results of the progressive operator which is present with the progressive but not with the plain habitual.  4.3.3  Temporariness and event time  In section 4.3.1 we discussed how both the habitual and the progressive habitual refer to an extended period of time giving the following examples:  (339) The Canucks are playing well.  146  (341) The Canucks play well.  It appears that the progressive habitual indicates a more temporary habit than the simple present does. In fact, the difference between the progressive habitual in (339) and the regular habitual in (341) is basically the same as the difference between a stative verb in the progressive and in the simple present:  (348)  a. Peter is living with his parents. b. Peter lives with his parents.  In the progressive sentence, a temporary situation is implied, whereas that is not the case with the simple present. I argued in chapter 3 that this is due to a temporariness implicature that arises with stative verbs in the progressive. The reason is that temporariness is a prototypical eventive property that does not occur with stative verbs, and so if the speaker wants to convey temporariness with a stative verb he can shift the verb to being eventive and by doing so implies one or more of the eventive properties. As habituals are in some sense states, as discussed in section 4.2.4, and as habituals in the progressive get the same temporary-reading as states in the progressive, the most plausible explanation is that, just like with states, the temporariness is an implicature. And in fact, (341) is true even if the Canucks always play well. However, most people that hear (341) would assume that the speaker had a reason for choosing the progressive in the sentence and most would understand the sentence to mean that the habit of the Canucks playing well was temporary. Let us explore this temporariness with the progressive habitual further. Suppose that we have a situation where the Canucks usually play well, and these days have been no exception. Then both (339) and (341) are true. However, if the Canucks usually play badly but for about a month they have been playing really well, then (339) is true but (341) is, if not false, at least inappropriate or misleading. It indicates that the Canucks play well in general, whereas they have only been playing well for a few games. As the simple present and the progressive habitual pick out different event times they can co-occur even when they describe two contradictory habits, as we saw earlier. Such sentences are better if adverbs like usually are present to highlight the contrast.  (349) a. ??The Canucks play well but these days they are playing horribly. b. The Canucks usually play well but these days they are playing horribly.  Notice also that the combination of the simple present and the progressive habitual is by far the best way to describe a situation such as the one represented in (349b). Other combinations were deemed either ungrammatical, or, at least, very questionable, by native speakers of English:95 95  The order of the two clauses does not matter. We could just as well say ‘The Canucks are playing horribly but usually they play well’. 147  (350) a. ?The Canucks usually play well but these days they play horribly. b. ?The Canucks are usually playing well but these days they are playing horribly. c. *The Canucks are usually playing well but these days they play horribly.  In (350a) we have both the regular habit and the temporary habit in the simple present. The sentence is not ungrammatical but most English speakers would, nevertheless, choose to use the progressive for the temporary habit. The sentence in (350b) is not ungrammatical either but very awkward at best. We could write (341) as: 96 (351) GEN[e] ([Play(e) Agent (e, Canucks)]; e’ [Manner(e, well)  M(e’)=e])  This can be read as follows: In general if there is an event e of the Canucks playing then that event is played well. This is true for the sentence in (341). However, in (349b) we are in the middle of a period where the Canucks are playing, but are actually playing badly and so for that time period the truth conditions of (351) do not hold. This is because we actually have a habit within the habit. In order for the sentence to work, we need a temporal adverb such as usually or generally to indicate that the usual situation is of a certain kind whereas the temporary situation is of a different kind. So we have established that the progressive habitual can co-occur with the regular habitual. Even though the two can convey what seem to be contradictory statements the fact that they pick out different intervals renders them compatible. In fact, habits are similar to states, as mentioned in section 4.2.4 (see also Carlson 2005). It is, therefore, not surprising that habituals are usually represented by the simple present, just like states (see detailed discussion of states in chapter 2). The following schema is intended to show this:  (352) ea  ea  ea  ea  ea  Habit = State: Simple present used  The schema in (352) tells us that repeated events of type a are taking place. The time line is unspecified and does not say anything about the beginning or the end. The progressive habitual is different, as it seems to pick out a much more defined reference time which instead of representing the time one particular event takes place, picks out a time that includes a  96  One option would be do write the sentence simply as: (i) GEN [e] ([Play(e) Agent(e, Canucks)]; Manner(e, well)) Technically this formula should be sufficient. However, as we require matching events when we relate two events to each other it is better for consistency to keep the matching effect also in sentences such as this one. Here M is an identity function; a special case of a matching function. 148  series of events, of which the state holds. That event time can be included in a bigger interval such as the one given for the regular habitual.  (353)  The progressive habitual ey  ey  ex  ex  ex  ex  ey  ey  Habit = State  SHIFT s: Progressive used  (353) shows a series of events of type x that together form a habit/state. They are surrounded by events of kind y where y stands for an unspecified event of a different kind than x. Progressive habituals are essentially states that have been shifted to an event, as already mentioned in section 4.2.4. Notice that this does not mean that play well is a state, but the habit of playing well is. Also see discussion of this shift in section 4.4.1 below. The reason the two sentences in (349) can co-occur is that (349b) would refer to the events marked ex in the schema in (353), whereas (349a) would refer to the events marked ey.  (354)  The habitual and progressive habitual together  ey  ey  ey  ey  ey  ey  Habit/State  ex  ex  ex  ey  ey  Habit/State  SHIFT s  Here ey stands for events where the Canucks play well and ex stands for events where the Canucks play horribly. The number of ey seems to justify playing well being seen as a general habit and the three or so horrible games can be seen as a temporary habit, which fortunately for the Canucks does not last too long.  4.3.4  Do habitual events have stages?  The main reason I gave in chapter 2 for why regular stative verbs do not occur in the progressive was that statives do not have stages, and according to Landman’s (1992) theory of the progressive, stages are necessary for an event to occur in the progressive. Furthermore, I argued in chapter 3 that when stative verbs are coerced to being eventive, they receive stages. The coerced states then are free to occur in the progressive. As habituals are stative, as previously mentioned, we would assume that habituals also do not have stages and hence cannot be put in the progressive. However, this may depend on how we define  149  stages. When we have a habit of the Canucks playing well, as in (339), we know that this habit can be broken up into individual events of playing (hockey). Each game could, for instance, be seen as one event. So we have Game1, Game2, Game3, etc. Landman’s definition of stages was that something was a stage of an event only if the bigger event was a more developed version of the first. We can hardly say that Game2 is a more developed version of Game1. Let us explore this further. Figure 4-1: Stages of the habit e’, ‘play well’  According to the picture in Fig. 4-1, we do not exactly see each individual game as a stage of e’ but instead Game1 is a stage, Game1 + Game2 is a stage, Game1 + Game2 + Game3 is a stage, etc. In that sense the second stage is a more developed stage than the first one as it includes both Game1 and Game2 instead of just Game2. However, I believe this is misleading. When one has developed a habit, is it really possible to say that every added event of the kind, together with all the previous events, is a further developed stage of that habit? Does it make a difference whether one walks to school fifty times or fiftyone times? Is the fifty-first time really a more developed stage than the fiftieth? If the answer to that is yes, should we not have stages with states as well? Should John’s love for his wife after nine years not be a more developed stage of his love for her after six years? No, I do not think anyone would claim that as there is no development. I do, therefore, suggest, that just like states, habits lack the ability to occur in the progressive. We know that habituals do occur in the progressive, contrary to what one might assume based on the previous paragraph. However, habits should not occur in the progressive for the same reasons that states should not – because they do not have stages and stages are required for the progressive. And yet states do, in fact, occur in the progressive. I argued in chapter 3 that when stative verbs occur in the progressive they have been coerced to being eventive. A special shift is applied to the stative that results in an eventive predicate. This new eventive predicate has stages and can, therefore, occur in the progressive. As stages are a requirement for an eventuality in order to occur in the progressive, we must assume that habituals in the progressive also have stages. And as habits do not have them, as previously argued, the most plausible explanation is that just like states, habits have been shifted to being events when they  150  occur in the progressive. I will turn back to the shift in next section, along with a formal analysis of the progressive habitual.  4.4 Formal analysis of the progressive habitual 4.4.1  The semantics of the progressive habitual  In the progressive habitual we have the progressive operator PROG applying to a habitual sentence like the one in (341), repeated here:  (341) The Canucks play well.  Under the combined Krifka et al. (1995) and Rothstein (2004) analysis we get the following truth conditions for the sentence:  (355) GEN[e] ([Play(e) Agent(e, Canucks)]; e’[Manner(e, well)]  M(e’)=e)  As habituals are in fact stative, as already argued, when we have a habitual sentence we have a state of a habitual generalization holding, such as the state of the Canucks usually playing well. The problem we encounter in the progressive habitual is that PROG cannot apply to the habitual sentence directly as GEN gives us a proposition and not a state. That is, GEN(P,Q) is a statement that can be either true or false so we cannot say that s=GEN(P,Q), as then we would be saying that an entity equals a truth-value. We need PROG to apply to the habitual state of the Canucks playing well. Therefore I introduce the predicate CHAR, instead of GEN. CHAR is a primitive function that is not definable in terms of anything else. Unlike GEN, CHAR is a three-place predicate involving the restrictor P, the nuclear scope Q and the state s. Intuitively, what CHAR(s,P,Q) means is that the state s is characterized by the generalization GEN(P,Q).  (356)  CHAR(s,P,Q) is true only if GEN[e] ([P(e) (e)  (s)], e’[Q(e’)  M(e’)=e])  This can be read in the following way: CHAR(s,P,Q) is true only if in general, for events of type P, where the runtime of the events is a subset of the runtime of the habitual state, there exists an event of type Q that matches the event of type P.97 I cannot give a full proper definition of CHAR in terms of GEN but we 97  Paul Portner (p.c.) has pointed out that my analysis of the progressive habitual has a lot in common with literature on quantification in situation semantics, such as that of Berman (1987) and Portner (1992) and see also von Fintel (1996). This indicates that my analysis could have been done within the theory of situation semantics with good results. 151  can say that if CHAR(s,P,Q) is to be true, then GEN(P,Q) must be true throughout the run-time of s. This cannot be turned around for even though a generalization holds throughout the runtime of a state s, s is not necessarily the state of the generalization holding. For instance, if it happened to be the case that the period during which the Canucks were playing well coincided exactly with Gregor Robertson’s tenure as mayor of Vancouver then it would still not be the case that the state of Gregor Robertson being mayor of Vancouver is the state of the Canucks playing well. The state and the generalization just happened to temporarily coincide. This means that not every formula that has CHAR can be rewritten into one that has GEN.98 When we apply CHAR to ‘The Canucks play well’, before aspect and tense, we get the following denotation:  (357)  s.CHAR(s, e [Play(e)  Agent (e, Canucks)], e’[Play(e’)  Agent(e’, Canucks)  Manner (e’, well)])  We have here the characterizing state of the Canucks playing well. However, our sentence is not yet ready for PROG to apply to it. Recall from chapter 3 that PROG can only apply to events whereas our formula in (357) denotes a set of states. So we need to shift from states to events using the shift rule from chapter 3. Let us look again at that rule.  (224)  [[SHIFT(VP)]]w,g,c = es[s  [[VP]]w,g,c  e=EV(s)]  (358) is the result of applying SHIFT to (357).  (358)  es[CHAR(s, e[Play(e)  Agent(e, Canucks)], e’[Play(e’)  Agent(e’, Canucks)  Manner(e’, well)])  e=EV(s)]  This can be read in the following way: s is a state which is characterized by the fact that during the runtime of s, an event of which it is true that there is a Canucks hockey is matched by an event of which it is true that the Canucks play well. Now we can apply PROG to the event:  98  GEN, as used by Krifka et al. (1995) is also an undefined primitive, so I am just replacing one undefined primitive with another one that is closely related to it, but that suits my purposes better. 152  (359)  [[PROG]]w,g,c (es[CHAR(s, e[Play(e) Agent(e, Canucks)], e’[Play(e’) Agent(e’, Canucks) Manner(e’, well)])  e=EV(s)) = tee’w’:<e’,w’>  CON(e,w)  s[CHAR(s, e[Play(e)  Agent(e, Canucks)], e’[Play(e’) Agent(e’, Canucks) Manner(e’, well)])  e’=EV(s)  t  (e)]   In order for (359) to be true we need to assert that there is a habit in progress, which is such that the Canucks are playing well. In (359) we have the progressive operator apply to a habitual predicate, the Canucks playing well, in the same way as it does with a simple event such as ‘running’. In ‘John is running’ we have a time period where it is true that the event of John running is ongoing and in ‘The Canucks are playing well’ we have a time period where it is true that a series of events of the Canucks playing well is ongoing. So what the sentence in (359) does is really to say that the habit of the Canucks playing well is in progress. In section 4.1 I mentioned that the habitual progressive is a different construction from the progressive habitual, so let us now look at the semantics of the habitual progressive in order to show where exactly this difference lies.  4.4.2  The semantics of the habitual of the progressive  Unlike the progressive habitual that needs CHAR instead of GEN, the habitual progressive is a regular progressive construction in the sense that it is the GEN operator that applies to it, as it applies after PROG. Consider the following example:  (360)  Whenever Paul arrives, John is eating.  Here the restrictor is ‘whenever Paul arrives’ and the matrix is ‘John is eating’. We need to apply the GEN operator to the sentence in order to get the habitual reading. GEN has scope over both the restrictor and the matrix. This is the rough denotation of the sentence ‘whenever Paul arrives, John is eating’, in a world w and a context c with an assignment function g. However details will need to be worked out.  (361)  GEN[e]([Arrive(e)  Agent(e, Paul)]; e’e’’w’[M(e’)=e <e’’,w’>  CON(e’, w)  [Eat(e’’)  Agent(e’’, John)]  (e)  ( e’)])  I have not included tense in the sentence but it would presumably come in after GEN and have scope over the whole sentence. It is also not spelled out how exactly the progressive relates the two events to one another instead of relating the event time to the reference time as aspect generally does. 153  According to this analysis, unlike in the progressive habitual, the habitual of the progressive does not directly stack the aspectual operators but instead the GEN operator quantifies over the whole sentence whereas the PROG operator only applies to the matrix. This is a bit sketchy because the detailed semantics of when-clauses is beyond the scope of my thesis and for the present purpose the main point is the difference between the progressive habitual and the habitual progressive.  4.5 General conclusions In this chapter I have discussed a construction that I call the progressive habitual. It is really the progressive of habitual sentences. Instead of saying that an event is in progress, as is usually said about the ‘standard’ progressive, we say that a habit is in progress. This is confirmed by the fact that the progressive habitual behaves in every way like a regular habitual sentence, except for the implicature of the event time being more temporary than it is with the standard habitual. I proposed a formal analysis of the progressive habitual, arguing that being aspectually stative, habits do not have stages and therefore cannot be progressivized like regular events. Therefore, in order to occur in the progressive, a habitual sentence first needs to be shifted to being an event, using the shift rule established in chapter 3 for stative verbs. I assume the habitual meaning of a progressive habitual is given by a quantificational operator, CHAR, which relates one set of conditions to another set, the restrictor and the matrix, giving us sentences that generalize over situations of a certain kind. The truth conditions of the habitual itself are those of a standard habitual but when the progressive operator is applied we get the reading that there is an ongoing habit at the reference time. Just like when we have states in the progressive, habits in the progressive implicate certain prototypical eventive properties. The use of the progressive indicates that the habit has been shifted to an event and the hearer assumes this has been done because the speaker wants to convey one or more of the prototypical eventive properties. With habits that is usually the property of temporariness. Basically, by stating that a habit is ongoing at the moment, the hearer understands the sentence to mean that this is not a permanent situation as otherwise the simple present would have been used. This is in fact a very plausible motivation for the use of the progressive.  154  5 The present participle progressive 5.1  Introduction  As mentioned in chapter 1, Icelandic has more than one way to indicate that an event is in progress. Most commonly the infinitive form of the verb is preceded by the copula, but just like in English, Icelandic can also use the present participle of the main verb. I will argue in this chapter that this construction is first and foremost used with repeated events in progress rather than single, ongoing events. In Icelandic, the present participle progressive is generally not used with event verbs in order to convey that an event is ongoing, unlike in English:  (362) a. #Jón er borandi. Jón is eating (pres part) ‘John is eating.’  b. #Jón er hlaupandi. Jón is running (pres part) ‘Jón is running.’  However, with adverbs like alltaf ‘always’, sífellt ‘constantly’, and the prefix sí ‘constantly’ the use of the present participle progressive is much the same as that of the habitual progressive.99  (363) a. Jón er sífellt  borandi.  (Present Participle)  Jón is constantly eating (pres part) ‘Jón is constantly eating.’  b. Jón er sífellt  a bora.  Jón is constantly to  (Infinitive)  eat (inf)  ‘Jón is constantly eating.’  In Jóhannsdóttir (2005, 2007a) I claimed that the present participle progressive requires one of these adverbs in order to be grammatical, but I will argue here that this is not the case, as the present participle 99  ráinsson (1974:46) talks about adverbs like alltaf ‘always’ and ávallt ‘always’ with the infinitival progressive. 155  progressive can get an iterative reading without an adverb present. However, if the context is clearly noniterative the present participle cannot be used. The exceptions to this generalization are verbs like sofandi ‘sleeping’ and sitjandi ‘sitting’, which will be discussed in section 5.2.3.100 The adverbs that appear in sentences with the present participle progressive tend to be a subset of temporal adverbs, which I will follow Csirmas (2009) in calling adverbs of quantity. These can be divided into three groups: frequency adverbs, adverbs of quantification and cardinal adverbs. With the present participle progressive their role seems to be to narrow down the quantity or frequency of the repeated event. Their function, however, varies somewhat based on the verbal form they appear with. For instance, I will show that when we have an adverb like stöugt ‘constantly’ in a simple past tense sentence with a relatively short event time, the event cannot be interrupted. If, on the other hand, the adverb appears with the present participle progressive, breaks are allowed:  (364) Jón hljóp stöugt Jón ran  frá  sjö  til ellefu í  morgun.  continuously from seven to eleven in morning  ‘Jón ran continuously from seven to eleven this morning.’  (365) Jón var stöugt  hlaupandi  frá  sjö  til ellefu í  morgun.  Jón was constantly running (pres part) from seven to eleven in morning ‘Jón was constantly running between seven and eleven this morning.’  I will argue that the verbs and the adverbs interact in such a way that which verbal form is used affects the reading of the adverb whose main role is to provide more details about the frequency of the iterative event. That is, when we get adverbs like stöugt ‘constantly’ with the present participle progressive, breaks are expected, but not when stöugt appears with the simple present/past. The main goals of this chapter are thus to answer the following three questions:  •  Why does the present participle progressive generally occur with adverbs of quantity such as always and constantly?  100  There is another use of the present participle that I am ignoring in this dissertation as it is not a progressive construction. These are statements like (i) that are particularly common on social networking websites like Twitter and Facebook as well as in captions of pictures showcased on the Internet: (i) Borandi franskar á sundlaugarbakkanum. Eating (pres part) french.fries on pool-side.the ‘Eating french fries on the side of the pool.’ http://www.flickr.com/photos/margretros/3392912462/  These are not complete sentences, as the auxiliary verb required by the present participle is missing. Instead we have here an adjunct.  156  •  Is the meaning of present participle progressive sentences with temporal adverbials the same as that of the infinitival progressive?  •  What effects do adverbs of quantity have on progressive and habitual sentences?  I will argue that the present participle progressive never refers to one ongoing event, but a series of ongoing events over a certain period of time. The present participle progressive, therefore, differs from the infinitival progressive, which can either denote one particular ongoing event or multiple events. This means that if we have multiple events, either progressive construction can be chosen, but if there is only one ongoing event the infinitival progressive is used. Furthermore, I will argue that the adverbs of quantity bring out the iterative nature of the sentences. Adverbs like stöugt ‘constantly’ get a different interpretation based on which progressive construction they occur with. The chapter is organized as follows. In section 5.2 I will discuss the present participle construction in Icelandic and argue that it is in fact used as a progressive construction in certain cases – when the verb denotes iterative events and not just one single event. In these cases, adverbs of quantity, such as alltaf ‘always’ and stöugt ‘constantly’ are frequently present in the sentence, though they are not necessary. I also show that the present participle does occur in a few cases where we do not have iterative events, but in these cases the verb has usually been nominalized or appears as an adjective and not as a verb. These cases are therefore quite different from the progressive ones. As the present participle progressive always occurs with iterative events, it appears to be very similar to the progressive habitual, discussed in chapter 4. As I argued there that the progressive habitual really is a habitual construction, it is worth asking whether the present participle progressive is a habitual construction as well. This is the topic of section 5.3 where I use the habitual diagnostics from chapter 4 on the present participle progressive. Interestingly enough, the present participle progressive passes all the habitual tests, and so it should be analysed as a habitual. However, there is a difference between the present participle progressive and the progressive habitual. For instance, the present participle progressive has a much more limited use and can only occur in a subset of the contexts that allow the progressive habitual. Because of the importance of adverbs of quantity with the present participle progressive I discuss these adverbs in section 5.4. As previously mentioned I divide them into three groups: frequency adverbs, adverbs of quantification and cardinal adverbs. Adverbs of quantification and frequency adverbs frequently occur with the present participle but I found only one example with a cardinal adverb. I will therefore not discuss cardinal adverbs further but focus on the other two types.101 In 5.4.2 I use some diagnostics to show the differences between the two groups and in 5.4.3 I provide their semantics. I argue 101  I am not sure why cardinal adverbs do not occur more commonly with the present participle progressive. It would be interesting to examine that further. 157  that frequency adverbs differ from adverbs of quantification in that they have a relevance operator that changes the reading of the adverb based on context. For instance, the reading of stöugt ‘constantly’ ranges from allowing no breaks, to allowing frequent breaks – all based on the length of the relevant time and the construction of the VP. In section 5.5 I discuss the semantics of the present participle progressive. I will argue that this construction differs from the infinitival progressive (and the progressive in English) in that it carries a presupposition that there is pluractionality. In 5.6 the conclusions of the chapter are drawn together.  5.2 The present participle construction 5.2.1  The present participle as a progressive construction  The present participle construction in Icelandic is similar to that of English as it is formed with the verb be and the present participle suffix –andi attached to the main verb. However, unlike in English, this is not the common way to show that an event is in progress. Instead Icelandic uses the infinitive for that. The Icelandic equivalent to a simple progressive sentence in English, like the ones in (366), would be (367) and not (368):  (366) a. John is eating. b. Anne is watching TV.  (367) a. Jón er a bora.  (Infinitival progressive)  Jón is to eat (inf) ‘Jón is eating.’  b. Anna er a horfa  á sjónvarp.  (Infinitival progressive)  Anna is to watch (inf) on TV ‘Anna is watching TV.’  (368) a. #Jón er borandi.  (Present participle)  Jón is eating (pres part) ‘Jón is eating.’  b. #Anna er horfandi  á sjónvarp.  (Present participle)  158  Anna  is watch (pres part) on TV  ‘Anna is watching television.’  So why is it then that I talk about the present participle progressive in the introduction to this chapter? Consider the following examples:  (369) a. Jón er alltaf borandi  egar ég kem í  Jón is always eating (pres part) when I  vinnuna.  come to work.the  ‘Jón is always eating when I come to work.’  b. Jón er alltaf a bora  egar ég kem í  vinnuna.  Jón is always to eat (inf) when I come to work.the ‘Jón is always eating when I come to work.’  In these sentences the present participle progressive and the infinitival progressive have the same meaning and the constructions seem interchangeable. We get the reading ‘Always when I come to work, John is eating’. It is clear that both sentences say that John’s eating has already started by the time I come to work and it may or may not continue after I arrive. This is the habitual of the progressive. Compare that to the simple present sentence in (370):  (370) Jón borar alltaf Jón eats  egar ég kem í  vinnuna.  always when I come to work.the  ‘Jón always eats when I come to work.’  The meaning of (370) differs from that of (369) as in (370) John starts eating when I come to work, instead of the eating already being in progress. So the present participle progressive gets a clear imperfective reading just like the infinitival progressive and unlike the perfective reading of (370). There are even cases where the present participle progressive and the infinitival progressive appear to be used interchangeably. The following sentence is interesting for the fact that both kinds of progressive are mixed together:  (371) Ef  vi erum ekki sofandi  á 159  erum vi borandi,  ef vi erum  If  we are  not  sleeping (pres part) then are á  Ekki borandi not  erum vi a  eating (pres part) then are  ekki a kaupa  í matinn  á  we to  we eating (pres part) if we are kaupa  matinn, og ef vi erum  buy (inf) in food.the and if we are  erum vi a horfa  not to buy (inf) in food.the then are  í  á family guy í  fartölvunni!  we to watch (inf) on Family Guy in laptop.the  ‘If we are not sleeping then we are eating, if we are not eating then we are buying groceries, and if we are not buying groceries, then we are watching Family Guy on the laptop.’ http://brynja86.bloggar.is/blogg/158106/Letin_gerir_engum_gott  5.2.2  The iterative nature of the present participle progressive  In section 5.2.1 I showed how the present participle progressive, which is usually not used with regular event verbs, becomes perfectly grammatical when there is an adverb like alltaf ‘always’ in the sentence:  (372) a. #Jón er borandi. Jón is eating (pres part) ‘Jón is eating.’  b. Jón er alltaf  borandi.  Jón is always eating (pres part) ‘Jón is always eating.’  However, the presence of alltaf is not necessary for verbs to occur in the present participle progressive, as long as it is clear that we have repeated events, as can be seen in the following examples:  (373) a. Ef If  maur ætlar one á  a yngja  sig almennilega á essum mat, arf  intends to make.heavier self properly ekki a  then not  to  vera borandi be  on this  maur  food needs one  24/7?  eating (pres part) 24/7  ‘If one intends to gain weight properly from this food, does not one have to be eating 24/7 then?’ http://vaxtarvorur.is/index.php/spjall/39-naeringarprogram-a-naeringarfraeei/9307-bulking-matur?limit=6start=6  160  b. ...og hún ekkir mig og veit and  a  ég er  ekki  she knows me and knows that I am not útum  allan bæ,  out.around all  í  berjandi  fólk  beating (pres part) people  nafni hundanna minna.  town in name dogs.the  mine  ‘...and she knows me and knows that I am not beating up people all over town, in the name of my dogs.’ http://www.hundaspjall.is/phpbb/viewtopic.php?p=142648sid=62296b72c24abf3bb59a424463a6779f  c. Ótrúlegt  hva maur tekur uppá öllu öru en  Unbelievable what one Prófin  takes up  nálgast!  Er  og  egar  all other than to study like.this when  t.d. búin  a vera me eitthverja tuskuski,  Exams.the come.near am e.g. finished to be skúrandi  a læra svona  rífandi  with some  útum  rag.sickness  allt heheheh veit  washing.floors (pres part) and cleaning (pres part) out.around all heheheh know a  a  eru margir sem kannast  that there are many  vi  etta.  that recognize with this  ‘Unbelieveable how one starts doing everything apart from studying when the exams close in! I’ve had some kind of rag-sickness, washing floors and cleaning all over the place, hehehehe, I know there are many that recognize this (behaviour).’ http://óreyjo.bloggar.is/blogg/page4  In (373a) the time adverbial 24/7 quantifies over times and in this case has a similar meaning to always as it means that there are many eating events that go on 24 hours a day, seven days a week. In (373b) we do not have a time adverbial, but the place adverbial út um allan bæ ‘all over town’ gives us many places and therefore many events – or actually lack of events. The speaker claims not to be beating up people all over town. This indicates that we have a frequency operator quantifying over events. In (373c) we have the locative út um allt ‘all over (the place)’ which on its own does not give us a repeated event. However, the use of the present participle progressive (and possibly the compound tuskuski (literally: rag-sickness)) indicate repeated cleaning events. The speaker claims to have been crazy with the rag, cleaning repeatedly. It is never said clearly that there were repeated events of cleaning but the combination of the choice of words and the present participle progressive make it evident. If the speaker had cleaned only once during the study period she would most definitely have used the simple past of the verb:  161  (374) ... er... búin  a vera me eitthverja tuskuski,  ... am... finished to be útum  with some  skúrai  og reif  rag.sickness washed.floors and cleaned  allt ...  out.around all... ‘...have had some rag-sickness, washed the floors and cleaned all over the place...’  The simple past gives us an episodic sentence where we have no reason to believe there was more than one event of washing and cleaning, whereas the present participle progressive rather indicates repeated events of such major cleaning. Notice that with adverbs like stöugt ‘constantly’ the present participle progressive still gives an iterative reading. This can be seen in example (364), repeated here:102  (364) Jón var stöugt  hlaupandi  frá  sjö  til ellefu í  morgun.  Jón was constantly/continuously running (pres part) from seven to eleven in morning ‘Jón was constantly running from seven to eleven this morning.’  Here we get the reading that there was more than one running event. John may have run to the store to buy milk. When he got back from there, he ran to his sister’s house, etc. We do not get the reading that he went out on a run that lasted four hours. This is clear when we give a context that restricts the reading of the sentence to one running:  (375) *Jón hljóp langa hlaupi sitt í Jón ran  morgun.  Hann  long run.the his in morning. He frá  sjö  var stöugt  hlaupandi  was constantly running (pres part)  til ellefu.  from seven to eleven ‘Jón ran his long run this morning. He was continuously running from seven to eleven.’  When the context is restricted in this way, where the present participle progressive refers to only one running event, Icelandic speakers reject the sentence and choose to use the simple past (here repeated from (365)): 103,104  102  Interestingly, the sentence ‘John was constantly running from seven to eleven this morning’ in English is not a perfectly acceptable sentence. The English constantly may require a longer event than the Icelandic stöugt needs. This is an interesting issue for further study. 103  The Icelandic adverb stöugt appears to differ based on the aspect of the verb and because of that it can be translated into English with either constantly or continuously, depending on its reading each time.  162  (365) Jón hljóp stöugt Jón ran  frá  sjö  til  constantly/continuously from seven to  ellefu í  morgun.  eleven in morning  ‘Jón ran continuously from seven to eleven this morning.’  In the simple past we get the reading that there was one long running event that started at seven and continued without stopping until eleven o’clock. Because of the different aspect we also get different interpretations for stöugt ‘constantly’. In the simple past sentence we expect John not to have stopped at all (except maybe while waiting on a red light) but in the present participle progressive where we have multiple running events we do not only allow there to be short breaks between events but we actually require it.105 As discussed in chapter 2, stative verbs do not occur in the progressive in English or Icelandic but can be coerced to being eventive and as such do occur in the progressive. The following examples are from chapter 3: ú kæmir  (376) a. Vi vorum a halda a We were  to  think  that  ekkert.  you would.come not  ‘We were thinking you were not coming.’  b. Ég er I  a elska etta lag.  am to love this  song  ‘I am loving this song.’  However, stative verbs cannot occur in the present participle progressive at all:  (377) a. *Vi vorum haldandi a We were  ú kæmir  ekkert.  believing that you would.come not  ‘We were thinking you were not coming.’  b. *Ég I  er  elskandi etta lag.  am loving  this  song  ‘I am loving this song.’ 104  When I asked Icelandic speakers about the meaning of the two sentences in (364) and (365) they all agreed on the iterative nature of the present participle and the ones that mentioned the adverb also all agreed that breaks were allowed in (364) but not in (365). 105  In Jóhannsdóttir (2005, 2007a) I argued that there was a relevance operator R in the truth function of frequency verbs that controlled how often an event had to take place and how many breaks there can be. 163  This is true even if there is an iterative adverb in the sentence:  (378) a. *Vi vorum alltaf We  were  always  haldandi a  ú  believing that you  kæmir  ekkert.  would.come not  ‘We were always thinking you were not coming.’  b. *Ég I  er  alltaf  elskandi etta lag.  am always loving  this  song  ‘I am always loving this song.’  If stative verbs can occur in the infinitival progressive after being coerced to being eventive, why can they not occur in the present participle progressive? The most plausible explanation is that the iterative nature of the present participle is the reason why the states do not occur in the present participle progressive as states are usually not iterative. However, if we do have an iterative state our theory would predict that it could occur in the present participle progressive. And (379) is certainly much better than (378):  (379) María er alltaf elskandi einhverjar fyllibyttur. María is always loving  some  drunks  ‘María is always loving some drunks.’  In (379) we have many instances of Mary loving a drunk. So there is the state of Mary loving drunk#1, then there is another instance of Mary loving drunk#2, etc. In (378) the subject of Mary’s affection, however, is a particular song. If she always loves the song then there must be one continuous loving. If there are times when she does not love the song then we can hardly say that she always loves the song. The fact that the subject of Mary’s affection in (378) is one particular song but in (379) it is many different drunks shows clearly the difference between the iterative and the non-iterative loving. Stative sentences are usually not iterative but when they are there should be nothing preventing them from occuring in the present participle progressive.106 106  I am not prepared to say that every state would be used in the present participle progressive as long as it could be seen as iterative. Let us say that someone has Alzheimers and so keeps forgetting certain facts. In the mornings she gets clearer and then remembers these facts that she keeps forgetting when she is more tired. In that sense we can say that she keeps on remembering these facts and yet María er yfirleitt munandi etta á morgnana ‘Mary is usually remembering this in the mornings’ is odd. The equivalent progressive sentence is perfectly fine in English, as long as it is a temporary habit. For instance if the nurse in the care home is telling the child what María is remembering in the mornings these days. But even though the sentence is bad in Icelandic, it is perfectly fine to say Jón er alltaf munandi hina 164  5.2.3  The bare present participle construction  Although I claimed in section 5.1 that the present participle progressive can only be used with multiple events, there are cases where this is not true. These are sentences such as the following:  (380) a. Jón er sitjandi. Jón is sitting (pres part) ‘Jón is sitting.’  b. María er sofandi. María is sleeping (pres part) ‘María is sleeping.’  Here we do not have multiple events but an ongoing situation of John being in a sitting position and Mary being asleep. Both are translated into English using the regular progressive construction and the meaning does indeed appear to be progressive. However, notice that in both sentences we have a stative situation, and as was discussed in chapters 2 and 3, stative verbs usually do not occur in the progressive without having first been coerced to being events.107 Höskuldur ráinsson (p.c.) has suggested that the Icelandic present participle forms in these sentences are actually adjectives.108 If that is true they resemble the English adjectival sentence in (381) and the French adverbial sentence in (382) and should be seen as stative:  (381) John is asleep.  undarlegustu hluti ‘John is always remembering the weirdest things’ if John tends to remember the weirdest things that his friends keep forgetting. 107 In chapter 3 I argued that posture verbs are eventive in English and stative in Icelandic. So only in Icelandic do posture verbs need to be coerced to being eventive in order to occur in the progressive. 108  The present participle also occurs as a noun in deverbal agentives such as eigandi ‘owner’, verjandi ‘defender; defence lawyer’ and nemandi ‘student’ (from eiga ‘own’, verja ‘defend’ and nema ‘study’, respectively), but in such cases a full lexicalization has taken place and the words really are nouns and not just verbal forms used as nouns. This is seen clearly by the fact that they have a full nominal inflectional paradigm: (i) Nom. sg. nemandi pl. nemendur Acc. nemanda nemendur Dat. nemanda nemendum Gen. nemanda nemenda  165  (382) Jean est debout. Jean is  stand (adv)  ‘Jean is standing.’  This adjectival use of these present participles can also be seen when they refer to the method of transportation such as gangandi ‘walking’ and keyrandi ‘driving’:109  (383) a. Fjöldi  fólks  skoar eldgosi  gangandi  ea úr  Number people watch eruption.the walking (pres part) or  lofti.  from air  ‘Many people go to see the eruption either by foot or by air.’ http://www.aeska.is/i/news/e_naesti_umsoknarfrestur_hja_aeskulydssjodi/?cal_month=3  b. Hann áætlar He  a  um  400 manns séu nú á lei gangandi  upp  estimates that about 400 people are now on way walking (pres part) up Fimmvöruháls... Fimmvöruháls  ‘He estimates that about 400 people are now on their way walking up Fimmvöruháls.’ http://www.ruv.is/frett/gonguhopar-leggja-fimmvorduhals  The adjectival use of the former verb is clear when it is used to modify nouns:  (384) a. Ég b í I  sem er tiltölulega nálægt éttbli  sveit  live in countryside that is rather ví  oft  a  close  og a gerist  urban.area and it  happens  hundaeigendur koma upp veginn, heim til mín í  therefore often that dog.owners keyrandi  göngutúr.  driving (pres part)  walk.trip  come up road.the home to me in  ‘I live in the countryside, fairly close to an urban area, and it happens often that dog owners come up the road to my place for/on a driving walk.’110 http://www.hugi.is/hundar/articles.php?page=viewcontentId=1632152  109  It might be just as accurate to call this construction adverbial rather than adjectival, as the meaning is similar to that of manner adverbs, but at the moment we do not really have any reason to analyse them as one over the other.  110  Keyrandi göngutúr ‘driving walk’ is a fairly new term referring to the situation where lazy dog owners take their dog out for a walk except instead of the owner walking with the dog he actually drives his car and the dog runs alongside it. 166  b. a er ekki eins og It  is not  like  keyrandi  konur á háum hælum sé eitthva  driving (pres part) women on high heels  ntt.  is something new  fyrirbæri. phenomenon ‘It is not like driving women in high heels is a new phenomenon.’ http://blogg.visir.is/yrr/2007/06/.../serhannair-skor-fyrir-keyrandi-konur/  The present participle forms of posture verbs (e.g. sitjandi ‘sitting’) are used to specify in what position someone or something is. Compare (385b) and (385c) to a simple locative like (385a). All give the bed as the location of Mary but (385b) and (385c) furthermore state that she is in a lying-down position:  (385) a. María er í  rúminu.  María is in bed.the ‘María’s in bed.’  b. María liggur í María lies  rúminu.  in bed.the  ‘María is lying in bed.’  c. María er liggjandi  í  rúminu.  María is lying (pres part) in bed.the ‘María is lying in bed.’  But even though the present participle in (380) and (385c) may be adjectival in Icelandic, this does not mean that it does not indicate an ongoing situation. However, whether we have here a regular progressive sentence or some kind of locative predication, with the present participle as some kind of manner modification, it is clear that these sentences differ considerably from the present participle progressive with frequency adverbials. For instance, these adjectival sentences do not need an adverb of quantity or any indication of iterativity, as can be seen in (380) and (385c), none of which include an adverb of quantity or refer to an iterative situation. These predicates are quite interesting but as I believe they are adjectival they are not the same construction as the present participle progressive and therefore we do not have any reason to expect iterativity with them. It is a worthwhile task for the future to give a detailed description of all the constructions in Icelandic in which the present participle appears.  167  5.3 The present participle vs. the progressive habitual The fact that the present participle progressive always denotes iterative events and that it can always be replaced by the progressive habitual makes one wonder whether the present participle progressive might be a habitual construction like the progressive habitual. Let us start by looking at the habitual diagnostics I introduced in chapter 4 and see whether they apply to the present participle progressive. These characteristics were:  • Characteristic of an extended period of time. • General habit but not universal quantification. • Event does not have to be ongoing at the moment.  5.3.1  Characteristic of an extended period of time  Both habituals and the progressive habitual, which is a special use of the infinitival progressive, refer to a habit that is characteristic for an extended period of time:  (386) a. Jón hjólar í Jón bikes  vinnuna.  to work.the  ‘Jón bikes to work.’ b. órsarar eru a spila  virkilega vel.  órsarar are to play (inf) really  well  ‘órsarar are playing really well.’  So, do sentences in the present participle progressive refer to an extended period of time as well? Look at the following example:  (387) María er alltaf  lærandi.  María is always studying (pres part) ‘María is always studying.’  (387) means that during a certain unspecified time period María is doing a lot of studying. The sentence is quite comparable to the sentences in (386) as in all cases we have an extended reference time, which includes multiple events of the same kind.  168  5.3.2  General habit but not universal quantification  As previously mentioned, habituals and the progressive habitual indicate a general habit but exceptions are allowed. This means that even if a universal quantifier is present in the sentence it does allow breaks. In (387) we have the reading: ‘During time period X María is always studying.’ The adverb alltaf ‘always’ is usually the universal quantifier and so the sentence might be expected to read: ‘For all times X, María is studying during that time’. However, in (387) that is not really the case as breaks are allowed and the sentence rather means something like ‘María is studying a lot.’ She does not have to be doing so at all times and she does not even have to be doing so at all relevant times. The sentence would be true even if there are, for instance, a couple of evenings a month where she watches television instead of studying.  5.3.3  Event does not have to be ongoing at the moment  The third characteristic of habitual sentences is that an event does not have to be ongoing at the time of evaluation. Again, this applies to the present participle progressive. We can easily say (387) even though María is not actually studying at the time of utterance. In chapter 4 I discussed how habitual sentences allow exceptions and it is clear that one of these exceptions can be taking place during the time of utterance. Imagine, for instance, parents looking at their teenage daughter crashed out on the couch:  (388) A: Æ, sjáu greyi. Aw see  Hún sefur.  poor.thing.the She  sleeps  ‘Aw, look at the poor thing. She is sleeping.’ B: a er elilegt a It  is natural  hún sé reytt. Hún er alltaf  that she is tired  she  lærandi.  is always studying (pres part)  ‘It is natural that she is tired. She is always studying.’  Here the girl is obviously not studying as she is sleeping, and yet the sentence uttered by B is perfectly felicitous.  5.3.4  Conclusion  From this section we can see that the characteristics of habituals also apply to the present participle progressive – it refers to repeated events during an extended period of time but the events that form the habit do not have to be happening constantly and they do not have to be happening at the time of  169  speaking. So the present participle progressive appears to be a habitual construction just like the habitual progressive. The present participle progressive however differs from the infinitival progressive in that the former has to be iterative whereas the latter does not. In that way the present participle progressive is similar to the progressive habitual as it picks out series of events of a certain type that take place over an unspecified period of time. Because of how commonly adverbs of quantity occur with the present participle progressive, I will now look at these adverbs, how they are sub-categorized and how each group interacts with the present participle progressive.  5.4 Adverbs of quantity with progressive sentences 5.4.1  Adverbs of quantity  As I have pointed out, event verbs are perfectly natural in the present participle progressive when an adverb like alltaf ‘always’ or sífellt ‘constantly’ is in the sentence, giving the sentence an iterative reading. I argued in section 5.2.2 that the present participle progressive requires an iterative reading and even though it does not require iterative adverbs, such adverbs almost always occur with the present participle progressive. It is not clear why that is. Their role is to give additional information on the frequency of the events within the habit. The adverbs interact with the rest of the sentence and the context in such a way that they influence the semantics and pragmatics of the sentence as a whole. Because of this I want to take a better look at the adverbs themselves. For instance, consider the following examples:  borandi  egar  (389) a. Jón  er  alltaf  Jón  is  always eating (pres part) when  ég  kem heim.  I  come home  ‘Jón is always eating when I come home.’  b. Páll er alltaf borandi  egar ég er  hjá honum.  Páll is always eating (pres part) when I am at  him  ‘Páll is always eating when I am at his place.’  There is a difference between the alltaf ‘always’ in (389a) on the one hand and the alltaf in (389b) on the other: The meaning of (389a) is something like ‘every time I come home, John is eating’, whereas the meaning of (389b) is ‘(every time) when I am at Paul’s place he is constantly eating’. I have previously argued (Jóhannsdóttir 2005, 2007a) that in these two sentences alltaf is occurring as two different kinds of adverb; it is an adverb of quantification in (389a) and quantifies over the whole sentence, whereas in  170  (389b) it is a frequency adverb that only modifies the VP. In this section I will take a closer look at the adverbs that occur with the present participle progressive and will argue that the present participle progressive occurs with any kind of adverb that conveys iterativity, whether that is a frequency adverb, an adverb of quantification or something else. Csirmaz (2009) has argued that adverbs that specify the quantity or frequency of multiple situations can be divided into three categories: Firstly there are adverbs of quantification (see e.g. Lewis (1975), Heim (1982), de Swart (1991)), such as always, often and sometimes, which quantify over occasions or times of occurrence and are shown in (390a). Then we have frequency adverbs, which are adverbs such as constantly and frequently, shown in (390b), and which Van Geenhoven (2005) defines as pluractional operators, similar to those of Lasersohn (1995). Thirdly we have adverbs like twice, five times, many times, which Van Geenhoven (2005) calls cardinal adverbs which are adverbs that specify the cardinality of multiple situations. Like frequency adverbs, cardinal adverbs can themselves introduce the plurality of a situation and so can be said to be pluractional operators. These are shown in (390c).  (390) a. Kenny always sings the harmony. b. Roly frequently plays his silver bass during concerts. c. Chris picked on Hershel at least twice during the concert.  Csirmaz (2009) groups these three types of adverbs together under the label adverbs of quantity. Notice that all these adverbs are also temporal adverbs but the group of temporal adverbs would also include adverbs like yesterday and tomorrow that directly locate events in time. As such adverbs have no impact on the availability of the present participle progressive with event verbs I will not discuss them further. I will also not discuss cardinal adverbs further as they do not in general occur in the present participle progressive, and instead focus on the other two types, adverbs of quantification and frequency adverbs.  5.4.2  The different behaviour of adverbs of quantity  There are several differences between adverbs of quantification and frequency adverbs. I will now discuss various tests that can be used to clearly show these differences.  5.4.2.1 Binding of indefinites (unselective binding) Lewis (1975) and Heim (1982) analysed adverbs of quantification in such a way that they form tripartite structures where the when-clause (or an if-clause) provides the restrictor, if present. Heim argued that adverbs of quantification are the only temporal adverbials that can unselectively bind indefinites. Donkey  171  sentences without an adverb of quantification have an implicit universal quantifier that we get for free; the translation of (391) therefore has the universal quantification.111  (391) Ef bóndi á  asna,  á  lemur hann hann.  If farmer owns donkey then beats  he  him  ‘If a farmer owns a donkey, he beats it.’ =x,y([farmer(x)  donkey(y)  owns(x,y)], [beats(x,y)])  This means that when we have an overt universal quantifier the reading is the same: egar bóndi á  (392) Alltaf  asna,  á  lemur hann hann.  Always when farmer owns donkey then beats  he  him  ‘Always, if a farmer owns a donkey, he beats it.’ =x,y ([farmer(x)  donkey(y)  owns(x,y)], [beats(x,y)])  However, it is also possible to have other adverbs of quantification with donkey sentences where the quantification changes accordingly:  (393)  Oft  egar bóndi á  Often if  asna,  á  lemur hann hann.  farmer owns donkey then beats he  him  ‘Often, if a farmer owns a donkey, he beats it.’ =OFTEN x,y ([farmer(x)  donkey(y)  owns(x,y)], [beats(x,y)])  Frequency adverbs do not bind indefinites, as can be seen in (394) where the adverb of quantification does not enforce a proportion of donkey-beaters among the donkeys, like an adverb of quantification would.112  111  I follow here the truth conditions of Heim (1982), who did not at the time use either times or events/situations in her framework. When I adapt the adverbs of quantification to my analysis later in this chapter, I will be using both. 112  Paul Portner (p.c.) has pointed out to me that it is not true that frequency adverbs cannot bind indefinites and gives examples like “When a farmer owns a donkey, he frequently beats it.” I do believe that in such cases frequently is, in fact, not a frequency adverb but an adverb of quantification and is in that way similar to adverbs like alltaf which can occur as either kind, and will be discussed in more detail in section 5.4.2.5. Constantly, however, never serves as an adverb of quantification.  172  (394) Ef If  bóndi á  á  asna,  lemur hann hann stöugt.  farmer owns donkey then beats  he  him  constantly  ‘If a farmer owns a donkey, he constantly beats it.’  In (394) we have an implicit universal quantifier and the sentence is interpreted as follows:  (395)  x,y ([farmer (x)  donkey (y)  owns (x,y)], [constantly beats (x,y)])  This is the same reading as when there is an overt adverb of quantification such as alltaf in the sentence:  (396)  Alltaf egar  bóndi á  always when  farmer owns donkey then beats  asna,  á  lemur hann hann stöugt. he  him  constantly  ‘Always, if a farmer owns a donkey, he constantly beats it.’ =x,y ([farmer (x)  donkey (y)  owns (x,y)], [constantly beats (x,y)])  The adverb of quantification has scope over the whole sentence but the frequency adverb, on the other hand, only modifies the verb of the matrix clause and says that the farmer ‘constantly beats the donkey’. The two adverbs (whether the universal quantification is represented by an overt adverb of quantification or not) apply to two different levels of the sentence.  5.4.2.2 Preposing In Icelandic, adverbs of quantification can be preposed, whereas frequency adverbs cannot, as shown in (397) and (398).  (397)  Subj – V – Adv – when-clause a. Jón borar oft Jón eats  egar  often when  ég  heimsæki hann.  I  visit  him  ‘Jón eats often when I visit him.’  b. Jón borar stöugt Jón eats  egar  constantly when  ég  heimsæki hann.  I  visit  ‘Jón eats constantly when I visit him.’  173  him  (398)  Adv – V – Subj – when-clause a. Oft  borar Jón egar  Often eats  Jón when  ég heimsæki hann. I visit  him  ‘Jón often eats when I visit him.’  b. *Stöugt  borar Jón egar  constantly eats  Jón when  ég heimsæki hann. I visit  him  ‘Jón constantly eats while I visit him.’ The same applies to cases where the whole when-clause has been preposed along with the adverb:113  (399)  Adv – when-clause – FV – Subj - V a. Oft  egar ég heimsæki hann borar Jón.  Often when I visit  him  eats  Jón  ‘Jón often eats when I visit him.’  b. *Stöugt  egar  Constantly when  ég  heimsæki hann borar Jón.  I  visit  him  eats  Jón  ‘Jón eats constantly while I visit him.’  These examples show that the syntactic position of frequency adverbs and adverbs of quantity differs.  5.4.2.3 The adverbs in the progressive When we have a sentence in the progressive, the duration of the interval represented by a when-clause influences the reading of the temporal adverb. We can divide the when-clauses into punctual and durational when-clauses. With punctual when-clauses the reference time is an instant, and the verb often is an achievement, like we have in ‘when I arrive’. With durational when-clauses the reference time is a longer interval, and the verb often is an activity or a state, as we have in ‘when I am at home’. The interesting fact is that when the when-clause is punctual, frequency adverbs are awkward, but when the when-clause is durational, adverbs of quantification are awkward.  113  Höskuldur ráinsson (p.c.) pointed out to me that frequency adverbs can be preposed in the so-called stylistic fronting construction, i.e. fronting to (apparently) subject position of impersonal or subjectless predicates. He gives an example where stöugt does indeed appear to be preposed: (i) etta er tilhneiging [sem stöugt hefur ___ fari vaxandi] This is inclination that constantly has gone growing ‘This is an inclination that has been constantly increasing.’ 174  Let us look at examples of the two progressive constructions, which behave exactly in the same way in these sentences. In (400), which has a punctual when-clause, the adverbs of quantification oft ‘often’ and alltaf ‘always’ are both fine whereas the frequency adverb stöugt is impossible. (400)  egar  ég kem heim.  Jón is often to eat (inf)/eating (pres part) when  I come home  a. Jón er oft  a bora/borandi  ‘Jón is often eating when I come home.’  b. Jón er alltaf  egar ég kem heim.  a bora/borandi  Jón is always to eat (inf)/eating (pres part) when I come home ‘Jón is always eating when I come home.’  c. *Jón er stöugt Jón  a bora/borandi  egar ég kem heim.  is constantly to eat (inf)/eating (pres part) when I come home  ‘Jón is constantly eating when I come home.’  Here alltaf is obviously an adverb of quantification and the meaning is ‘for all times t at which I come home, John is eating at t’. The frequency reading, ‘in a typical situation when I come home, John is eating constantly’, is not possible. The most likely explanation for this is that the when-clause provides just a point in time (the time of the arrival) and not an interval. For frequency adverbs to be applicable we need more than one, and probably more than two, eating events and we cannot have more than one eating event of John’s taking place at the moment I come home. When the when-clause is durative the frequency adverb stöugt is fine but now the adverb of quantification oft is problematic:  (401)  a. #Jón er oft Jón  egar ég er  a bora/borandi  heima.  is often to eat (inf)/eating (pres part) when I am home  ‘Jón is often eating when I am at home.’  b. Jón er stöugt  egar  a bora/borandi  Jón is constantly to eat (inf)/eating (pres part) when  ég er  heima.  I am home  ‘Jón is constantly eating when I am at home.’  The reading of (401b) is that while I am visiting John he is constantly eating. With an adverb of quantification, the sentence in (401a), however, claims that it happens often that while I am at home John is in the middle of eating. That is, the reference time is within the event time, just like in any imperfective  175  sentence, and since ‘I am home’ indicates a fairly long reference time it implies an even longer event time. John’s eating needs to last longer than my stay at home. As oft can also have a frequency reading one would expect it to be available here and have the meaning that ‘when I am at home John eats often’ but we do not seem to get that reading with the progressive, and instead the simple present is needed.  (402)  Jón borar oft Jón eats  egar ég er  heima.  often when I am home  ‘Jón eats often when I am at home.’  Here the frequency reading of oft becomes available, but it can also have the adverb of quantification reading. So the sentence is ambiguous. It can mean that ‘(it happens) often, (that) John eats while I am at home’, or ‘when I am at home John eats often’. This indicates that the adverb of quantification reading with the progressive and a durative when-clause is problematic as it indicates a very long event time. As soon as we give the sentence an event that can naturally be seen as lasting a long time the sentence is actually fine:  (403)  Jón er oft  a reyna a grenna  Jón is often to try  to slim  sig  egar  ég  er  self  when  I  am home  heima.  ‘Jón is often trying to lose weight while I am at home.’  An overview of the adverbs and how the interact with different when-clauses is shown in the following table. Table 5:1: Adverbs of quantity with when-clauses  Punctual when-clause  Durational when-clause  Adverbs of quantification    *  Frequency adverb  *    5.4.2.4 In the scope of for-adverbials Van Geenhoven (2004:148) points out that the following sentences with indefinite singular complements are somewhat odd:114  114  This fact was originally discussed by Carlson (1977) but the two main kinds of explanations for it can be found in Dowty (1979) and Krifka (1989, 1992). 176  (404)  a. ?Mary ate a sandwich for an hour. b. ?Mary discovered a flea on her dog for ten minutes.  Zucchi and White (2001:249) point out that this oddness disappears if we add frequency adverbs to the sentences:  (405)  a. Mary ate a sandwich every five minutes for an hour. b. Mary discovered a flea on her dog every day for a month.  Zucchi and White call this the “aspectual effect of frequency adverbs”. Van Geenhoven takes this as one of the reasons for why frequency plays a major role in the semantic interpretation of for-adverbial sentences. Van Geenhoven (2005) points out that only frequency adverbs can occur in the scope of a foradverbial, while adverbs of quantification cannot. Consider the following examples from Icelandic:  (406)  Jón var stöugt  a finna/finnandi  Jón was constantly to  flær á  hundinum sínum í  find (inf)/finding (pres part) fleas on dog.the  his  in  heilan mánu. whole month ‘Jón was constantly discovering fleas on his dog for a whole month.’  (407)  ?Jón var oft Jón  a finna/finnandi  was often to  flær  á  hundinum sínum í  find (inf)/finding (pres part) fleas on dog.the  his  for  heilan mánu. whole month ‘Jón was often discovering fleas on his dog for a whole month.’  Here the frequency adverb stöugt ‘constantly’ occurs easily with both kinds of progressive, but the adverb of quantification oft ‘often’ is awkward, further showing the difference between frequency adverbs and adverbs of quantification.  5.4.2.5 The ambiguity of alltaf As I pointed out in Jóhannsdóttir (2005, 2007a), the adverb alltaf ‘always’ in Icelandic can either occur as an adverb of quantification or as a frequency adverb. This is clear when we use the diagnostics presented in the previous sections. In fact, the same seems to apply to some other adverbs as I will show. By using the Icelandic adverb alltaf I will demonstrate how this ambiguity becomes clear when applying the tests  177  already introduced. Firstly, alltaf behaves indeed like an adverb of quantification in occurring as an unselective binder: (396)  Alltaf egar Always when  bóndi á asna, á lemur hann hann stöugt. farmer owns donkey then beats he him constantly  ‘Always, when a farmer owns a donkey, he constantly beats it.’ It can be preposed like adverbs of quantification:  (408)  Alltaf  er Jón borandi  egar ég heimsæki hann.  Always is Jón eating (pres part) when  I visit  him  ‘Jón is always eating when I visit him.’  And it can occur with punctual when-clauses in progressive sentences: egar  ég kem heim.  Jón is always eating (pres part) when  I come home  (400c) Jón er alltaf  borandi  ‘Jón is always eating when I come home.’  However, it can also occur as a frequency adverb. For instance, it can occur in the scope of for-adverbials, like frequency adverbs, but unlike most adverbs of quantification:  (409)  Jón var alltaf finnandi  flær  á hundinum sínum í  Jón was always finding (pres part) fleas on dog.the  his  heilan mánu.  for whole month  ‘Jón was constantly discovering fleas on his dog for a whole month.’  It can occur with durative when-clauses in the progressive, unlike most adverbs of quantification, but like frequency adverbs:  (410)  Jón er alltaf  borandi  egar  Jón is always eating (pres part) when  ég er  heima.  I am home  ‘Jón is eating constantly when I am at home.’  And it can, in fact, co-occur with other adverbs of quantification:  178  (411)  Oft  egar ég er  í  heimsókn hjá Jóni er hann alltaf  Often when I am in visit  at  Jón  is he  étandi.  always eating (pres part)  ‘Often when I am visiting Jón he is constantly eating.’  So it seems clear that alltaf ‘always’ can occur as either an adverb of quantification or as a frequency adverb. In fact, I suggest the same applies to always in English. In (412) we have always as an adverb of quantification, occurring as a selective binder and occurring with a punctual when-clause.  (412)  a. Always, if a farmer owns a donkey, he constantly beats it. b. John is always eating when I come home.  And in (413) always is a frequency adverb appearing in the scope of a for-adverbial, appearing with a durative when-clause and co-occurring with an adverb of quantification.  (413)  a. John was always discovering fleas on his dog for a whole month. b. John is always eating when I am at home. c. Often when I am visiting John he is always eating.  It is possible that more adverbs of quantification can occur as frequency adverbs in some contexts. In fact, ráinsson (2007) categorizes oft ‘often’ as a frequency adverb and shows oft in a postverbal position, unlike adverbs like aldrei ‘never’ and ekki ‘not’:  (414)  a. Hún hafi lesi She  had  leibeiningarnar vandlega/oft.  read (perfect) instructions.the carefully/often  ‘She had read the instructions carefully/often.’  b. *Hún hafi lesi She had  leibeiningarnar aldrei/ekki.  read (perfect) instructions.the never/not  ‘She had never/not read the instructions.  Furthermore, Lisa Matthewson (p.c.) has pointed out to me that sentences like (415) are perfectly grammatical in English and the meaning is: ‘when I am at home John eats often’:  (415)  John is often eating when I am at home.  179  In the simple present the frequency reading of oft/often is even clearer. In fact, in English, the frequency reading is the only one available when the adverb is post-verbal. When the adverb is pre-verbal the quantificational reading is the dominant one.  (416)  a John eats often when he is sad. b. John often eats when he is sad.  The Icelandic sentence is more ambiguous as the adverb can only follow the verb when the subject is in first position of the sentence. So the reading of (417) can either be that John eats often when he is sad, or that it happens often when John is sad that he eats.115  (417)  Jón  borar oft  Jón  eats  egar  often when  hann er dapur. he  is sad  ‘Jón eats often when he is sad.’  It appears, therefore, that not only alltaf ‘always’ but also oft ‘often’ can occur as a frequency adverb in addition to being an adverb of quantification, as well as the English adverbs always and often. Furthermore, as mentioned in section 5.4.2.1, footnote 114, the English adverb frequently sometimes behaves like an adverb of quantifiction and not just as a frequency adverb  5.4.2.6 Conclusion In this section I have shown that adverbs of quantification and frequency adverbs have some different characteristics that make it possible to distinguish them from each other. The temporal adverbs alltaf ‘always’ and oft ‘often’, however, show certain behaviour that belongs to frequency adverbs rather than adverbs of quantification. I conclude from this that these adverbs are in fact ambiguous.  115  Gunnar Ól. Hansson (p.c.) has pointed out to me that in Icelandic we use different focus with each reading: ‘John EATS often’ would give us the quantification reading and ‘John eats OFTEN’ the frequency reading. It would be interesting to look in more detail at the role of focus with different adverbs. See also discussion about focus in Rooth’s analysis, represented in 5.3.3.1. 180  Table 2: Quick overview of adverbs of quantity  Binding Preposing Progressive with punctual when-clause Progressive with durative when-clause Scope of for-adverbial  Adverbs of quantification    * *  Frequency adverbs * * *    alltaf/oft       This is the pattern we expect if alltaf ‘always’ and oft ‘often’ really are ambiguous. Since alltaf can either occur as an adverb of quantification or as a frequency adverb it is always good, whenever either adverbs of quantification or frequency adverbs are good. We might also want to consider whether there really are any pure adverbs of quantity or whether all these adverbs can be used as either frequency adverbs or adverbs of quantification, on at least some occasions. If we can always find examples where adverbs of quantification can be used for instance in the progressive with durative when-clauses, or in the scope of for-adverbials, or if we can find frequency adverbs that can occur with for instance punctual when-clauses, then it indicates that we do not have any pure adverbs of quantity. However, I do not think this is the case as adverbs like constantly and continously do not ever seem to behave like adverbs of quantification, as already mentioned in footnote 114, section 5.4.2.1, and adverbs like never do not seem to have a frequency reading. Even though the two kinds of adverbs of quantity differ in their nature, they both can occur with the present participle progressive, providing more detailed information about the frequency of the events in question. A detailed examination of the syntax of adverbs in both English and Icelandic might be able to explain why that is. For instance, it is not unlikely that the difference between adverbs of quantification and frequency adverbs could be derived from the syntactic positions in the structure. If adverbs of quantification are base-generated at the IP-level, for instance, but frequency adverbs are either adjuncts to or within the VP we might be able to explain this difference in quantification. Take the tripartite structure of adverbs of quantification, for instance. They have scope over both the restrictor and the matrix whereas frequency adverbs tend to only have scope over either one. ráinsson’s (2007) discussion of the syntax of adverbs supports this idea even though he does not divide the adverbs in the same way as has been done in this dissertation. I will now look at the semantics of the adverbs of quantity.  181  5.4.3  The semantics of adverbs of quantity  5.4.3.1 Adverbs of quantification As already mentioned in section 5.4.2.1, in an analysis along the lines of Lewis (1975) and Heim (1982), adverbs of quantification, such as always and often form a tripartite structure where the first one is the restrictive clause and the second one is the nuclear scope.  (418)  Adverb of quantification (,)  So if we apply (419) to (420) the quantifier will in effect bind the two variables, x and y.  (419)  “always (,)” is true iff every assignment to the free variables in  which makes  true also makes  true.  (420)  Always1,2 ((x1 is a farmer  y2 is a donkey  x1 owns y2), x1 beats y2)  The truth conditions of (420) then equal those of the formula in (421) where the unselective binder has been replaced by standard selective ones:  (421)  xy [[farmer (x)  donkey (y)  own (x,y)]  beat (x,y)]  However, we run into the same problem as we did with Krifka et al.’s (1995) analysis of GEN in chapter 4, section 4.2.5, as we need to use events in order for the adverbs to be compatible with the theory I have chosen for this dissertation. So I will be adapting Lewis’ and Heim’s analysis to work with events.116 Again, I will look towards Rothstein’s (1995) analysis of adverbial quantification over events and the matching function M, already discussed in chapter 4, section 4.2.5. As previously said, M matches one event with another and so in a sentence like ‘Mary opens the door every time the bell rings’ M maps events of door-openings onto events of bell-ringings. The universal quantifier then defines the range of M as including the entire set of events of bell-ringings (Rothstein 1995:16):  116  Heim (1990) herself later re-wrote the theory within situation semantics where she argued that adverbs of quantification quantified only over situations, an idea originally from Berman (1987). von Fintel (1996) later revised and extended Heim‘s analysis.  182  (422)  e[[RING(e)  Th(e)=THE BELL]  e’ [OPEN(e’)  Ag(e’)=Mary  Th(e’)=THE DOOR  M(e’)=e]]  So if M maps every door-opening event onto a bell-ringing event, then the adverb of quantification most of the time would map most door-ringing events onto a bell-ringing event, and adverbs like often should map many door-opening events onto a bell-ringing event. Using Partee’s (1988) work on the quantifier many we can then give the semantics of (423a) as in (423b):  (423)  a. Mary often opens the door when the bell rings. b. {e: RING(e)  Th(e)=THE BELL  e’ [OPEN(e’)  Ag(e’)=Mary  Th(e’)=THE DOOR  M(e’)=e}  p  Where p is a contextually determined proportion  {e: RING(e)  Th(e)=THE BELL}  The proportion of all bell-ringing events that are matched by events of Mary opening the door is equal to or greater than p, where p is a contextually determined proportion.  5.4.3.2 Frequency adverbs Frequency adverbs such as stöugt ‘constantly’, margsinnis ‘repeatedly’ and reglulega ‘regularly’ indicate at what frequency certain events take place. Let us start by looking back at (364) where we have stöugt ‘constantly’ in the perfective:  (364)  Jón hljóp stöugt Jón ran  frá  sjö  til  constantly from seven to  ellefu í  morgun.  eleven in morning  ‘Jón ran continuously from seven to eleven this morning.’  Stöugt indicates that there should not be any breaks between running events. And this is true for (364); if John took a break during the hours of his running the sentence would be false. We have one running event that takes place without any interruption and yet, most people would not understand (424) to mean that John never stopped running during the time that Mary was in London. It simply means that for most of his waking hours he was running.  183  (424) Jón hljóp stöugt  á mean María var  í  Jón ran constantly on while María ‘Jón ran constantly while María was in London.’  London. was  in  London  It may appear as if the adverb changes its semantics depending on the length of the interval denoted by the adverbial. If the when-clause refers to something that took a short time no breaks are allowed, but if it refers to something that took a longer time, breaks are allowed. This is not very plausible. It is more likely that pragmatics plays an important role in indicating whether there can be interruptions and if so, how often and how long those interruptions can be. Aspect interacts with the adverb in such a way that it influences the overall interpretation of the sentence. Compare the simple past sentence in (364) to the present participle progressive sentence in (365), repeated here:  (365)  Jón var  stöugt  hlaupandi  frá  sjö  til ellefu í  morgun.  Jón was  constantly running (pres part) from seven to eleven in mor ning  ‘Jón was constantly running from seven to eleven this morning.’  Unlike in (364), in (365) we get the feeling that we have more than one running event and that, in fact, Jón was constantly on the move. Therefore it appears that the truth conditions of sentences with stöugt differs depending on the aspect of the sentence. We can account for this different reading in terms of a well known pragmatic phenomenon, sometimes called the problem of quantifier domain restriction: When two people are speaking it is not always enough to rely on the permanent features of the words used to understand the meaning of the sentence. For instance if one hears someone say the sentence in (425) it is very unlikely that the person is referring to every bottle in the universe being empty; rather it is likely that she means that every one of a restricted set of bottles is empty. It could be the bottles in this room, or the bottles that she had recently bought, and so on.  (425)  Every bottle is empty.  Westerståhl (1985, 1989), von Fintel (1994), Stanley and Szabó (2000) and others, use contextual variables to treat the problem of quantifier domain restriction. A sentence like (425) could then, for instance, receive the representation in (426):117 117  Actually, Stanley and Szabó’s (2000) formulation is somewhat more complex as they use f(i), instead of R, where the i is a variable provided by the context and the value of f is a function provided by the context that maps objects onto quantifier domains. As I do not use any bound-variable examples a simpler representation is sufficient.  184  (426) x [[bottle(x)  R(x)]  empty(x)] In (426) R supplies the restriction on the quantified expression ‘every bottle’ in (425), relative to context. Instead of asserting that every bottle in the world is empty, (426) says that every bottle in a particular set, referred to by R, is empty. R is a free predicate variable that receives its value from the utterance context. I propose that in the formulation of adverbs like stöugt ‘constantly’ there is a contextual variable, similar to the ones suggested by Westerståhl (1985, 1989), von Fintel (1994), and Stanley and Szabó (2000), that restricts the adverb. The semantic representation for frequency adverbs include R as a one-place predicate over times. An interesting and a fruitful approach to frequency adverbs has been proposed by Van Geenhoven (2005:6) who defines frequency adverbs as “pluractional operators, each expressing a way of distributing sub-event times over the overall event time of an utterance”. Consider (427) as an example:  (427)  John frequently drove past my house yesterday.  The scenario we get here is that yesterday there was a time when John drove past my house, followed by a time where John did not drive past my house. That time was in turn followed by a time where John did again drive past my house. How many times John drove past my house is not specified but the choice of the adverb frequently indicates that it did at least take place fairly often. Without the frequency adverb in the sentence we have only one instance of John driving by the house:  (428)  John drove past my house yesterday.  Van Geenhoven adopts Link’s (1983) distributive star operator *, which transforms a predicate P that holds of individuals into a corresponding predicate *P that holds of sums of individuals. This meets the principle of cumulative reference: “If a predicate P holds of two plural individuals x and y, then *P holds of their sum xy as well” (Van Geenhoven 1995:6). If CAT denotes the set of all individuals who are cats, then *CAT denotes the set of all the pluralities of cats, or in other words, all the sums of cats. The star operator also holds over events so if EAT denotes the set of all atomic eating events then *EAT denotes the set of all pluralities of eating events. Van Geenhoven (2005:112) presents a semantic analysis of aspect markers, where she interprets them as pluractional operators. The markers are markers of plurality in the domain of verbs and event times. She proposes that we can think of pluractional morphemes on verbs as “star operators, each expressing a way of distributing sub-event times over the overall event time of an utterance and each meeting the principle of cumulative reference”. These operators show repetitive components in the sense  185  that between each two actions described by the verb they refer to, there is a hiatus during which that action does not take place. To illustrate with an example, West Greenlandic has a frequentative aspect marker -tar‘repeatedly’, which Van Geenhoven argues can be interpreted as a pluractional marker. It contributes three meaning components, a distributive one, a repetitive one and a component of cumulativity. Consider the following example from West Greenlandic (Van Geenhoven 2005:109).  (429)  Nuka ullaap tungaa tamaat sanioqquttarpug. Nuka  ullaa-p  tunga-a  tama-at saniuqqut-tar-pug  N.ABS morning-ERG direction-3SG.SG.ABS all-3SG  go.by-repeatedly-IND[-tr].3SG  ‘Nuka went by repeatedly the whole morning.’  Van Geenhoven suggests that the marker -tar- contributes a distributive operator, which she calls ‘crystal’ star . When this crystal star is applied to a verb there will be a hiatus between each pair of subevents expressed by the tar-ed verb. The truth conditions of -tar are shown in (430). In (431) we have applied tar to the verb saniuqqut ‘to go by’ (Van Geenhoven 2005: 113).118  (430)  -tar-  Vtx(t V (x) at t) where t V(x) at t= 1 iff t’ (t’ t  V(x) at t’  number (t’) > 1  t’ (t’  t  V(x) at t’  t’’ (t’’  t  (t’’ > t’  t’’ < t’) V(x) at t’’  t’’’ (t’ < t’’’ < t’’  t’’ < t’’’ < t’  ¬V(x) at t’’’))))  (431)  saniuqqut-tar  tx(t go by(x) at t)  When the crystal star is applied to the verb saniuqqut ‘go by’ we get the reading that there exists more than one going-by event that are a part of some time t and for each of those going-by events there exists another going-by event that is either before or after the first one, and there exists a third event that is not a going-by event and which takes place in between the two going-by events. Even though the formula may work well for West Greenlandic –tar-, it is too vague to be useable for the Icelandic frequency adverbs, as it just says that every going-by event is either preceded or followed by another going-by event. That would be true even if there were only two going-by events. It will not, therefore, suffice for the description of frequency adverbs like frequently or constantly. Van Geenhoven’s other distributive operators, the ‘flower’ star operator , which expresses high frequency, the ‘white’ star , where there is no hiatus, and the ‘black’ star operator , which like  expresses  118 t  here represents time and so start (with different stars) represents the number of repeated subevent times. 186  continuity but is silent about the absence of hiatuses, do not do better on the Icelandic data. The white star and the black star have the same problem as the crystal star as they do not actually force there to be more than two events, whereas the flower star specifies that the number of going-by events is high. The flower star, thus, comes closest, but is subject to another problem, namely that there is no relevance to context. The number of events needed for something to be frequent varies from one context to another, such as different situations, different places, different expectations, etc. Adding a contextual variable is important as it would pick out only the relevant times. For this reason, I assume that R is a contextual variable ranging over predicates over intervals that gets its value from the utterance context. By adding a contextual variable R, the truth conditions of stöugt ‘constantly’ could be written out as in (432).119 Notice that Van Geenhoven used times rather than events or situations, so I have modified her analysis somewhat, in order to be more consistent with the analysis that I use throughout this dissertation. P stands here for a predicate of events.  (432)  [[stöugt]]w,g,c = Pet[t = (e)  [t’ [t’  t  R(t’)]  e’ [t’  (e’)  P(e’)  e’e]]]  (432) says that for all relevant intervals t’ that are subintervals of t, there is an event e’ such that t’ is included in the runtime of e’, and P is true of e’, and e’ is a subinterval of e. So the semantics of a simple habitual sentence like Jón hleypur stöugt ‘John runs constantly’ can be written in the following way:  (433)  [[PRES2 Jón hleypur stöugt]]w,g,c = e[g(2) = (e)  [t’ [t’  g(2)  R(t’)]  e’ [t’  ( e’)  run (e’)  Agent (e’,Jón)  e’e ]]], where g(2) o tc  This says that for all relevant intervals t’ that are subintervals of g(2), there is an event e’ such that t’ is included in the runtime of e’, and ‘John is running’ is true of e’, and e’ is a subinterval of e’’. What R does here is to restrict the set of intervals that stöugt is quantifying over to those that are in some sense relevant in the utterance context, such that for instance bathroom breaks can be ignored. I have presented a basic idea of how these adverbs work, and in chapters 3 and 4 I presented an analysis of the infinitival progressive. It is an interesting and important question how the two interact but I do not really know yet and therefore I have to leave a full compositional analysis for further research. The formula in (434) is, however, what the end-result should be.  119  In Jóhannsdóttir (2005) I gave the truth condition for stöugt in terms of times.  187  (434)  [[PRES2 PROG Jón hleypur stöugt]]w,g,c = ee’’w’:<e’’,w’>  CON(e,w)  [g(2) = (e’’)  g(2) o tc  [t’ [t’  g(2)  R(t’)]  e’ [t’  (e’)  run (e’,w)  Agent (e’,Jón)  e’e’’ tc  (e’’)]] ]  We can read this in such a way that ‘John is constantly running’ is true in a world w relative to an assignment g and a context c iff there is an event e whose runtime includes the time of c and which in w has a continuation branch <e’,w’> where it is true that for all relevant intervals t’ that are subintervals of t, there is an event e’ such that t’ is included in the runtime of e’, and John is running is true of e’, and e’ is a subinterval of e’’. If we look back at the examples in (365) and (424) we can see that the interpretation varies, based on the context.  (435) Jón hljóp stöugt  á mean María var  í  London.  Jón ran constantly on while ‘Jón ran constantly while María was in London.’  (365)  hlaupandi  frá  María  sjö  til ellefu í  was  Jón var  stöugt  Jón was  constantly running (pres part) from seven to eleven in morning  in  London  morgun.  ‘Jón was constantly running from seven to eleven this morning.’  Whereas it is natural to assume more or less constant running in (365) it is not natural to expect constant running in (424). This is because in (424) only times that are suitable for running are included in R, excluding times when John is at work, when he is sleeping etc.  5.4.4  General conclusions  In this chapter we have discussed adverbs of quantity and frequency adverbs and seen that in certain contexts they behave quite differently. Adverbs of quantification bind indefinites, can be preposed, and can occur in the progressive with punctual when-claues, whereas frequency adverbs can occur in the progressive with durative when-clauses and in the scope of for-adverbials. This different behaviour provides good support for categorizing the adverbs into different classes. However, some of these adverbs, such as alltaf ‘always’ and oft ‘often’ in Icelandic, and always, often and frequently in English can occur as either kind, depending on context. That indicates that these adverbs are ambiguous between the two classes and further study is needed in order to show if other adverbs show the same ambiguity. Furthermore, it would be interesting to explore the syntactic position of the two kinds of adverbs as the position in the structure can influence the meaning.  188  5.5 Pluractional presupposition In section 5.2.2 we established that the present participle progressive in Icelandic always gives a pluractional reading, that is, a series of events is ongoing at the reference time. Just like with habituals none of these events has to be taking place at the reference time but what is important is that the reference time falls within some particular event time that includes this series of events. Most of the time there is an adverb of quantity in the sentence but sometimes the pluractionality is made clear by context. We could suggest that the present participle progressive contains a pluractional progressive operator that renders the sentence pluractional, but that is an unlikely explanation as it would predict that we could get a present participle progressive sentence without any adverbial and without any iterative context. As previously mentioned, the sentences from (368) above are at least strange if said out of the blue.  (368) a. #Jón er borandi. Jón  is eating (pres part)  ‘Jón is eating.’  b. #Anna er horfandi Anna  á sjónvarp.  is watch (pres part) on TV  ‘Anna is watching television.’  Unlike a sentence in the infinitival progressive that without anything indicating iterativity conveys a single, ongoing event, and a not a series of events, a sentence in the present participle progressive without an adverb of quantity or any other iterative adverbial would indicate a plural event. That does not happen as the sentences in (368) are simply bad. This indicates that the present participle progressive itself does not give a pluractional reading and therefore there is no reason to assume that its truth conditions are in any way different from the progressive operator that occurs with the infinitival progressive. And yet the present participle progressive is not possible with single events. If the progressive operator with the present participle progressive is the same or similar to the one with the infinitival progressive, could we possibly assume the same kind of semantics for the present participle progressive as we have with the progressive habitual? That would mean we have CHAR in the sentence and a shift from a state to an event before PROG applies. The benefits of such an analysis are obviously that the progressive habitual and the present participle progressive are built in the same way, with the same operators applying, which would explain why they get more or less the same meaning. The problem, however, is that the progressive habitual and the present participle progressive are not parallels; in the progressive habitual we can pair up the habitual progressive with the habitual sentence and even with a simple, episodic progressive sentence:  189  (436)  a. The Canucks play well.  (Habitual)  b. The Canucks are playing well (right now).  (Progressive)  c. The Canucks are playing well (these days).  (Progressive habitual)  With the present participle progressive there are no such parallels. There is no related habitual sentence. We cannot, therefore, say that the present participle is a habitual sentence that has been shifted to being progressive. If we do not assume that the pluractionality is built into the present participle progressive operator, nor do we assume the same kind of analysis as we have with the progressive habitual, what exactly is the present participle progressive? I will argue that it is a progressive with a presupposition of pluractionality. The nature of this pluractionality generally needs to be spelled out further which is the reason why we generally get adverbs of quantity or other pluractional adverbials present in the sentence. What this means is that if there is an iterative adverb in the sentence, it will serve to satisfy the pluractionality presupposition of the present participle progressive. If there is no adverb, but context gives sufficient clues that the event is iterated, then the presupposition is satisfied by context. Let me explain this further. By using the present participle progressive in (363) the speaker makes an implicit assumption that there are multiple events of eating and not just the one eating.  (363) a. Jón er sífellt  borandi.  (Present Participle)  Jón is constantly eating (pres part) ‘Jón is constantly eating.’  The adverb sífellt ‘constantly’ reiterates the pluractionality saying that there were so many eating events that they were near continuous. No hearer would understand the sentence to mean that there was one continuous eating event, even though the Icelandic adverb sífellt can mean continuously as well as constantly. This is even clearer in (364) where we have the reference time from seven to eleven in addition to the adverb stöugt:  (364) Jón var stöugt  hlaupandi  frá  sjö  til ellefu í  morgun.  Jón was constantly/continuously running (pres part) from seven to eleven in morning ‘Jón was constantly running from seven to eleven this morning.’  Notice that the reference time is not an unreasonably long time for a long-distance runner so the length of the reference time does not make the events pluractional. The adverb stöugt can also mean ‘continously’ 190  as well as ‘constantly’ and so the pluractionality cannot come from that either. And yet all the Icelandic speakers that were asked about the meaning of the sentence claimed that it indicated a number of runnings, and not just one long running event.  The use of the present participle presupposes the  pluractionality so by hearing the present participle progressive used the hearer automatically makes the assumption that we have multiple events of running. A presuppositional analysis predicts infelicity if the presupposition is not met, so if there is neither an adverbial nor a contextual satisfaction, accommodation is required (see e.g. Lewis (1979) and Heim (1983)). It is not clear to me why the accommodation is so difficult in these sentences as they tend to be unacceptable without the adverbial or contextual satisfaction, but as seen in both (363) and (364), even an ambiguous adverbial is enough. However, different kinds of presuppositions differ considerably in how easy they are to accommodate, and this variation is at present not very well understood. This analysis solves some problems that we have with the two alternatives suggested above. It does not predict that we have a parallel non-progressive habitual sentence with the present participle progressive as we would if the present participle progressive was formed in the same way as the progressive habitual. Furthermore, unlike what we would expect if the pluractionality was a part of the truth conditions of the present participle progressive this analysis does not predict a sentence like Jón er borandi ‘Jón is eating’ to be automatically grammatical. It would need accommodation and there could be any number of reasons for why that accommodation is so hard to get. More importantly, we do not run into the same problem of compositionality as we get if the pluractionality was a part of the truth conditions of the present participle progressive. If both the frequency adverb and the present participle progressive by itself had pluractionality in their truth conditions, we would have two pluractional operators in the sentence and would end up with something like a multiplicity of a multiplicity of events. By assuming that the pluractionality of the present participle progressive is a mere presupposition and not a part of the truth conditions, the presupposition will be satisfied by the pluractionality built into the meaning of the adverb, or accommodated because of the context. So even though I do not provide a compositional analysis for now, such an analysis should, in principle, be possible. What such an analysis looks like will be a topic for future research.  5.6 Conclusion In this chapter I have discussed the present participle progressive in Icelandic and argued that it is a pluractional progressive construction that has the meaning that a series of events is ongoing instead of just a single event as the regular infinitival progressive does. The present participle progressive is in that way habitual just like the progressive habitual discussed in chapter 4. However, the pluractionality of the construction is not a part of the truth conditions of the present participle progressive but a mere presupposition that gets satisfied by adverbs of quantity that further pick out the frequency of the events 191  in question. When there are no adverbs of quantity in the sentence the presupposition is usually satisfied by context.  192  6 Closing Remarks 6.1 Results A simple description of the progressive might characterize it as expressing an incomplete action in progress at a specific time. It is a non-habitual, non-stative, imperfective aspect. However, as has been previously pointed out, and as I clearly show in this dissertation, stative verbs sometimes can appear in the progressive (e.g. Hirtle (1967, 2007), Mufwene (1984), Brinton (1988), Hirtle and Bégin (1991), Kakietek (1997), mieciska (2002/2003), ráinsson (2005), Torfadóttir (2004), Kranich (2010)) and the progressive sometimes does allow a habitual interpretation (e.g. Hirtle (1967:49ff), Leech (1971:33), Palmer (1988), Kranich (2010)). This might indicate that the previous descriptions of the progressive, such as those of Poutsma (1926), Lakoff (1966, 1970), Einarsson (1967), Visser (1973) Comrie (1976) and many others, have simply not been accurate. Corpus-based studies, such as that of Kranich (2010) for English, have shown that the use of the progressive has increased but that the use of stative verbs in the progressive has only increased slightly. The Icelandic ScanDiaSyn project (ráinsson (2010)) has shown that younger speakers tend to use stative verbs in the progressive more than older speakers. We might have an ongoing change where states and habituals are increasingly used in the progressive but that is not completely a recent change and, therefore, any reasonable description of the progressive should take states and habituals into account. Most theories of the progressive, such as those of Bennett and Partee (1972), Dowty (1979), Parsons (1990), Landman (1992) and Portner (1998) focus on the imperfective paradox and attempt to explain how an event can be said to be in progress if it does not get completed. The completion is only relevant to accomplishment verbs and does not affect either states or habituals in the progressive.120 For our purpose, therefore, all that matters is the description of an event being in progress, and how that affects states and habituals. We also need the theory to allow for an explanation of why when we have statives or habituals in the progressive, they get an implicature of temporariness, dynamicity and/or control. While I chose to use Landman’s analysis in this dissertation, Portner's (1998) theory of the progressive could also be adapted to allow for this. What I have argued is that when we have stative verbs in the progressive, they have been coerced to being eventive. Control, dynamicity and temporariness are prototypical eventive properties that are in general not available to stative verbs. However, if a speaker wants to convey any of these properties with a stative verb he can coerce the verb to being eventive and by doing so implicate one or more of these  120  Technically it also affects achievement verbs but as Rothstein (2004) has argued, achievement verbs in the progressive have really been coerced to being accomplishments. 193  prototypical eventive properties. The former stative verb now has to occur in the progressive in order to get a non-habitual reading – just like event verbs. In order to establish what kind of a shift rule is at work when states are coerced to being eventive, I evaluated the arguments for stative verbs having a Davidsonian argument (Landman 2000 and Mittwoch 2005), and against that hypothesis (Katz 1997, 2000, 2003 and Maienborn 2003, 2004, 2007). I found that the arguments are stronger in favour of there being a Davidsonian argument with stative verbs, but it is nevertheless clear that it must differ considerably from the event argument, which partly explains why it is so much harder to argue for the existence of such an argument with stative verbs. By arguing for the existence of a state argument our shift rule can be fairly simple, as it requires the minimal change from a state argument to an event argument. If stative verbs did not have a Davidsonian argument a more elaborate shift would be required. Ultimately, what we have when we hear stative verbs in the progressive is a former-stative coerced to being eventive and therefore behaving like any eventive verb would, which includes having one or more of the prototypical eventive properties implicated. However, as I have argued, not all eventive verbs have all the eventive properties and so it is generally up to the hearer to figure out which of the prototypical eventive properties the speaker intends to convey when shifting the state to an event. The implicature is often made clear by context but sometimes the hearer simply has to use his intuition. An implicature is also present with habitual sentences in the progressive. A general habit is expected to hold indefinitely unless otherwise specified. When the habitual sentence is in the progressive, however, the hearer understands it so that the speaker is implying one or more of the prototypical eventive properties, usually temporariness. This gives the reading that the time referred to by the progressive sentence is a subinterval of some other interval where the habit does not hold. The hearer can infer that the speaker had a reason not to use the simple present. So this is a mere implicature as nothing in the progressive or the habitual says that the habit is necessarily temporary. The habitual operator CHAR applies to the sentence, rendering it habitual so we no longer have one particular event in progress, but a series of events. In order for the PROG operator to apply, the shift rule that shifts states to events also applies to the habitual, shifting it to an event and then PROG can apply. Just like with states in the progressive, when we apply the progressive to a habitual construction we get the implicature that the habit is temporary. In the habitual progressive we have GEN and PROG applying in the opposite order from CHAR and PROG in the progressive habitual, as GEN applies to a sentence with PROG in the restrictor. In Icelandic there is an alternative way to convey that a series of events is in progress – by using the present participle progressive. The construction presupposes pluractional events and therefore never has the reading that a single event is in progress In this way it differs from the infinitival progressive. Usually there is an iterative adverbial present in these sentences, indicating at what frequency the events are taking place. Most commonly this is an adverb of quantification or a frequency adverb. Some adverbs, 194  such as alltaf ‘always’ and oft ‘often’ play a dual role as they can either occur as adverbs of quantification or frequency adverbs. One of the main differences is that adverbs of quantification appear to be sentential and take scope over both the restrictor and the matrix, whereas frequency adverbs apply to the VP only. A further study would be required in order to give precise composition of adverbs of quantity, relating to their different possible syntactic positions.  6.2 Suggestions for further study Even though this dissertation answers many questions there are others yet unanswered that will make for interesting further studies. For instance, it will be interesting to look in more detail at the third kind of progressive in Icelandic, the one that uses a posture verb in conjunction with the main verb as in Jón situr og les ‘John sits and reads’. This is a commonly used progressive construction in the other Scandinavian languages but is not much used in Icelandic, even though it does exist. I would like to see how systematically it is used as a progressive construction and where the progressive meaning comes from. I argued in Jóhannsdóttir (2007b) that it was the stativity of the posture verb that gave the progressive reading of the event and it would be interesting to explore that idea further. In order to give a more exhaustive description of the progressive in Icelandic it is also necessary to continue with ráinsson’s (1974) analysis of the infinitival form vera a + V, which forms the infinitival progressive. For instance, are there cases of the vera a + V construction where we do not have a progressive reading? ráinsson (2001) believes this is the case and names examples like: (437) Hann var ekkert a hanga yfir ví. He was not  to hang over it  ‘He did not bother spending time on it.’  He argues that such sentences are not progressive sentences even though they use the vera a + V construction. Just as with the English present participle, the Icelandic present participle is used in nonprogressive contexts and a general analysis of the constructions would make it easier to give us a full picture of the progressive in Icelandic. For instance, is Jón er sofandi ‘Jón is asleep’ more like Jón er alltaf borandi ‘Jón is always eating’ or like Jón er reyttur ‘Jón is tired’? Formally it is obviously more like the former but categorically perhaps more like the latter. In general, a thorough examination into the present participle in Icelandic is needed – particularly since, to my knowledge, no one has so far studied the present participle in detail. As very little research has been done on the syntax and semantics of adverbs in Icelandic, further study in that area would help us not only explain the duality of adverbs like alltaf ‘always’, but also 195  understand in more detail the difference between, say, adverbs of quantification and frequency adverbs. The more we know about these adverbs the easier it will be to look into the interaction of adverbs and aspect. As Icelandic and English are the only two Germanic languages that have fully grammaticalized progressive constructions the development of the progressive in these languages is interesting in the broader spectrum. For instance, it is not impossible to imagine that when the progressive has become fully grammaticalized in other Germanic languages we will start seeing similar cases of stative verbs and habituals in the progressive.  196  Bibliography Abusch, Dorit. 1988. Sequence of tense, intensionality, and scope. In The Proceedings of WCCFL 7. CSLI, Stanford, 1–14. Albright, Adam. 2009. Lexical and morphological conditioning of paradigm gaps. In Curt Rice and Sylvia Blaho (eds): Modeling ungrammaticality in Optimality Theory. Equinox, London. Arregui, Ana and Lisa Matthewson. 2001. A cross-linguistic perspective on the expression of manner. In Rachel Hastings, Brendan Jackson and Zsofia Zvolensky (eds.): SALT XI. Cornell University, Ithaca, NY, 1-19. Austin, Peter. 1998. Crow is sitting chasing them: Grammaticalisation and the verb “to sit” in the Mantharta languages, Western Australia. In Anna Siewierska and Jae Jung Song (eds.): Case, typology and grammar: in honour of Barry J. Blake. John Benjamins, Amsterdam, 19-36. Bach, Emmon. 1986. The algebra of events. Linguistics and Philosophy 9: 5-16. Bache, Carl. 1997. The study of aspect, tense and action. Towards a theory of the semantics of grammatical categories. 2nd revised edition. Lang, Frankfurt/Main. Barwise, Jon. 1981. Scenes and other situations. Journal of Philosophy 78.7: 369-397 Bauer, Georg. 1970. The English “perfect” reconsidered. Journal of Linguistics 6: 189–198. Benediktsson, Hreinn. 1976. Ísl. vera a + nafnh.: Aldur og uppruni. In Lars Svensson, Anne Marie Wieselgren and Åke Hansson (eds.): Nordiska studier i filologi och lingvistik. Festskrift tillägnad Gösta Holm på 60 årsdagen 8 juli 1976. Cal Bloms Boktryckeri, Lund, 25-47. Bennett, Michael and Barbara Partee. 1972. Toward the logic of tense and aspect in English. University Linguistics Club, Bloomington, Indiana. Berman, Steve. 1987. Situation-based semantics for adverbs of quantification. In Juliette Blevins and Anne Vainikka (eds.): University of Massachusetts Occasional Papers 12:45-68 Bertinetto, Pier Marco, Karen Ebert and Casper de Groot. 2000. The progressive in Europe. In Östen Dahl (ed.): Tense and aspect in the languages of Europe. (Eurotyp: Typology of languages in Europe, 20-6.) Mouton de Gruyter, Berlin, 517-558. Binnick, Robert. 1991. Time and the verb. Oxford University Press, Oxford. Boertien, Harmon S. 1979. The double modal construction in Texas. Texas Linguistic Forum 13: 14-33. Brinton, Laurel J. 1987. The aspectual nature of states and habits. Folia Linguistica XXI: 195-214. Brinton, Laurel J. 1988. The development of English aspectual systems. Cambridge University Press, Cambridge. Buyssens, Eric. 1968. Les deux aspectifs de la conjugaison anglaise au XXe siècle. Presses universitaires de Bruxelles, Bruxelles. Bybee, Joan and Östen Dahl. 1989. The creation of tense and aspect systems in the language of the world, Studies in language 13:51-103. Carlson, Gregory N. 1977. Reference to kinds in English. Doctoral dissertation, University of Massachusetts, Amherst. Carlson, Gregory N. 2005. Generics, habituals and iteratives. In Keith Brown (ed.): The encyclopedia of language and linguistics, 2nd Ed. Elsevier. http://www.ling.rochester.edu/people/carlson/download/generics/habituals.pdf Carlson, Lauri. 1981. Aspect and quantification. In Philip J. Tedeschi and Annie Zaenen (eds.): Tense and aspect (Syntax and Semantics 14). Academic Press, New York, 31-64. Chomsky, Noam. 1982. Some concept and consequences of the theory of Government and binding. MIT Press, Cambridge, MA.  197  Chung, Sandra, and Alan Timberlake. 1985. Tense, aspect, and mood. In Timothy Shopen (ed.): Language typology and syntactic description: Grammatical categories and the lexicon. Vol. 3. Cambridge University Press, Cambridge, 202–258. Comrie, Bernard. 1976. Aspect. Cambridge textbooks in linguistics 2. Cambridge University Press, Cambridge. Comrie, Bernard. 1995. Tense and aspect. In Joachim Jacobs, Arnim von Stechow, Wolfgang Sternefeld and Theo Vennemann (eds.): Syntax. Ein internationales Handbuch zeitgenössischer Forschung. An international handbook of contemporary research.De Gruyter, Berlin, 1244-51. Copley, Bridget. 2004. Aspect and scope in future conditionals. Ms, University of Southern California. Copley, Bridget. 2005. Ordering and reasoning. In Jon Gajewski, Valentine Hacquard, Bernhard Nickel and Seth Yalcin (eds.): New work on modality. MIT working papers in linguistics 51. Csirmaz, Aniko. 2009. Adverbs of quantity: Entities of different kinds. University of Pennsylvania Working Papers in Linguistics 15 (1): 65-74. Dahl, Östen. 1985. Tense and aspect systems. Blackwell, Oxford. Davidson, Donald. 1967. The logical form of action sentences. In Nicholas Rescher (ed.): The logic of decision and action. University of Pittsburgh Press, Pittsburgh. Davies,