Blackfoot prominence: insights from morpho-phonology Natalie Weber and Blake H. Allen April 15, 2012 1 Introduction Blackfoot is a Plains Algonquian language spoken by about 4,315 speakers (Russell and Genee 2006) in southern Alberta and northwestern Montana. It has a system of syllabic prominence primarily correlated with an increase in F0 (Van Der Mark 2003), as well as increased duration and amplitude. This prominence has been referred to in the literature as stress (Taylor 1969), pitch accent (Frantz 1997; Frantz and Russell 1995; Kaneko 1999), or tone (Stacy 2004). Although previous studies have noted various char- acteristics of Blackfoot prominence (Kaneko 1999; Stacy 2004), none have resulted in a complete analysis of word-level prominence. This study is a first step towards such a motivated and predictive analysis. We investigate how paradigmatic forms of Blackfoot nominals affect the pattern of prominence in a phonological word and make inferences about the morpho-phonology of Blackfoot. Motivation for this study came from several observations. Though rare, there are contrastive pairs of words that differ solely in the placement 1 of prominence. Some examples of contrastive pairs are noted in Frantz (1997:3). However, these pairs always differ in morphological composition, and we suspect that these underlying morphological differences lead to dif- ferent surface patterns of prominence. Morph Surface Gloss aohkii aohḱı́ı ‘water’ á-ohkii áóhkii ‘he is barking’ Furthermore, there are numerous examples where accent of a phonolog- ical word does not seem to be a fixed property of the nominal stem, but changes depending on the prefixes it combines with. In the example below, accent falls on the first, second, or third syllable of the stem imitaa ‘dog’1, or not at all, suggesting the accent cannot be a lexical property of the stem itself. Prefix Stem Gloss imitáá ‘dog’ kaak- oḿıtaa ‘just a dog’ pok- ómitaa ‘small dog’ ómahk- omitaa ‘big dog’ 2 Previous investigations Previous grammars and descriptive studies of Blackfoot referred to the prosodic prominence as a ‘stress’ or ‘accent’ system (Uhlenbeck and Cornelius 1978; Taylor 1969; Frantz 1971). Later investigations of Blackfoot describe 1See section 4.1 for an explanation of the vowel change i to o in the stem. 2 it as a ‘pitch accent’ system and assume a simple distinction of accented ver- sus unaccented syllables; the accented syllables are usually marked in the or- thography by an acute accent (for example, Frantz 1971, 1997; Frantz and Russell 1995). In recent years there have been three masters theses (Kaneko 1999; Van Der Mark 2003; Stacy 2004) which deal specifically with aspects of Blackfoot prominence. These are discussed in more detail below. Kaneko (1999) was the first to look at Blackfoot pitch accent in depth. She noted that the pitch accent of nominal roots tended to be attracted to heavy syllables and relied on a metrical stress analysis to accomodate Black- foot’s quantity-sensitive system. She found that the surface pitch accent of nominal compounds (multimorphemic nominals) did not depend on princi- ples of metrical stress, but is predictable based on whether the components are bound or free, and whether or not they have a lexically-specified accent. Van Der Mark (2003) and Stacy (2004) both focused on categorizing the language as a ‘pitch accent’, ‘tonal’, or ‘stress’ system. Van Der Mark (2003) investigated the acoustic correlates of Blackfoot pitch accent, and found that it was strongly correlated with higher pitch, as well as length and duration to a lesser degree. She found that vocal fold tension, which occurs in stress languages such as English and Dutch, was not a variable associated with Blackfoot pitch accent, leading Van Der Mark to conclude that Blackfoot may be categorized as a pitch accent, but not as a stress language. Stacy (2004) addressed individual claims made by Kaneko (1999) and showed examples of how each generalization is violated in Blackfoot, as well as how Blackfoot violates most features normally associated with metrical stress systems. She argued that since Blackfoot does not exhibit rhythmic or accentual metrical properties, it cannot be a stress or pitch accent language. 3 She concluded it must be tonal in nature and offers a few observations of tonal behavior in Blackfoot, including alternation of coda glottal stops with a falling tone, and accent spread. 3 Methodology Over the course of numerous elicitation sessions, we prompted our informant— a native speaker of Kainai Blackfoot from Alberta—with English bare nouns and prefix-noun phrases2. In order to control for any free variation or speaker memory errors, most tokens were requested during at least two different elicitation sessions. The following prefixes were used to form the nominal compounds we elicited: omahk- ‘big’ kaak- ‘only, just’ sik- ‘black’ ksikk- ‘white’ pok- ‘small, young’ inno- ‘long’ naapi- ‘white’ apo- ‘white’ saahk- ‘young, short’ 4 Orthography All Blackfoot forms in this paper follow Frantz’s orthographic conventions (Frantz 1978, 1997), laid out in Tables 1 and 2. Blackfoot contrasts both long and short vowels and long and short consonants. The segment /s/ may be syllabic (Derrick 2007; Denzer-King 2009), and short vowels before a velar fricative (ah, ih, oh) are devoiced. Blackfoot phonotactics are highly restricted. In general, the only non-syllabic segments 2In future elicitations we will be evaluating the usefulness of image pairs as prompts. 4 Labial Coronal Dorsal Glottal Stops p pp t tt k kk ’ /P/ Fricatives s ss h /x/ Affricates ts ks Nasals m mm n nn Glides w y /j/ Table 1: Blackfoot orthography which can occur as the first consonant in a cluster are a glottal stop (<’>) or velar fricative (). Front Back High i i: Mid o o: Low a a: Table 2: Blackfoot vowel system 4.1 Nominals Many noun stems exhibit morphologically-conditioned allomorphy. One al- lomorph (which we refer to as α) always occurs word-initially and frequently also compound-internally. The other (β) is only found compound-internally, and usually differs from the first in partially predictable ways: while a subset of stems allomorphs are suppletive, most reflect phonotactically-motivated segmental changes. These β forms are typically distinguished by vowel epenthesis or consonant loss, but may also reflect an initial vowel change, or historical gemination (Thomson 1978). Table 3 illustrates the stems found in this paper. We assume that β forms have the same lexical accent as α 5 α β Gloss Note aakii -aakii ‘woman’ No change aṕı’si -aṕı’si ‘coyote’ poos -ohpoos ‘cat’ Epenthetic oh. pokon -ohpokon ‘ball’ kiááyo -ohkiááyo ‘bear’ ṕı́ıtaa -iṕı́ıtaa ‘eagle’ Epenthetic i. pookaa -ipokaa ‘child’ akkssin -okkssin ‘bed’ Initial vowel change. imitaa -omitaa ‘dog’ apáni -opáni ‘butterfly’ ńınaa -́ınaa ‘man’ Initial nasal loss. Table 3: Forms of elicited Blackfoot nominal stems forms. Our analysis of the epenthetic oh and i as belonging to the stem allomorph— rather than to the prefix allomorph or as material inserted late in the phonological derivation—relies on the observation that a given stem will always select for the same epenthetic vowel regardless of the choice of pre- fix. Although our consultant will occasionally offer forms with the atypical epenthetic vowel, prompting her with the form including the typical vowel invariably causes her to remark that she prefers the form with the typical vowel. Of the atypical vowels that our consultant offers, we have not noticed a tendency to supply either oh or i more often than the other. However, in some cases she seems to have reanalyzed the epenthetic oh as part of the prefix, so that [Prefix + oh] is used as an allomorph of [Prefix]. This forms 6 a separate phonological word from the following stem, as evidenced by the fact that she forms consonant clusters that are otherwise not found word internally, such as in sikoh’apáni ‘small butterfly’. 5 Findings Four generalization arose from comparisons of segmental and accentual dif- ferences in the nominal paradigms we elicited. These four generalizations capture the majority of compound accents patterns found. As described be- low, two of those generalizations pertain to distinct lexical subclasses, while the other two describe segmentally-conditioned accent shift. 5.1 Two stem classes A novel generalization to come from our examination of Blackfoot com- pounds relates to the status of lexicalized accent among nominal stems. We analyze each nominal stem as belonging to one of the following two classes: (1) Class 1: Bears no lexical accent. Class 2: One syllable is associated with an accent. All Blackfoot nominal stems uttered in isolation have a local pitch max- imum on one of the syllables. For some nominal stems, this local pitch max- imum is correlated with a phonological lexical accent. However, a subset of stems whose final syllable bears a local pitch maximum does not pattern like other stems in various nominal compounds. Prefixes like pok - ‘small’ act as a noun class litmus test: accented nouns (Class 2) keep their lexical accent when pok - is prefixed to them, while the accent of other nouns (Class 1) is determined by the prefix. Specifically, accent will fall on the first or second 7 syllable of the compound, in accordance with generalizations to be discussed later. Accented Root ‘small’ + Root Gloss iṕı́ıtaa poks-iṕı́ıtaa ‘eagle’ apáni pok-apáni ‘butterfly’ Unaccented Root ‘small’ + Root Gloss aakii pok-ááki ‘woman’ omitaa pok-ómitaa ‘dog’ ohpokon pók-ohpokon ‘ball’ This discrepancy between words like aakii and those whose accent lo- cation does not change upon prefixation leads us to believe that members the noun class described above differ from other nouns in lacking lexically- specified accent. Two other facts encountered in our elicitations support this claim. First, the local pitch maximum of unaccented stems is realized as a sharp pitch fall pre-pausally, when the word is uttered in isolation or at the end of a syntactic phrase. Three examples of such nouns are given below with diacritics to indicate their pitch contours: (2) aakii → aaḱıi ‘woman’ pokon → pokón ‘ball’ imitaa → imitáa ‘dog’ Phrase-medially, these stems exhibit only a gradual pitch rise and no fall, as would be expected of unaccented lexical items. Moreover, the Class 1 8 stems ending in a long vowel are the only cases in the language of a pitch fall across a long syllable; elsewhere, the entirety of the long syllable is pronounced with a steady or rising high pitch. Given these data, we conclude that this word-final pitch fall is due to a phrasal-level accent rather than being associated with any particular morphology. Additionally, although these Class 1 stems bear no lexical accent, there does exist at least one final-accented stem in Class 2, owá́ı ‘egg’. Although the addition of a prefix generally does not affect the location of accent on Class 2 nominal stems, we are still faced with the task of describ- ing the distribution of accent locations among Class 1 stem compounds. At this point the number of paradigms we have collected is still too small for a conclusive analysis, and in some cases multiple theoretically distinct hy- potheses make the same predictions about accent placement, but the three claims will make below do together describe a significant majority of the observed forms. 5.2 Three prefix classes Blackfoot nominal prefixes divide naturally into three distinct classes. Specif- ically, positing prefix classes based on which syllable of a compound they target for accent insertion has allowed us to account for nearly all the forms we have encountered. While two categories of predictable and phonetically- grounded exceptions will be discussed in the subsections below, we use the following conventions to describe these classes: (3) Class 1: Accents the compound’s first syllable. Class 2: Accents the compound’s second syllable. Class 3: Accents the compound’s third syllable. 9 Among the prefixes we elicited, the ones that pattern as Class 1 types are omahk - ‘big’, ap- ‘white’, and naapi - ‘white’. Example compounds are given below. Class 1 Prefix Prefix + Root Gloss omahk- ómahk-omitaa ‘big dog’ ap- áp-inaakii ‘white woman’ naapi- náápi-aakii ‘white woman’ There is no reason related solely to these data to describe the Class 1 prefixes as adding an accent to the first syllable of the compound, rather than simply bearing initial lexical accent, e.g. as ómahk -, áp-, and náápi -. Indeed, it may emerge during future research that this analysis is preferable. For the sake of having symmetrical classes, however, we are assuming that the mechanism by which these prefixes add accent parallels those used by the other two classes, which as described below are more accurately described by reference to syllable count within a compound. Class 2 prefixes add accent to the second syllable of compounds formed from unaccented stems: Class 2 Prefix Prefix + Root Gloss pok- pok-ómitaa ‘small dog’ inno- innó-ómitaa ‘long dog’ also pok- pok-áákii ‘small woman’ It happens to be the case that all Class 2 prefixes we found are either monosyllabic (e.g. pok -) or V-final disyllabic (e.g. inno-), and so since word- 10 internal β stems are always V-initial, we are unable to state conclusively that Class 2 prefixes add accent to the second syllable of the compound rather than the first syllable of the stem. For example, in cases like inno- + ohpoos, the final o of the prefix and the initial o of the stem fuse together into a long o, which then expresses the accent as a continual high pitch across the entire V: span. Based on our small set of morphemes, Class 3 prefixes appear to be fewer in number than the first two classes: we found only two, kaak ‘only/just’ and ksikk ‘white’. Class 3 Prefix Prefix + Root Gloss kaak- kaak-omı́taa ‘just a dog’ ksikk- ksikk-omı́taa ‘white dog’ We have analyzed these prefixes as causing accent to fall on the third syllable of the compound, but another valid analysis of these data would be that it accents the second syllable of the stem. So far we have been unable to identify phonological or semantics patterns that can predict which class a given prefix will belong to. 5.3 Syllable weight We have also found that in some cases, unaccented stems with Class 3 prefixes have an accent on the peninitial syllable rather than the expected third-syllable accent. In all these compounds, however, this second syllable is phonologically heavy—either (C)VV or (C)VC. 11 Root ‘just’ + Root Gloss aakii kaa.káá.kii ‘just a woman’ i’towaaki kaa.ḱı’.to.waa.ki ‘just a chicken’ Based on these forms, we propose a principle of weight-based accent attraction. It is remarkable that only the accents derived from Class 3 prefixes appear to be affected by syllabic weight. We speculate that this discrepancy could be due to the fact that the accent created by Class 3 stems is segmentally and syllabically farther from its source (the prefix)— perhaps this distance causes instability in the accent location, rendering it susceptible to influence from factors like syllable weight. 5.4 Voiceless syllables Blackfoot has voiceless syllabic segments, including as well as devoiced vowels before the velar fricative . Since Blackfoot prominence is sig- nalled primarily by a higher F0, accent cannot fall on a voiceless syllable. Stacy (2004) found that accent which is predicted to fall on a voiceless syl- lable is realized one syllable to the left, and our study confirmed this. The prefix pok- normally accents the second syllable of a word, exempli- fied in the table below. Root ‘small’ + Root Gloss issk pokśıssk ‘small bucket’ aakii pokáákii ‘small woman’ omitaa pokómitaa ‘small dog’ i’towaaki poḱı’towaaki ‘small chicken’ However, when the second syllable is voiceless, accent shifts left to fall on the first syllable of the word (in this case, on the prefix itself). 12 Root ‘small’ + Root Gloss ohpoos pó.koh.poos ‘small cat’ ohpokon pó.koh.po.kon ‘small ball’ ohmokoyi pó.koh.mo.ko.yi ‘small wolf’ 6 Conclusion We have presented evidence that Blackfoot nominal stems may be divided into two classes (accented and unaccented) which interact with prefixes in different ways. Prefixation does not affect the accent of an accented stem. In contrast, the choice of prefix determines the pattern of prominence of a nomi- nal compound when combining with an unaccented stem. We also noted two phonological factors that influence accent placement: voicing and syllable weight. This analysis offers new insights into Blackfoot morpho-phonology and is an important contribution to an understanding of Blackfoot promi- nence. 7 Future research A full analysis of Blackfoot prominence would account not only for the word- level prominence of nominals which we investigated here, but would examine the prominence of those words in phrasal contexts. From preliminary ob- servations, it seems that not every word-level prominence is also prominent during natural speech. Which accents remain in all contexts and which can be influenced by the phrasal prosody is the subject of future research. We are also interested in exploring a footed analysis of Blackfoot. In such an analysis, footing in Blackfoot would be an organizing principle and would interact with the prominence in particular ways. Some evidence of 13 a possible footed structure influencing the placement of prominence comes from the Class 3 prefixes discussed above. Prominence falls on the third syllable of the word if the second is light, but on the second syllable if it is heavy. Additionally, both Class 3 prefixes are heavy syllables themselves. This suggests that Blackfoot words could be organized into heavy and light syllables with iambic feet, and that the prominence falls on the head of the foot as follows: Morph Prosody Gloss kaak-omitaa (kaa).(ko.mı́).(taa) ‘just a dog’ kaak-aakii (kaa).(káá).(kii) ‘just a woman’ kaak-i’towaaki (kaa).(ḱı’).(to.waa).ki ‘just a chicken’ Though unexplored, some morphemes in our transcriptions of story- boards done in class do show vowel length alternation, and it would be interesting to see if these also follow a rhythmic alternating pattern that would correspond to feet. Aside from these explorations, we would also like to explore the promi- nence patterns of Blackfoot verbs. We take as our null hypothesis that prominence in verbs will follow the same sorts of patterns found in nomi- nals. However, many languages show different patterns for verbs and nouns, and it would be interesting to see if something similar occurs in Blackfoot. Bibliography Denzer-King, R. 2009. The distribution of /s/ in blackfoot: An optimality theory account. Master’s thesis, University of Montana. 14 Derrick, D. 2007. Syllabification and Blackfoot /s/. In Proceedings of NWLC 22, SFU Working Papers in Linguistics, 62–76. SFU. Frantz, D. 1971. Toward a generative grammar of blackfoot:(with particu- lar attention to selected stem formation processes). Summer Institute of Linguistics, University of Oklahoma. Frantz, Donald G. 1978. Abstractness of phonology and Blackfoot orthog- raphy design. In Approaches to language, anthropological issues: Papers written for the IXth International Congress of Anthropological and Eth- nological Sciences, Chicago, 1973 , ed. W. McCormack and S.A. Wurm, 307–325. Mouton Publishers. Frantz, Donald G. 1997. Blackfoot grammar . University of Toronto Press. Frantz, Donald G., and N.J. Russell. 1995. Blackfoot dictionary of stems, roots, and affixes. University of Toronto Press. Kaneko, I. 1999. A metrical analysis of Blackfoot nominal accent in opti- mality theory. Linguistics, British Columbia . Van Der Mark, S.C. 2003. The phonetics of Blackfoot pitch accent. Master’s thesis, University of Calgary. Russell, Lena Heavy Shields, and Inge Genee. 2006. The Blackfoot lan- guage: current position and future prospect. Talk at the 38th Algonquian conference, University of British Columbia. Stacy, E. 2004. Phonological aspects of Blackfoot prominence. Doctoral Dissertation, University of Calgary. 15 Taylor, A.R. 1969. A grammar of Blackfoot. Doctoral Dissertation, Univer- sity of California, Berkeley. Thomson, Gregory E. 1978. The origin of Blackfoot geminate stops and nasals. In Linguistic studies of Native Canada, ed. Eung-Do Cook and Jonathan Kaye. University of British Columbia. Uhlenbeck, C.C., and C. Cornelius. 1978. A concise Blackfoot grammar based on material from the southern Peigans, volume 41. AMS Press. 16 Blackfoot Pitch Accent: Insights from Morpho-Phonology Natalie Weber and Blake Allen Department of Linguistics, University of British Columbia Prominence in Blackfoot Blackfoot is a Plains Algonquian language spoken by about 4,315 speakers [5] in southern Alberta and northern Montana. It has a system of syllabic promi- nence primarily correlated with an increase in F0 [7], as well as increased duration and amplitude. This prominence has been referred to in the literature as stress [8], pitch accent [3, 2, 4], or tone [6]. Blackfoot Nominals Many Blackfoot noun stems have two allomorphs. One usually occurs word-initially, while the other reflects phonotactically-motivated vowel epenthesis, initial change, or a length alternation. Some roots are suppletive. All Blackfoot formsfollow Frantz’s orthographic conventions [1, 3]. α β Gloss kiááyo -ohkiááyo ‘bear’ píítaa -ipíítaa ‘eagle’ akkssin -okkssin ‘bed’ ponoka -innoka ‘elk’ Research Questions How do prefixes affect the prominence patterns of Blackfoot nominals? What can paradigmatic forms tell us about their morpho-phonology? Our Approach We elicited both bare nouns and nouns with the following prefixes, using English as a prompt. omahk- ‘big’ kaak- ‘only, just’ sik- ‘black’ ksikk- ‘white’ pok- ‘small, young’ inno- ‘long’ Acknowledgements Thanks goes to the 2011/2012 Field Methods class at UBC and especially to Beatrice, who so patiently teaches us Blackfoot. Nitsíkohtaahsitakihpinnaan! Findings 1 Two Stem Classes Class 1: Bears no lexical accent. Class 2: One syllable is associated with an accent. Prefixes like pok- ‘small’ act as a noun class litmus test: accented nouns keep their lexical accent and unaccented nouns are accented as determined by the prefix. Accented Root ‘small’ + Root Gloss ipíítaa poks-ipíítaa ‘eagle’ apáni pok-apáni ‘butterfly’ omitáíkoan pok-omitáíkoan ‘puppy’ Unaccented Root ‘small’ + Root Gloss aakii pok-ááki ‘woman’ omitaa pok-ómitaa ‘dog’ ohpokon pók-ohpokon ‘ball’ All bare forms of nominals are pronounced with at least one accent. Pre-pausally, underlyingly accent- less stems are pronounced with a pitch fall across the last syllable of the word. aakii → aakíi ‘woman’ pokon → pokón ‘ball’ imitaa → imitáa ‘dog’ 2 Three Prefix Classes Class 1: Accents the compound’s first syllable. Class 2: Accents the compound’s second syllable. Class 3: Accents the compound’s third syllable. The prominence patterns of compounds containing unaccented roots are determined by the prefix. Dif- ferent prefixes cause accent to fall on either the first, second, or third syllable of the entire word. Class 1 Prefix Prefix + Root Gloss omahk- ómahk-omitaa ‘big dog’ Class 2 Prefix Prefix + Root Gloss pok- pok-ómitaa ‘small dog’ inno- innó-(ó)mitaa ‘long dog’ Class 3 Prefix Prefix + Root Gloss kaak- kaak-omítaa ‘just a dog’ ksikk- ksikk-omítaa ‘white dog’ These prefixes only bring about accents on their as- sociated compounds when added to unaccented roots. Accented roots generally keep their lexical accent and are not prosodically affected by prefixes. 3 Syllable Weight Post-peninitial accent is attracted to the second syllable when it is heavy. Prefixes which normally accent the third syllable ac- cent the second syllable if it is heavy. Root ‘just’ + Root Gloss aakii kaa.káá.kii ‘just a woman’ i’towaaki kaa.kí’.to.waa.ki ‘just a chicken’ 4 Voiceless Syllables Voiceless syllables cannot carry pitch. If accent is predicted to fall on a voiceless syllable, accent shifts one syllable to the left. The prefix pok- normally accents the second syllable. When the second syllable is voiceless, accent shifts left to fall on the prefix itself. Root ‘small’ + Root Gloss ohpoos pó.koh.poos ‘small cat’ ohpokon pó.koh.po.kon ‘small ball’ ohmokoyi pó.koh.mo.ko.yi ‘small wolf’ References [1] Frantz, D.G. 1978. Abstractness of phonology and Blackfoot orthography design. In Approaches to language, anthropological issues: Papers written for the IXth International Congress of Anthropological and Ethnological Sciences, Chicago, 1973, eds McCormack, W. and Wurm, S.A. Mouton: 307–325. [2] Frantz, D.G. and Russell N.J. 1995 [2nd edition]. Blackfoot dictionary of stems, roots, and affixes. University of Toronto Press. [3] Frantz, D.G. 1997. Blackfoot grammar. University of Toronto press. [4] Kaneko, I. 1999. A metrical analysis of Blackfoot nominal accent in optimality theory. MA thesis, Department of Linguistics, University of British Columbia. [5] Russell, L.H.S. and Genee, I. 2006. The Blackfoot language: current position and future prospect. Talk at the 38th Algonquian conference, University of British Columbia. [6] Stacy, E. 2004. Phonological aspects of Blackfoot prominence. PhD thesis, Department of Linguistics, University of Calgary. [7] Van Der Mark, S.C. 2003. The phonetics of Blackfoot pitch accent. MA thesis, Department of Linguistics, University of Calgary. [8] Taylor, A.R. 1969. A grammar of Blackfoot. PhD thesis, Department of Linguistics, University of California Berkeley.