UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

On testing the psychological reality of phonological rules Reid, Heather Jean 1976

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

Item Metadata

Download

Media
831-UBC_1976_A6_7 R43.pdf [ 8.29MB ]
Metadata
JSON: 831-1.0093706.json
JSON-LD: 831-1.0093706-ld.json
RDF/XML (Pretty): 831-1.0093706-rdf.xml
RDF/JSON: 831-1.0093706-rdf.json
Turtle: 831-1.0093706-turtle.txt
N-Triples: 831-1.0093706-rdf-ntriples.txt
Original Record: 831-1.0093706-source.json
Full Text
831-1.0093706-fulltext.txt
Citation
831-1.0093706.ris

Full Text

ON TESTING THE PSYCHOLOGICAL REALITY OF PHONOLOGICAL RULES by HEATHER JEAN REID B.A., Simon Fraser University, 1974 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE i n THE DIVISION OF AUDIOLOGY-AND•SPEECH SCIENCES in THE DEPARTMENT OF PEDIATRICS We accept t h i s thesis, as conforming to the required standard Andre-Pierre Benguerel John Gi l b e r t THE UNIVERSITY OF BRITISH COLUMBIA August, 19 76 ( c T ) Heather Jean Reid, 1976 In presenting th i s thes is in pa r t i a l fu l f i lment of the requirements for an advanced degree at the Un ivers i ty of B r i t i s h Columbia, I agree that the L ibrary sha l l make it f ree l y ava i l ab le for reference and study. I further agree that permission for extensive copying of th i s thes is for scho lar ly purposes may be granted by the Head of my Department or by his representat ives. It is understood that copying or pub l i ca t i on of th is thesis fo r f inanc ia l gain sha l l not be allowed without my written permission. Department of I A^Yl C S j The Univers i ty of B r i t i s h Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 ABSTRACT This investigation was motivated by the growing d i s s a t i s -f a ction with the inconsistent use of empirical methodology in transformational generative phonology (TGP) and by the r e s u l t i n g l i m i t e d value which TGP has for other s c i e n t i f i c f i e l d s of study. The investigation i s concerned with judging a p a r t i c u l a r experimental paradigm for i t s v a l i d i t y as a confirmation/disconfirmation procedure with the intention of consequently confirming or disconfirming the psychological r e a l i t y of c e r t a i n phonological rules. As revealed i n the review of the relevant l i t e r a t u r e , one could j u s t i f y the v i o l a t i o n of TGP's ideal speaker-listener framework—which would r e s u l t from te s t i n g some of TGP 1s hypotheses—by using r e a l speaker-listeners. Previous t e s t i n g for the use of c e r t a i n of Chomsky and Halle's (1968) phonologi-c a l rules has raised doubts about the v a l i d i t y of claims con-cerning these rules' psychological r e a l i t y . The method used i n the present study consisted of requiring subjects to derive and pronounce novel words (without the use of pencil and paper) from e x i s t i n g English stem-words and suffixes a u r a l l y presented to them. One group of subjects was exposed to e x i s t i n g English derivations which exemplified sound patterns accounted for by Chomsky and Halle through the rules under investigation. This group was also exposed to example derivations which showed no phonetic change. A second group of subjects were exposed only to examples showing no phonetic change. Analysis of the res u l t s show, f i r s t (with respect to the present experiment's design), that the d i s t r i b u t i o n of the number of predicted responses ( i . e . the responses predicted by the rules under investigation) i n each group of subjects i s very s i m i l a r . It i s concluded that each group showed a similar a b i l i t y in performing the novel derivation task and that the subjects were representative samples of the population under study. The greater occurrence of predicted sound patterns in the responses of the f i r s t group of subjects i s attributed to that group's exposure to example derivations showing predicted phonetic changes. The o v e r a l l production of predicted sound patterns i n each group cannot be attributed to just a few subjects. A trend appears i n which stem-suffix sets which were most often involved i n given predicted phonetic changes were the same i n both groups' responses. Conclusions are also drawn with respect to the v a l i d i t y of the experimental paradigm as a v a l i d procedure for confirming or disconfirming the phonological rules i n question. F i r s t , the l o g i c a l argument which uses the po s i t i v e consequences of an hypothesis, known as "the f a l l a c y of affirming the conse-quent," i s i n v a l i d . Thus, none of the various possible s t r a t -egies of sound pattern production which were considered (in order to account for subjects' responses) could be affirmed. Secondly, a v a l i d argument of the type modus t o l l e n s can be used when the consequences of an hypothesis are negative. The v a l i d conclusion permitted by t h i s argument i s the disconfirma-tion of the hypothesis. Some of the problems encountered with - i v -t h i s argument are discussed: (a) i t i s impossible to deter-mine the exact number of times that an hypothesis i s discon-firmed in a set of data i n which some of the data consist of positi v e consequences; (b) the argument must be c a r e f u l l y quantified i n order to permit v a l i d conclusions to be drawn from data that i s derived from r e a l ( i . e . non-idealized) conditions of the world; and (c) there exists no c r i t e r i o n frequency of (non-)use for the (dis)confirmation of the 'psychological r e a l i t y of a phonological r u l e ' . If i t were possible to e x p l i c i t l y specify the extension of a rule's use, such a c r i t e r i o n frequency of a rule's (non-)use might be well motivated. In i t s absence, the psychological r e a l i t y of phono-l o g i c a l rules could not even be disconfirmed i n t h i s study. - V -TABLE OF CONTENTS Page ABSTRACT . i i LIST OF TABLES ix LIST OF FIGURES x i ACKNOWLEDGMENTS x i i Chapter I INTRODUCTION 1 Chapter II REVIEW OF THE LITERATURE 9 2.1 The meaning of the phrase 'psychological r e a l i t y of phonological rules 1 9 2.1.1 Introduction 9 2.1.2 The psychological nature of the description 9 2.1.3 Rules in the description.. 11 2.1.4 Knowledge of rules 13 2.1.5 The r e a l i t y of a theory... 16 2.2 Whether experiments on r e a l speakers can apply to theories on ideal speaker-listeners 18 2.2.1 Motivation for adhering to the i n e x p l i c i t theory based on the ideal speaker-listener frame 18 2.2.2 Whether Chomsky and Halle's mentalistic and empirical theory can be maintained.. 21 Steinberg doubts the v a l i d i t y of the theory's mentalistic nature 21 Botha questions TGP's status as a mentalistic and as an empirical theory 2 3 2.2.3 How experiments on rea l speakers cannot apply to theories about ideal speaker-l i s t e n e r s 35 - v i -Page 2.2.4 Possible j u s t i f i c a t i o n of experimentation external to the ideal speaker-l i s t e n e r frame 36 2.3 Independent v e r i f i c a t i o n procedures for phonological rules 38 2.3.1 Direct experiments as a means-of v e r i f i c a t i o n of phonological rules 39 2.3.2 Evidence from other sources which i s relevant to the findings of d i r e c t experi-mentation 65 2.4 Conclusions 6 8 CHAPTER III METHOD 69 3.1 Aims of the investigation 69 3.2 Experimental plan 72 3.3 Subjects 73 3.4 Composition of the word l i s t s .. 76 3.4.1 Practice task s t i m u l i .... 76 3.4.2 Experimental task s t i m u l i . 7 8 3.4.3 V e r i f i c a t i o n of the ap-propriateness of combining the stem-words and su f f i x e s 80 3.5 Administration and recording of the experiments 82 '3.5.1 The stimuli 82 3.5.2 Directions to the subjects 84 3.5.3 Order of presentation i n the experiments 86 3.5.4 Method of presentation ... 88 3.5.5 Recording the data 89 - v i i -Page CHAPTER IV ANALYSIS AND RESULTS OF THE DATA 90 4.1 Checking the r e l i a b i l i t y of the transcriptions 90 4.2 Estimations of the lengths of response latencies 91 4.3 The "productivity" of the phonological rules 9 3 4.3.1 Introduction 93 4.3.2 The occurrence of sound patterns in each group of subjects 94 The vowel laxing rule (VLR.. ) : V t-tense]/ CVCV 106 The primary stress rule (PSRV.: V ->• [1 s t r e s s ] / C + a f f i x 109 The vowel tensing rule (VTR,): V -* t+tense]/ V 109 The vowel tensing rule (VTR ): V [+tense]/ CiV 112 [-high] The velar softening rule (VSOR): /kd/.;.+ s / _ j 115 The underlying vowel in the second s y l l a b l e of obtain and pertain 115 The /s/-voicing rule (/s/VR): /s/ -> [+voice]/ V V 118 [+tense] Summary 120 4.4 Attempts to determine the means by which a sound pattern was produced 123 4.4.1 The influence of example derivations on 'similar' stem-suffix sets 123 4.4.2 The influence of a previous response on a l a t e r response 126 - v i i i -Page 4.4.3 The influence of d i f f e r e n t example derivations on responses involving one stem-word 12 8 4.5 Further observations 133 4.5.1 Subjects' productivity with sound patterns and with phonological rules 133 4.5.2 Stem-suffix sets most and least often involved i n pre-dicted sound patterns 138 4.5.3 Suffixes and vowels most . often involved i n predicted sound patterns 142 CHAPTER V DISCUSSION AND CONCLUSIONS 144 BIBLIOGRAPHY 156 APPENDIX I 159 APPENDIX II 164 - i x -LIST OF TABLES Table Page I Arrangement of the experiments into groups, the contents of each experiment and the number of subjects p a r t i c i p a t i n g in each experiment 74 II Estimated range, median and upper q u a r t i l e values of response latencies 92 III V-» .[-tense]/ CVCV. The number of responses in each category and the percentage of occurrence of each category for the l i s t s of stimuli relevant to the rule mentioned. The data for the short l i s t of stimuli are on the l e f t side of each column while the data for the long l i s t are on the r i g h t side 107 IV V -* [1 s t r e s s ] / C + a f f i x . (See Table I l l ' s legend.) The 'S' in SSSS+SS in the categories above stands for ' s y l l a b l e . ' 110 V V + [+tense]/ V. (See Table I l l ' s legend.) I l l VI V [+tense]/ Civ. The number of [-high] responses in each category and the percentage of occurrence of each category for the single l i s t of stimuli relevant to the rule mentioned 114 VII /k d/ -> s/ I. (See Table I l l ' s legend.) 116 VIII V ->• [-tense]/_ (cvcvl . A b i l i t y to extend the I C C j double consonant sound pattern in the context -ion was assumed. (See Table V I 1 s legend.) 119 IX s -*[+voice]/ V V. (See Table VI's [+tense] legend.) 121 X The matrix used for the chi-square measure of both groups' responses to slave + - i f y 124 XI Example of a matrix arrangement for an i n d i v i d u a l group's responses to a pair of stem-suffix sets... 127 2 XII X and r t l e v e l s of significance of d i f f e r e n t pairs of stem-suffix sets. 129 -X-Table Page XIII The percentages of occurrence of unpredicted sound patterns in the responses (long l i s t s ) of Group B which were relevant to each r u l e . . . 137 XIV Ranges in the percentages of unpredicted re-s sponses, the mean percentages and the average deviations from the means for Group B (long l i s t s ) 139 XV Stem-suffix sets whose responses most often showed the predicted sound pattern. The % i s based on the number of times (out of the t o t a l number of responses to that set) that the set e l i c i t e d a predicted response. Note that /s/VR i s not l i s t e d since Group B did not produce any response predicted by i t 140 - x i -LIST OF FIGURES Page An example of an argument. Figure from Botha, 1970 . . . 26 Cumulative number (N) of predicted responses (short l i s t ) for each group 102 Cumulative percentage of predicted responses (short l i s t ) for Group A ( ) and Group B (——) 103 - x i i -ACKNOWLEDGEMENTS I wish to thank Prof. A.-P. Benguerel for his constant guidance, encouragement and humour during the course of t h i s project. In addition, I am appreciative of h i s work with a d i g i t a l computer in programming the various s t a t i s t i c a l mea-sures and in determining the d i s t r i b u t i o n of tetrachoric c o r r e l a t i o n c o e f f i c i e n t values. I would also l i k e to thank Prof. J.H.V. Gi l b e r t for his examination of the thes i s . I wish to thank Prof. J.B. Delack who amiably offered advice and suggested readings to aid i n the investigation. I am grate f u l also to Amy Fleming for her help i n retranscribing several of the subject interviews and to Marion Jacques for her competent typing of the thes i s . And of course I extend my thanks, l a s t but not least, to the students who par t i c i p a t e d as subjects in the study. One further acknowledgment: I have greatly appreciated the opportunity during the l a s t two years to study alongside stu-dents from d i f f e r e n t f i e l d s . CHAPTER I INTRODUCTION Tranformational generative grammar has as i t s goal the description of a system of rules which s p e c i f i e s the sound-meaning correspondences in language (Chomsky and Halle, 1968, p. 3 ) . The sound-meaning correspondences refer to the r e l a t i o n between an ideal phonetic form and an associated i n t r i n s i c semantic interpretation of each sentence within the set of possible sentences for a language. The rule-system specifying a l a n - . guage's sound-meaning relationships i s assumed to be a native speaker's i n t e r n a l i z e d "knowledge" of the language. This "knowledge" i s also c a l l e d the "competence" of the native speaker. Whereas a speaker's "performance" refers to the actual use of the language i n concrete situations (Chomsky, 1965; p. 3 ) • the knowledge- or^competence•of the speaker-is found in an ide a l i z e d speaker-listener r e l a t i o n s h i p . This r e l a t i o n -ship involves an i d e a l i z e d speaker-listener who (a) i s part of a completely homogenous speech community, (b) knows the l a n -guage p e r f e c t l y and (c) i s unaffected by grammatically i r -revelant conditions such as memory l i m i t s , d i s t r a c t i o n , i n a t -tention and n o n - l i n g u i s t i c knowledge and b e l i e f s i n applying his "knowledge" of the language i n actual performance. Chomsky believes that i t i s only within the framework of the ideal speaker-listener r e l a t i o n s h i p that i t i s possible for a speaker's performance to d i r e c t l y r e f l e c t his competence (Chomsky, 1965, p. 4 ) . The d i s t i n c t i o n between competence and performance i s believed to permit the study of uncontaminated -1--2-l i n g u i s t i c knowledge by abstracting away other n o n - l i n g u i s t i c factors involved i n the use of language. As Chomsky recently said, "The s c i e n t i s t wanting to study rules [of grammar] must clean away a l l of the interactions..." or no n - l i n g u i s t i c phen-omena from the i n t e r n a l i z e d knowledge of the speaker. The d i s t i n c t i o n between competence and performance derives motivation from r a t i o n a l i s t philosophy which holds that the mind (or reason) i s the sole source of human knowledge (Lyons, 1970/ p. 96) . In t h i s view i t is" hypothesized that much of the human's l i n g u i s t i c knowledge i s innate. The innate l i n g u i s t i c knowledge i s c a l l e d universal grammar and i s thought to be comprised of p r i n c i p l e s stating what kinds of rules for l a n -guages may e x i s t . Exposure to a given language permits the c h i l d with his innate or universal knowledge to determine language s p e c i f i c rules or competence. The ultimate goal of transformational generative grammar seems to be to study universal grammar which hopefully can be inferr e d from the study of the p a r t i c u l a r grammars of many la n -guages. In t h i s r a t i o n a l i s t t r a d i t i o n i t i s thought that one may approach the age-old epistemological problem of gaining knowledge of what one knows. That i s to say, by s t r i p p i n g away the phenomena of language performance and considering only the i d e a l i z e d speaker-listener's competence, i t i s thought possible to i n f e r the p r i n c i p l e s of universal grammar. Both universal grammar and the grammar for a p a r t i c u l a r language are mentalistic i n nature. Botha (1971, p. 117) quotes Chomsky as saying: -3-Thus, at several l e v e l s the l i n g u i s t i s involved i n a construction of explanatory theories, and at each l e v e l there i s a clear psychological interpretation for his t h e o r e t i c a l and descriptive work. At the l e v e l of a p a r t i c u l a r grammar, he i s attempting to characterize the knowledge of a language, a c e r t a i n cognitive system that has been developed—uncon-sciously, of course—by the normal speaker-hearer. At the l e v e l of universal grammar he i s t r y i n g to e s t a b l i s h c e r t a i n general properties of human in t e l l i g e n c e . L i n g u i s t i c s , so characterized, i s simply the s u b f i e l d of psychology that deals with these aspects of mind. (Chomsky, 1964, p. 24.) The metaphysical d i s t i n c t i o n between competence and per-formance dictates that the rule systems of a language w i l l be considered from the point of view of the i d e a l speaker-listener r e l a t i o n s h i p . This means that the empirical study of grammar w i l l exclude v e r i f i c a t i o n of i t s theories by evidence gained from experimentation done on r e a l speaker-listeners. Recall that t h i s i s because Chomsky i s not interested i n studying phenomena intera c t i n g with and "contaminating" competence i n a speaker's performance. The study of the phonological sub-component of transforma-t i o n a l generative grammer i s c a l l e d transformational generative phonology (and i s henceforth referred to as TGP). Many re -searchers i n phonetics and phonology are opposed to TGP's metaphysical constraints denying them access to experimental paradigms outside the realm of the i d e a l speaker-listener. Two general ideas from t r a d i t i o n a l empiricism are represented in the objections to the p r o h i b i t i o n of v e r i f i c a t i o n procedures external to the ideal speaker-listener frame. One idea i s that a l l theories must be tested experimentally or else r i s k maldevelopment. L i n e l l (1974, p. v i i ) claims that -4-i A theory runs the r i s k of degenerating into a mere systematization of data, unless one t r i e s to f i n d empirical interpretations for the various theoret-i c a l e n t i t i e s that are proposed. Ohala likewise proposes that " . . . a l l hypotheses require experi-mental v e r i f i c a t i o n " (1973, p . l ) . In cases of TGP where i t has been empirically concluded without experimentation that phono-l o g i c a l rules and rule constraints are part of the r e a l speaker-l i s t e n e r ' s knowledge, Ohala (1974,.p. 19) claims that" the r e s u l t seems to be a " . . . t h e o r e t i c a l e d i f i c e that i s enormous, elabor-ate and very, very f r a g i l e . " A second idea representing objections to the i n a c c e s s i b i l i t y of TGP to experimentation on non-ideal speakers i s that a theory must play a role i n explaining i t as well. The route chosen by Chomsky to explain the data i s one where there i s f i r s t developed a rule system or s t r u c t u r a l description for a l a n -guage which i s i n accordance with the p r i n c i p l e s of universal grammar (Greene 1972, p. 34). Universal grammar may be derived by studying d i f f e r e n t languages with the plan of i d e n t i f y i n g p r i n c i p l e s common to them a l l which could plausibly be con-sidered to be part of the c h i l d ' s innate l i n g u i s t i c knowledge. Chomsky also believes that the p r i n c i p l e s of universal grammar may be discovered through the use of the "evaluation measure" which assigns a value to a grammar or to a sequence of rules. The function of the evaluation measure i s to select the most highly valued of competing alt e r n a t i v e grammars of a language on the assumption that t h i s grammar w i l l be the one which children would develop in learning the language. Greene (p. 30) describes Chomsky's b e l i e f that children must have some innate -5-l i n g u i s t i c a b i l i t y permitting them to choose one type of grammar which i s appropriate for analyzing language i n general. Chomsky holds, for instance, that t h i s one type of grammatical analysis that children are programmed to develop must be univer-sal to a l l languages. "This universal grammatical theory would give an account of the grammatical forms and relations that are common to a l l languages..." (Greene, 1972, p. 31). Uni-versal grammar therefore accounts for or 'explains' (in what L i n e l l (1974, pp. 147-149) c a l l s a weak sense of "'explanation'") the grammatical forms and re l a t i o n s of a language s p e c i f i c grammar which in turn accounts for or explains the observable phenomena. According to L i n e l l , generative theory presupposes a sense of explanation in which observable phenomena are subsumed under a "'covering'" t h e o r e t i c a l p r i n c i p l e . The th e o r e t i c a l p r i n c i p l e must co r r e c t l y predict the observable phenomena. L i n e l l believes that explanations such as these are weak since correct predictions can be produced by f a l s e theory just as well as by true theory and by theories which are intended to represent r e a l i t y i n varying degrees. Explanations of phen-omena (by means of a th e o r e t i c a l p r i n c i p l e which c o r r e c t l y predicts the phenomena) can be strengthened i f independent reasons are found which lead one to believe that the theoret-i c a l p r i n c i p l e i s true. Some l i n g u i s t s prefer to 'explain' the data f i r s t in t h i s stronger sense and then proceed to develop universal p r i n c i p l e s for language. For instance, Lindblom doubts the explanatory power of Chomsky and Halle's -6-(1968) phonological theory wherein l i n g u i s t i c form has primacy over the "variables of language use and i t s substantive bases" (1971, p.5). Lindblom (1971, p.8) disagrees with Chomsky and Halle that one should t r y to develop an abstract t h e o r e t i c a l apparatus to account for phenomena without r e l a t i n g the postu-lated mental structures and processes to the phys i o l o g i c a l mechanism. Concerned with explaining the physical phenomena of language, Lindblom (pp.7-8) suggests re-evaluating the notion of ' l i n g u i s t i c a l l y relevant f a c t s ' (and l i n g u i s t i c a l l y relevant phonetic facts) while keeping i n mind the p o s s i b i l i t y of assigning phonetics a less peripheral role i n l i n g u i s t i c inquiry. His alte r n a t i v e to the abstract theory of Chomsky and Halle would be a theory which uses phonetics to predict phonological phenomena by beginning with hypotheses f o r phy s i c a l l y based preconditions of speech communication and development. Ohala (1974,,pp. 1-3) states that the f i r s t task i n phonology i s to discover the sound patterns and that the second task i s to give a causal explanation of some aspect of the patterns. Language and speech, he reasons (Ohala 1974, pp. 18-19), are phys i c a l l y and psychologically r e a l systems and therefore are limi t e d systems. He believes that "The problem, then, i s to constrain the range of hypotheses we entertain i n the same way that the r e a l world i s constrained." Since he believes there are s t r i c t e r or more numerous constraints on what can be ex-plained by physical phonetic factors, Ohala recommends the use of phonetics (over psychology, history or other possible -7-sub-disciplines of phonology) as a tool for explaining aspects of phonological r u l e s . An implication of the above objections to TGP 1s immunity to experimental v e r i f i c a t i o n on non-ideal speakers i s the l i m i t a t i o n of the number and form of independent research strategies available to those who would study phonology within the i d e a l i z e d speaker-listener frame. The absence of research strategies of a non-formal nature i s a possible r i s k for the generality of any theory which, from the point of view of t r a d i t i o n a l empiricism, must be derived from greater numbers of '"independent motivation, argumentation or reasoning V . Empirical science i n general requires, i n Botha's words (1971, pp. 229-230 and fn. 29), that "...postulation and use of... concepts must be motivated by showing that they are required for diverse reasons, no two of which are interdependent." The main objectives of t h i s project are f i r s t to carry out an experiment on r e a l speakers to test the psychological r e a l i t y for that group of speakers of some of Chomsky and Halle's (1968) general phonological rules. Another objective i s to draw conclusions on the v a l i d i t y of the experimental paradigm i t -s e l f as a confirmation of disconfirmation procedure. The ex-periment w i l l consist of giving a word-derivation t a s k — t h a t of s u f f i x a t i o n — t o adult, native English speakers to see i f the phonetic form of t h e i r responses r e f l e c t s (or does not r e f l e c t ) the use of certain general phonological rules. It w i l l be h e l p f u l to further c l a r i f y the motivation for and the goals of research of those groups of l i n g u i s t s who are -8-opposed i n t h e i r views toward v e r i f i c a t i o n procedures i n TGP. The experiment of the present study would not be included by Chomsky and Halle as a method of v a l i d a t i o n within the frame-work of TGP. Therefore, following the recommendation of Botha (1968/ p. 104), attempts will-be made (a) to e x p l i c i t l y state how experiments on non-ideal speaker-listeners can (or cannot, as the case may be) apply to theories on i d e a l speaker-listeners; (b) to give e x p l i c i t c r i t e r i a for confirming or disconfirming the mentalistic claims of certain phonological rules; and (c) to give an explanation (as well as the method for a r r i v i n g at that explanation) of the occurrence of both p o s i t i v e and nega- ' t i v e evidence for the phonological rules within the responses of single speakers and groups of speakers. 9 CHAPTER II REVIEW OF THE LITERATURE 2.1 The meaning of the phrase 'psychological r e a l i t y of phonological r u l e s ' 2.1.1 Introduction In a recent paper e n t i t l e d 'Perspectives i n Phonology', Fischer-JjzJrgensen (1975, p.221) characterizes the present state of phonological theory as one i n which most of the basic assumptions are the object of serious c r i t i c i s m and i n which many points are being revised by professed adherents of the theory. Some of the points on which l i n g u i s t s disagree are involved i n "the very basis of generative phonology, the claim that the description has psychological r e a l i t y " (Fischer-Jo'rgensen, 1975 , p. 219) . Before describing the attempts of th i s and other projects to fi n d evidence for or against the psychological r e a l i t y of phonological description, comment must be made on the various interpretations of the phrase 'the psychological r e a l i t y of phonological rules.' 2.1.2 The psychological nature of the description It was mentioned in the introduction that both the un i -versal grammar and the grammar for a p a r t i c u l a r language are mentalistic i n nature. Botha (1971, p. 116) points t h i s out in saying that the object of study of a p a r t i c u l a r grammar and that of the general l i n g u i s t i c theory are mental f a c u l t i e s . The mental entity which a "language-particular" (that i s , language-specific) grammar represents i s l i n g u i s t i c compe-tence, the i n t e r n a l i z e d rule-system by means of which the ideal speaker-listener relates the sound signals of an i n d e f i n i t e -9--10-number of sentences to t h e i r respective semantic i n t e r p r e t a -tions and vice versa. Universal grammar i s meant to o represent the 'faculte de langage' of an i d e a l speaker-listener which i s taken to be a language independent mental'! capacity enabling the speaker-listener to acquire l i n g u i s t i c competence. Chomsky does not make clear which aspects—form, substance, structure, or anything e l s e — o f the 'mental f a c u l t i e s ' are characterized by the universal and language-specific grammars. Other l i n g u i s t s have been more e x p l i c i t concerning the r e l a t i o n of grammars and mental e n t i t i e s . "Katz regards the r e l a t i o n -ship between the structure of the l i n g u i s t i c theory and that of the mental mechanism as one of isomorphism" (Botha, 19 71, p. 119). Kiparsky, on the other hand, believes that a grammar should c o r r e c t l y represent both the substance and the form of the fluent native speaker's language-specific knowledge. Botha concludes that the isomorphic r e l a t i o n between theory and men-t a l faculty found in the approaches of Katz and Kiparsky means that l i n g u i s t i c theory for them i s a conceptual analogue to mental structures. In his metaphysical assumptions for the l i n g u i s t i c theory's representation of psychological r e a l i t y , L i n e l l proposes that the grammar' s i n t e r n a l structure "be_ isomorphic to the speaker' s underlying psychological r e a l i t y " with respect to forms (such as the d i f f e r e n t forms or l e x i c a l units which are there) and o structures (such as properties of the forms, r e l a t i o n s between and generalizations over the forms, etc.) ( L i n e l l , 1974, p . l ) . -11-2.1.3 Rules i n the description Another fundamental concept i n TGP which e l i c i t s disagree-ment amongst l i n g u i s t s i s the notion of 'rule'. L i n e l l (1974, pp. 30-32) reports that 'rule' i s a vague notion i n science and i n ordinary language. In general, a 'rule' i s a proposition, the formulation of which includes indications of a class of human actions, a class of persons who perform these actions and a modality for the actions involved (required, forbidden, permitted). ( L i n e l l , 1974, p. 30.) L i n e l l refers to Black who states that the uses of the term range from " ' e x p l i c i t norms'" to the "'degenerate'" sense in which 'rule' i s almost synonymous with "'generalization'" and "'general assertions about matter of f a c t . ' " L i n e l l mentions some performance evidence c i t e d by Black for the 'psychological r e a l i t y ' of rules. The evidence consists of speakers' readiness to correct themselves and others, t h e i r willingness to believe that there i s a rule even i f they do not know what i t i s , and t h e i r endorsement of rules a f t e r they have been formulated by an onlooker. ( L i n e l l , 1971, p. 31.). To t h i s L i n e l l adds that second language learners r e a l i z e that i t i s not enough merely to be understood, as there are rules for what counts as correct which go beyond that. L i n e l l men-tions further possible support for the existence of rules which i s supplied by Searle. Searle proposes that phonological rules are not just r e g u l a r i t i e s which may be described by an observer; he feels that they are also "'constitutive'" of language since "'we recognize departures as mispronunciations'" and since the rule projects to cover new cases ( L i n e l l , 1974, p. 31).. -12-A question which i s fundamental to the use of 'rule' i n l i n g u i s t i c theory i s whether language behaviour i s rule-guided or whether, on the other hand, i t i s merely rule-conforming ( L i n e l l , 1974, p.31). In the l a t t e r case the rules would only represent r e g u l a r i t i e s that an observer could extract from language behaviour. Chomsky believes that l i n g u i s t i c behaviour occupies an intermediate position between conscious rule g u i -dance and mere conformity to rules. This b e l i e f might be paraphrased by saying that the speaker i s directed by rules but i s not consciously aware of the rules. The sense of rule which L i n e l l prefers i s the weak sense i n which rules are only r e g u l a r i t i e s in the observed data. He reaches t h i s conclusion af t e r deciding that there are probably many types of awareness covered by the verb 'know' when l i n -guists say that a speaker 'knows' the .'rules' of his language. That i s to say, L i n e l l does not appear to be prepared to con-sider 'rule' as being constitutive of human language u n t i l there i s c l a r i f i c a t i o n of the term 'to know' (a r u l e ) . L i n e l l q u a l i f i e s his notion of the "weak" or "regulative" sense of rule by r e j e c t i n g Katz's idea that rules are isomor-phic to causally e f f i c i e n t neurophysiological processes. He states that "rules" and "causal processes" are on d i f f e r e n t l e v e l s of explanation i n the same way as are "reasons" and "causes." His f i n a l notion of 'rule' i s that rules govern l i n g u i s t i c behaviour i n the sense that the speaker chooses (consciously or unconsciously, deliberately or habitually) to follow them. L i n g u i s t i c competence i s not a causally e f f i c i e n t force i n speech production; rather i t defines the l i n g u i s t i c conditions which must be met...(pp.31-32). -13-Line 11 1s interpretation of rules governing l i n g u i s t i c behaviour i s also ambiguous i n the sense that i t i s not clear whether the speaker chooses either "consciously 1 1 and "deliber-ately" or "unconsciously" and "habitually"; or whether i t would be possible for the speaker to choose "consciously" and "habitually" or "unconsciously" and "deliberately"; and whether the choosing i s consistently of one nature, whatever that might be. 2.1.4. Knowledge of rules Rules are the things which a native speaker i s assumed to know, or i n other words, to be aware of. In order to discuss further the sense of rule i t i s appropriate to discuss l i n -guists' use of the following frequently interchanged terms: "knowledge," "to know," "awareness" and "to be aware of". Once again, as i n the case of "rule," the words are assigned d i f f e r e n t meanings which are not always e x p l i c i t l y stated by the authors. Zimmer (1969, p.309) refers to the phrase "to know a l i n -g u i s t i c r u l e " i n the sense of knowing a mathematical formula. He does not say whether knowing a formula involves the sense of 'knowing how1 to use the formula, the sense of 'knowing what' the formula i s or the sense of merely 'knowing that' the formula e x i s t s . Hockett (1968) interprets Chomsky's (1965) sense of "know-ledge" to be an epistemological sense, whatever that might be. In Language and Mind (1972) Chomsky refers to "knowledge" as an underlying system of b e l i e f s . An epistemological sense of -14-"knowledge" i s referredto by Carterette and Friedman (1974, p.7). They state that the Cartesian (that i s , r a t i o n a l i s t ) doctrine of perception and b e l i e f holds that knowledge i s a psychophysical judgement that i s self-warranted. Chomsky (1965) states that when a speaker "knows" the grammar of h i s language he has "knowledge" of that grammar which cannot e x p l i c i t l y be described by the speaker. He believes that t h i s knowledge, considered as part of the mind's proper-t i e s and content, may not even be accessible to the speaker's introspection ( L i n e l l , 1974, p.13). L i n e l l (1974, p.162, fn.23) describes another l a t e r discus-sion by Chomsky of his use of the word'"'knowledge. ' " Chomsky purports to meaning neither "'knowledge that' "nor "'knowledge how.'" Rather, he intends "'something i n between' " such as "' t a c i t or unconscious knowledge.'". Just as i n defining the sense of'''rule,' Chomsky here situates h i s notion i n between two polar senses and then r e s t r i c t s the i n d e f i n i t e meaning by r e f e r r i n g to the t a c i t nature of the object of d e f i n i t i o n (such as 'rule' or 'knowledge' ) . Some authors' interpretations of 'knowledge' r e f l e c t the view which sees language as "something which the speaker uses for communication with other people and with himself, and... [which] i s shaped by i t s functions i n communication" ( L i n e l l , 1974, p.27). They define and use l i n g u i s t i c 'knowledge' from a "behavioural perspective." For instance, Hockett (1968, p. 63) states h i s preference as the "know how to" sense of the word -15-'know.' Rather than use the ambiguous word 'knowledge' Hockett substitutes the word "habit." "The term 'habit' i s l i t t l e more than a paraphrase of the know how to...sense of 'to know'." Hockett finds the phrase "to have the habit of one's grammar" as less misleading, from an empirical point of view, than Chomsky's "have knowledge of" sense. This i s because although a habit may provide the basis for and the pre-dictions from a given generalization, i t may not, as part of empirical evidence, endure forever and so i s subject to modi-f i c a t i o n . In other words, a habit i s not a "Law of Nature" as Hockett feels Chomsky's sense of self-warranted "knowledge" i s apt to imply. L i n e l l (19 74, p.27) contends that l i n g u i s t i c knowledge should include what he c a l l s "grammatical competence" and "general l i n g u i s t i c competence." Grammatical competence refers to the 'know that' sense of knowledge. The speaker 'knows' a system of obligatory conditions which utterances must meet in order to be recognized as grammatical. L i n e l l q u a l i f i e s his sense of 'knowledge that' by stating that t h i s knowledge i s not always e x p l i c i t . "Sometimes the speaker knows ex-p l i c i t rules, but t y p i c a l l y he i s not able to formulate rules e x p l i c i t l y " . (p.. 162, fn.23). General l i n g u i s t i c competence, on the other hand, concerns the speaker's s k i l l or a b i l i t y "to manipulate the language a c t i v e l y , with precision and va r i a t i o n within the grammatical r u l e s . " This sense of know-ledge i s of course the 'know how to' sense and i s ind i c a t i v e of a 'behavioural perspective' on language. -16-Wang (1968, p.707) believes that a speaker has d i f f e r e n t degrees of "awareness" (or 'knowledge') of l i n g u i s t i c r u l e s. He suggests that the d i s p a r i t i e s in awareness may be dependent on several factors: the formal character of the r u l e , the complexity of the ru l e , the number of items i n the lexicon for which the rule i s relevant, a combination of the preceding factors or possibly some other factor. Unfortunately Wang does not say which sense of knowledge he i s r e f e r r i n g t o — t h e 'know how to' sense, the 'know that' sense or some amalgamation of the two. F i n a l l y , Ladefoged (1970) expresses a concern with the term "knowledge" which i s founded on his b e l i e f that a speaker's "knowledge" may have d i f f e r e n t sources: a p r e l i t e r a t e source, a spelling-influenced source and a grammatically (that i s to say, l i n g u i s t i c a l l y ) influenced source. Ladefoged complains that the vaguesness of the notion of "knowledge" prevents the grammarian from knowing which knowledge of which d i f f e r e n t sources should be included i n or excluded from phonological models. 2.1.5 The r e a l i t y of a theory It has been seen that many l i n g u i s t s do not agree on t h e i r interpretations of "psychological" theory, "rule" and "know-ledge." Some interpretations are vaguely defined. In p a r t i c u -l a r , Chomsky's interpretations, which are most relevant to the aims of t h i s study, are i n d e f i n i t e . He makes no statement re-garding which aspect of the mental faculty the psychological l i n g u i s t i c descriptions should refer to. His description of -17-"rule" as an unconscious b e l i e f that i m p l i c i t l y guides the speaker i s somewhat more clear. But the interpretation of "knowledge" as being "in between 'knowledge th a t 1 and 'know-ledge how1" ( L i n e l l , 1974, p.162, fn.23) leaves his readers with the task of imagining the nature of t h i s intermediary concept of knowledge. The vague nature of the phrase under discussion i s yet increased by the permissive d e f i n i t i o n of the ' r e a l i t y ' of a theory accepted by philosophers of science. L i n e l l (1974, p.12) refers to a statement made by Harre that a " ' r e a l i s t i c ' " or "'representational'" concept of a theory can be supported without claiming that a l l of the components of the theory are r e a l . Yet representational theory contends that "the t h e o r e t i c a l e n t i t i e s and processes refer to r e a l (though non-observable or inaccessible) e n t i t i e s and processes which are assumed to stand in a causal r e l a t i o n to the observable phenomena" ( L i n e l l , 1974, p.155, f n . l ) . Thus the theory i s permitted to depict or repre-sent only c e r t a i n causal r e l a t i o n s i n an inaccessible r e a l i t y . L i n e l l states that "Chomsky, Halle, Postal and other gen-e r a t i v i s t s conceive of generative phonology as a representa-t i o n a l theory of psychological r e a l i t y " . (1974, p.12). He adds that Chomsky and Halle prefer to think of the r e l a t i o n be-tween competence and performance as rather abstract and so they might describe t h e i r theory as being one of "moderate" realism'. They therefore wish to take advantage of the f l e x i -b i l i t y of representational theory to represent only certain -18-causal r e l a t i o n s i n r e a l i t y . On the other hand, a claim i s sometimes made for a l l of the components of generative grammar to have mentally r e a l referents. L i n e l l c i t e s a statement by Katz who says that "'every aspect of the mentalistic theory involves psychological r e a l i t y ' " ( L i n e l l , 1974, p.13). He also mentions Whitaker's b e l i e f that "'hypotheses about the phonological, syntactic and semantic structure of language in fact represent an underlying psychological r e a l i t y ' " ( L i n e l l , 1974, p.13). In conclusion, the terms used i n the phrase 'the psycho-l o g i c a l r e a l i t y of phonological rules' are often assigned d i f f e r e n t senses by d i f f e r e n t l i n g u i s t s . The whole phrase therefore takes d i f f e r e n t senses depending on the l i n g u i s t who uses i t or on the reader who must- f i l l i n for himself some of the i n d e f i n i t e points of the phrase. The terms and the statement when used by Chomsky are vague on c e r t a i n points mentioned e a r l i e r . This i s due i n part to the statement and i t s terms having been derived from a body of theory which i s ri d d l e d with t h e o r e t i c a l ; components which are i n e x p l i c i t l y described. One naturally asks, f i r s t , why such an i n e x p l i c i t l y defined theory i s maintained and second, how the theory i s maintained i n an empirical inductive approach to problem solving by an orthodox school of l i n g u i s t s . 2.2 Whether: experiments on r e a l speakers- can apply to theories  on ideal speaker-listeners" ~ ' 2.2.1 Motivation for adhering to the i n e x p l i c i t theory  based on the i d e a l speaker-listener frame Linguists who d i f f e r i n t h e i r opinions on the use of the -19-ide a l speaker-listener frame for v e r i f i c a t i o n of generative grammar's theories none the less would probably share Chomsky's view that man i s endowed with a number of special f a c u l t i e s which play a c r u c i a l r o l e i n our ac q u i s i t i o n of knowledge and enable us to act as free agents, undetermined (though not necessarily unaffected) by external s t i m u l i i n the environment (Lyons, 1970, p.98). They do not, however, agree with Chomsky's categorical r e -je c t i o n of Skinnerian behaviourist psychology and his conse-quent dismissal of experiential phenomena such as instrumental and impressionistic data gained from observing the language performance of a r e a l speaker. Chomsky's opponents f e e l that i t i s important to consider 'experience' for numerous reasons. One reason i s that the use of language, in communication with others and with s e l f , probably has a r o l e in shaping language. Chomsky might answer that the use of language may shape language-s p e c i f i c grammars but would not influence universal grammar, the discovery of which i s the goal of his theory. From a synchronic point of view, language may not a f f e c t the human "faculte de langage." However, speaking from a broad d i a -chronic view, that i s to say, an evolutionary view, language use would be expected to play a role in shaping the innately endowed grammar i n man. To resolve t h i s potential area of disagreement l i n g u i s t s should decide whether the cognitive system of man, i n which the "faculte de langage" i s presumed to play some rol e , i s to be considered on an evolutionary or a s t a t i c basis. Another reason why experiential language data i s not -2 0-ignored by some of Chomsky's opponents i s the b e l i e f , men-tioned before, that the range of hypotheses on language must be constrained in the same r e a l way that language systems are delimited i n humans. Since language i s p h y s i c a l l y observable, the data used for hypothesis formation should also deal with the p h y s i c a l l y observable phenomena of speech and language. Another motive for considering experiential data i s t h e i r reproval of Chomsky's repudiation of behaviourist theory as a means of explaining human behaviour. Chomsky rejects the behaviourist approach since i t does not deal with (nor t r y to describe) the p r i n c i p l e s of innate knowledge. However, just because behaviourist theory i s "primitive" i n the foregoing sense does not require i t to be rejected by a type of "how-else argument." According to Botha, the "how-else" argument of Chomsky and Halle would state that the abstract i d e a l i z e d listener-speaker frame i s the correct t h e o r e t i c a l approach since i t i s the only approach which can describe the p r i n c i -ples of innate knowledge. Botha refutes the "how-else" argument by pointing out that The fact that, in a given f i e l d of inquiry, there i s at a s p e c i f i c moment only one theory that works and that no alternative can be proposed i s no more than a coincidence (1971, pp.125-126). The d i f f e r e n t approaches to the study of TGP give im-portance either to abstract mental p r i n c i p l e s and e n t i t i e s , on the one hand, or to the sounds actually pronounced and per-ceived ( L i n e l l , 1974, p.150). Transformational generative theory, with i t s obscure and i n d e f i n i t e aspects, i s maintained -21-by certain l i n g u i s t s because i t successfully propagates ab-stract mental p r i n c i p l e s and e n t i t i e s which are r a t i o n a l i s t i c constructs. Chomsky (1965) regards his general l i n g u i s t i c theory as "a s p e c i f i c hypothesis, of an e s s e n t i a l l y r a t i o n a l -i s t case, as to the nature of mental structures and processes" (p.53). Chomsky and Halle (1968) f e e l that since the empirical v e r i f i c a t i o n of abstract mental e n t i t i e s w i l l require i n d i r e c t and subtle means of v e r i f i c a t i o n , t h e i r abstract theory should be maintained. They say (p. 332) that "for the foreseeable future, the study of language and mental processes w i l l have to be c a r r i e d out at such a l e v e l of abstraction i f i t i s to make s i g n i f i c a n t progress." 2.2.2 Whether Chomsky and Halle's mentalistic and empirical  theory can be maintained Steinberg doubts the v a l i d i t y of the theory's  mentalistic nature C r i t i c i s m of transformational generative grammar i n general and of TGP i n p a r t i c u l a r as v a l i d mentalistic and empirical theories i s growing. Steinberg (1975) claims that Chomsky has invalidated his position as a mentalist by including some "formalisms" i n his intended mentalistic theory. A "mentalis-t i c " theory i s held by Steinberg to be one in which a l l aspects of the theory are held to be psychologically s i g n i f i c a n t ; that i s to say, a l l aspects of the theory relate to ideas of the speaker. A " f o r m a l i s t i c " theory, on the other hand, requires only certain aspects of the theory to be psychologically s i g n i f i c a n t . It was stated e a r l i e r that Chomsky and Halle's -22-concept of the ' r e a l i t y ' of a theory was a 'representational' concept i n which, indeed, not a l l components of the theory are claimed to be r e a l . Steinberg proposes that a certain functioning of Chomsky's the o r e t i c a l grammar—the competence process of constructing derivations—does not relate to the psychological function of the speaker: while a t h e o r e t i c a l grammar i s said to construct derivations according to a process which begins with the symbol S and proceeds to the Base and from there to the Transformational, Semantic, and Phonological (sub)components, a grammar in t e r n a l i z e d by a speaker i s not hypothesized to have t h i s process. Deriva-tions are said to be constructed by speakers i n some other way. Chomsky indicates that a speaker would require a set of h e u r i s t i c s or [else] Use rules, i n addition to the grammar, i n order to produce deriva-tions (1975, p.246). The fact that psychological significance i s not given to the process by which the grammar provides derivations r e s u l t s in inconsistency for a theory which i s supposed to be concerned with mentalism. This formalism i n Chomsky's mentalistic theory also r e s u l t s i n a r b i t r a r i n e s s as no j u s t i f i c a t i o n i s given for treating the process underlying the grammar as a psychologically i n s i g n i f i c a n t component. Steinberg believes that even i f j u s t i f i c a t i o n were given for incorporating formalisms into a mentalistic theory, (for instance Chomsky might claim that h i s theory was meant to be a representational one of r e a l i t y ) , the positing of those formalisms could s t i l l r e s u l t i n the psycho-l o g i c a l i n v a l i d i t y of several components of the theory. For instance, given that the underlying process of competence i s denied psychological significance, the nature of the process -23-a f f e c t s the character and composition of structures in a derivation. The r e s u l t i n g derivations would be psychologically i n v a l i d . In addition, he states that i t would be absurd to think that a psychologically i n v a l i d process could provide a determining basis for the discovery of psychologically v a l i d rules. Steinberg also believes that the organization of the grammar—that i s , the input and output relations among the syntactic, semantic and phonological components—is rendered psychologically i n v a l i d . This follows f i r s t from the fact that the grammatical organization i s i n accord with the mentally i n v a l i d generative process which underlies the grammar; second, from the fact that the grammar's derivations are a r e f l e c t i o n of the generative process and grammatical organiza-t i o n . F i n a l l y , Steinberg proposes that l i n g u i s t i c generativity cannot be psychologically meaningful i n l i n g u i s t i c s when the derivations which a grammatical theory generates are psychologi-c a l l y i n v a l i d . Steinberg o f f e r s two d i f f i c u l t remedies for "correcting the inherently contradictory theory which Chomsky has developed" (p.250). One solution i s for Chomsky to "renunciate mentalism" and declare that the grammar and i t s output derivations are non-psychological formalisms. Another solution involves a r a d i c a l r e v i s i o n of the theory, which Steinberg only b r i e f l y outlines, so that i t conforms with how speakers might reasonably be claimed to construct derivations. Botha questions TGP' s status as a mentalistic and as  an empirical theory TGP's status both as a mentalistic and as an empirical -24-theory has been censured by Botha i n numerous works (1968, 1970 and 1971). In general, his c r i t i c i s m i s based on a l i n e of reasoning which points out that the patterns of argumenta-? tio n used to j u s t i f y l i n g u i s t i c hypotheses are i n v a l i d . In The Methodological Status of Grammatical Argumentation (19 70) Botha derives his c r i t e r i a for a v a l i d argumentation pattern from Toulmin. (Although Toulmin's arguments are mainly concerned with jurisprudence, Botha claims that his conclusions on the l o g i c a l structure of an argument are v a l i d for argumentation i n empirical science (1970, p.19).) The components of Toulmin 1s well-formed argument are: (1) a claim or conclusion, (2) data, (3) a warrant, (4) a q u a l i f i e r , (5) conditions of reb u t t a l , and (6) a backing. The claim or conclusion i s an hypothesis of which the corre c t -ness or incorrectness i s affirmed or denied. The data are the facts provided to support or contradict the claim. The warrant i s a "rule" or "inference-licence" given to show why the data can be taken as support or contradiction of the claim. Botha describes the warrant as a "bridge-like hypothetical state-ment that has the form: i f X, then Y" (p.20). Different warrants can be seen to confer d i f f e r e n t degrees of force on the claim which they attempt to authorize. The q u a l i f i e r i s therfore b u i l t into the argument to specify e x p l i c i t l y the degree of force which the data confer on the claim through the use of the warrant. In the case where a p a r t i c u l a r q u a l i f i e r i s used, the conditions of rebuttal state the reasons for using -25-i t . That i s , the conditions of rebuttal specify the circum-stances under which the warrant has no authority. F i n a l l y , to know why a warrant i s generally v a l i d , the grounds from which i t derives i t s v a l i d i t y i s s p e c i f i e d e x p l i c i t l y , as a "categorical assertion of f a c t , " by the backing. See Figure 1 for Botha's exemplification of the components of t h i s argument. A well-formed argument requires that a l l of the components of the argument be e x p l i c i t l y presented and related i n a proper way. To ensure that the conclusion of an argument i s correct, the argument must both be well-formed and have correct state-ments presenting the data, backing and conditions of r e b u t t a l . Botha finds that the i n v a l i d i t y of many l i n g u i s t i c argu-ments stems i n part from the lack of proof of the correctness of the backings of grammatical warrants. He concludes that grammatical argumentation f a i l s to q u a l i f y as a confirmation procedure also because "the correctness of many claims about hypothetical data cannot be established beyond doubt" (19 70, p.62). To discuss whether TGP i s v a l i d as a mentalistic theory Botha investigates i n what ways the mentalistic hypotheses of the theory can be v e r i f i e d . F i r s t he outlines the basic types of norms used i n empirical science for the v e r i f i c a t i o n of theories. Then he enumerates the means by which mentalist l i n g u i s t s attempt to v e r i f y t h e i r theories. F i n a l l y , he discusses whether or not the v e r i f i c a t i o n procedures are acceptable from a l o g i c a l and epistemological point of view. The norms of truth used i n v a l i d a t i n g empirical theories -26-DATA: CLAIM/CONCLUSION: Harry was born in Bermuda. WARRANT; -» so, presumably, Harry i s a B r i t i s h subject. CONDITIONS OF REBUTTAL: since unless If a man was born in Bermuda, then he w i l l be a B r i t i s h subject. (i) His parents were al i e n s , ( i i ) He has become a naturalised American. BACKING: On account of the fact that The B r i t i s h and Bermudan statutes and lega l provisions A,B,C, and D specify that a man born i n Bermuda w i l l be a B r i t i s h subject. Figure 1. An example of an argument. Figure from Botha, 19 70. -27-are norms of correspondence, norms of coherence and pragmatic norms. Correspondence norms are based on a concept of truth in which "the truth of a statement depends on whether the state of a f f a i r s referred to by t h i s statement corresponds to what i t asserts about the state of a f f a i r s " (Botha, 1971, p.122). Therefore a statement w i l l be true i f i t " f i t s the f a c t s . " Correspondence norms are thought to be "more fundamental" than pragmatic norms or coherence norms since science i s generally governed by a r e a l i t y p r i n c i p l e . The truth of correspondence norms i s c a l l e d empirical truth since the correspondence be-tween what a theory asserts and the state of a f f a i r s i s deter-mined by experience. Norms of coherence are based on the epistemological thesis that the truth of a theory depends on whether i t f i t s into a body of theory or a system of previously-established s c i e n t i f i c statements. With these norms of truth, a statement i s considered to be true i f i t "coheres with others, and i f these are known to be true" (Botha, 1971, p.123). Botha mentions two problems a r i s i n g with the use of coherence norms. One i s that an ex-p l i c i t c r i t e r i o n i s needed by which to determine whether two theories "cohere." Another problem i s that of determining which of two or more opposing bodies of i n t e r n a l l y consistent theory i s the " v a l i d " t h e o r e t i c a l system. F i n a l l y , pragmatic norms specify a concept of truth wherein the truth of a theory i s r e f l e c t e d by the way i n which the theory guides and stimulates the continuation of s c i e n t i f i c inquiry. Botha states, however, that although s c i e n t i s t s do -28-appreciate theories for t h e i r " f r u i t f u l n e s s , " they do not equate f r u i t f u l n e s s with truth. Consequently pragmatic norms are not advocated by many philosophers of science as conceptions of truth. Botha goes on to describe the modes of argument used by mentalist l i n g u i s t s to v e r i f y t h e i r theories. (His detailed discussion and schematization of these arguments may be found in Chapter 4 of Methodological Status of Tranformational  Generative Phonology (1971).) One mode of argument i s the "how-else argument" i n which l i n g u i s t s argue that t h e i r theory i s v a l i d because there i s presently no other theory which accounts for the data. Botha rejects t h i s argument on a number of counts: i t contains no backing; the norm of truth used i s the "primitive" pragmatist norm; and there i s a p o s s i b i l i t y for other appropriate theories to be l a t e r developed by other scholars. A common mode of v e r i f i c a t i o n i n l i n g u i s t i c study i s tes t i n g the predictions of grammars. Norms of correspondence are used i n t h i s v e r i f i c a t i o n procedure. The warrant for the argument i n t h i s v e r i f i c a t i o n process can be stated i n two ways. In one case, the warrant might state that a theory i s correct i f the data i t predicts are correct. Botha, however, states that a theory which c o r r e c t l y predicts some events does not necessarily c o r r e c t l y describe or represent the structure of the mechanism whose operation r e s u l t s in the predicted events. He c i t e s examples from the history of science and from transformational generative grammar to support his rebuttal (1970, p.130). These examples are instances of competing hypotheses which a l l predicted correct data. He also c i t e s instances where a given hypothesis has been preferred over r i v a l hypotheses even though the former's predictions were incorrect. On the other hand, the warrant could be interpreted as the statement that "unobservable events cause measureable e f f e c t s . " This i s a statement of "causal e f f i c a c y " used i n ontological arguments—that i s , arguments which attempt to prove the existence of some unobservable e n t i t y . Psycholin-g u i s t i c experiments often use ontological arguments. The warrant, as a statement of causal e f f i c a c y , i s l o g i c a l l y i n -v a l i d and i s known as "the f a l l a c y of affirming the conse-quent "(Botha, 1971, p.134). The warrant i s generally of the form " i f an empirical generalization 'g' follows from hypothesis 'h', and i f 'g* i s true, then 'h* i s true." I t i s not v a l i d , from a l o g i c a l point of view, to simply assume that 'h' implies 'g' and *g* implies 'h' without careful q u a n t i f i c a t i o n of the l o g i c a l statement. Other types of arguments which are based on norms of co-herence are c a r e f u l l y outlined by Botha. These arguments obtain t h e i r evidence from numerous sources such as history, psychology, metatheory, neurophysiology, and other f i e l d s of study. Each argument discussed i s refuted as a v a l i d v e r i f i c a -t i o n technique for mentalistic hypotheses. In general i t might be said that the c r i t i c i s m s of these arguments stem from the arguments' f a i l u r e to meet conditions of "well-formedness" mentioned e a r l i e r . (The reader i s referred once again to -30-Chapter 4 (Botha, 1971) for further details.) After demonstrating the conditions d i s q u a l i f y i n g grammatical arguments as v a l i d v e r i f i c a t i o n procedures, Botha amplifies the serious nature of the sit u a t i o n i n two ways. F i r s t , he pro-poses that even i f the arguments using coherence and correspon-dence norms were v a l i d , the supporting evidence of psycholin-g u i s t i c t esting i s "scant" and often "mutually contradictory" (Botha, 1971, p.167). Secondly, Botha's discussions of grammatical argumentation assume that the theory's formal de-vices " c a t e g o r i c a l l y " characterize l i n g u i s t i c competence. A categorical characterization of competence rejects the notion that the speaker's acqusition of language i s instantaneous. Rather, the categorical statement of l i n g u i s t i c competence assumes that language acqu i s i t i o n takes place over a period of time and Requires that mentalistic l i n g u i s t i c theories take into account the order i n which primary l i n g u i s t i c data are used by the c h i l d . Chomsky and Halle specify that the mentalistic assertions . of t h e i r hypotheses are not categorical statements of l i n g u i s t i c competence but are instead "conditional assertions" (1968, p.331) of competence. The conditional assertion of competence r e l i e s on the a p r i o r i assumption that a speaker acquires l a n -guage instantaneously. Chomsky and Halle state that i t would be more d i f f i c u l t to study the categorical statement of a speaker's competence and the mentalistic nature of hypotheses r e s u l t i n g from that statement since the order i n which primary l i n g u i s t i c data are used by the c h i l d would have to be taken -31-into account. Botha then argues that i f there are thus far no v a l i d pro-cedures of v e r i f i c a t i o n for those hypotheses (which he discus-sed) involving the more complex categorical statement of l i n -g u i s t i c mentalism, there could not presently e x i s t any pro- . cedure to test Chomsky and Halle's conditional statement of mental r e a l i t y for TGP. Botha claims to r e j e c t the thesis of mentalism i n transfor mational generative grammar. He does claim, however, that transformational generative grammar as a mentalistic l i n g u i s t i theory i s untestable at present. He suggests that the theory' mentalistic claims might become testable i f l i n g u i s t s would i d e n t i f y and correct the weak methodology of argumentation. In other words, the arguments must meet the conditions of well formedness and the statements of the arguments* warrants and backings must be correct. Consequently Botha goes on to laboriously consider the empirical status of TGP as a non-mentalistic theory. Chomsky and Halle (1968) state e x p l i c i t l y that they wish t h e i r theory of TGP to be an empirical one. They say that even with the assumption that the grammar s h a l l consider deviations from grammaticalness as "correct," the problem to be solved i s one of an empirical nature: a set of formal devices and an evalua t i o n measure must be selected which j o i n t l y meet the empirical condition that the highest valued grammar of the appropriate form i s , i n f a c t , the one selected by the c h i l d on the basis of primary l i n g u i s t i c data (Chomsky and Halle, 1968, p.331). -32-They continue to stress the empirical nature of the theory in saying that such a theory which sp e c i f i e s formal devices and an evaluation procedure can be proven false...by confronting i t with empirical evidence r e l a t i n g to the grammar that a c t u a l l y underlies the speaker's performance. There i s such a grammar, and i t i s an empirical problem to discover it...However d i f f i c u l t i t may be to f i n d relevant evidence for or against a proposed theory, there can be no doubt what-soever about the empirical nature of the problem. Before considering the empirical status of TGP as a non-mentalistic theory Botha describes the aim of a theory which makes no claims for mentalism. Such a theory aims to construct formal devices which can c o r r e c t l y "account f o r " the data which constitute l i n g u i s t i c r e a l i t y . The devices do not need to c o r r e c t l y describe i n some way the mental mechanism underlying l i n g u i s t i c data. Botha approaches the issue of TGP's merits as an empirical theory by o u t l i n i n g the factors which determine and bear an influence on i t s empirical status. One factor which determines that a theory i s empirical i s the fact that i t i s testable. Botha's requirements for a testable theory are borrowed from Hempel (Botha, 1971, p.174). Hempel believes that a theory has " t e s t a b i l i t y " i f e x p l i c i t t e s t implications can be derived from that theory and i f the experimenter can state what would constitute favourable and unfavourable evidence for those test implications. T e s t a b i l i t y would be confirmed by confronting t e s t implications with experimental evidence. Another variable which influences the empirical status of a theory are certain factors which can adversely a f f e c t the theory's t e s t a b i l i t y . Such a factor would arise i n a theory -33-which i s so i n e x p l i c i t l y stated that no test implications can be derived from i t . Another factor i s the p o s s i b i l i t y that i t could be impossible to specify just what evidence—both relevant and r e l i a b l e — w o u l d be able to confirm or disconfirm a theory. In addition, i t i s possible for a theory to be constructed so as to enable i t to be upheld even i n the face of evidence which contradicts i t . In t h i s s i t u a t i o n ad hoc hypotheses are i n -corporated into the theory for the single purpose of protecting the theory against contradictory evidence. In Chapter 5 Botha (19 71) investigates whether the evidence available for some of TGP's formal devices i s relevant and r e l i a b l e . Also, he discusses the inclusion of ad hoc hy-potheses in certain p r i n c i p l e s of the theory. His main con-clusions on the a v a i l a b i l i t y of evidence are, f i r s t , that the evidence i s indeed not available for the hypothesized form and function of the evaluation measure. ( B r i e f l y , the evaluation measure i s a s i m p l i c i t y metric used for evaluating rules. "The 'value' of a sequence of rules i s the r e c i p r o c a l of the number of symbols i n i t s minimal representation " (Chomsky and Halle, 1968, p.335).) TGP often depends on the evaluation measure to play a key role i n l i n g u i s t i c v e r i f i c a t i o n pro-cedures as the warrant of the grammatical argument. Secondly, he finds that Chomsky and Halle (196 8) do not specify what evidence would confirm phonetic representations i f t h e i r composite phonetic features were viewed as being either mental instructions to the a r t i c u l a t o r y apparatus or as representing aspects of vocal t r a c t behaviour. Further, the -34-empirical status of Chomsky and Halle's phonetic representations i s found by Botha to be questionable when the representations are considered to be a description of perceptual r e a l i t y . This follows from what he considers to be i r r e l e v a n t physical e v i -dence, as well as i r r e l e v a n t and unreliable perceptual evidence, which i s offered by Chomsky and Halle to confirm or discon-firm the representations as r e f e r r i n g to perceptual r e a l i t y . Botha also points out several d i f f i c u l t i e s inherent i n develop-ing an i n d i r e c t mode of v a l i d a t i n g phonetic representations on the basis of the notion "systematic theory." This v a l i d a t i o n procedure would involve a l i m i t e d number of data about the placement of primary stress being offered as evidence for the correctness of complex stress contours. The pattern of argu-mentation based on the notion of "systematic theory" i s not proposed and defended by Chomsky and Halle; i t s existence i s only suggested by them and thereafter explored by Botha. Botha discusses four theories i n TGP from the point of view that they contain ad hoc hypotheses which block adverse e v i -dence from applying to them. (These theories are the p r i n c i p l e of the transformational cycle, l i n g u i s t i c universals, the theory of exceptions, and devices accounting for stress phe-nomena. For the d e t a i l s of his discussion the reader i s referred to Botha, 1971, Chapter 5.) The ad hoc hypotheses or "blocking devices" involved i n the four cases he describes serve to block the process of t e s t i n g the theories which i n -corporate them by n e u t r a l i z i n g contradictory evidence. There-fore the theories' t e s t a b i l i t y i s denied and Botha d i s q u a l i f i e s -35-them as empirical theories. Chomsky and Halle f e e l that i t i s quite usual i n empirical science to employ blocking devices to protect hypotheses of great generality. Botha refutes t h e i r attempt to j u s t i f y the use of blocks by saying that the mere use of a p r i n c i p l e in empirical science does not imply that i t i s l o g i c a l l y or ep i s -temologically sound. It i s important to mention that Botha feels that his con-clusions, i f correct, do not "show that TGP i s a completely unempirical approach to the study of phonological properties of natural language" (1971, p.247). He states c l e a r l y that h i s discussion dealt with only a small number of t h e o r e t i c a l de-vices used i n phonological theory. "If some of the hypotheses constituting a theory have a questionable empirical status, i t does not automatically follow that the remainder are also sus-pect with respect to t h e i r t e s t a b i l i t y " (p. 247). 2.2.3 How experiments on r e a l speakers cannot apply to  theories about ideal speaker-listeners Given that doubt has been cast on the v a l i d i t y of TGP as a mentalistic theory and on the empirical status of some com-ponents of TGP, certain ideas behind the requests for experi-mentation outside the ide a l speaker-listener frame are reinforced. Those ideas are, f i r s t , that a l l theories should be tested, since a theory does not have empirical status unless i t i s testable and since an untested theory r i s k s maldevelopment. The second idea i s that a l l theories should be tested and v e r i f i e d i n as many independent ways as possible i n order to increase the theories' generality. -36-The i n d i v i d u a l studying TGP i s therefore greatly tempted to expose various hypotheses to experimentation on the true source of a l l l i n g u i s t i c data, the r e a l speaker. Yet the bothersome obstacle to the v e r i f i c a t i o n of transformational generative grammar's theories through language performance data remains: i n p r i n c i p l e , a theory system which i s based on an i d e a l i z e d speaker-listener i s immune to evidence gained from the performance of a r e a l speaker. A r e a l speaker, r e c a l l , does not meet the conditions of coming from a "'homogeneous'" speech community; i t i s quite possible that he does not know the language "'perfectly'"; i t i s possible that he w i l l not remain uninfluenced by memory l i m i t s , d i s t r a c t i o n s , inattention and n o n - l i n g u i s t i c knowledge and b e l i e f s . Short of hoping that Chomsky and his followers w i l l have a sudden turn-about and either renounce the mentalistic nature of transformational generative grammar or discard the ideal speaker-listener frame-work, one i s i n c l i n e d to s i t back and ponder Botha's systematic dismantling of the theory system and hope for a miracle. 2.2.4 Possible j u s t i f i c a t i o n of experimentation external to  the i d e a l speaker-listener frame However, i t may be possible to reach some compromising so-l u t i o n to the si t u a t i o n which, given the truth of Botha's conclusions on the non-empirical status of some of TGP's hypotheses, w i l l continue to propagate untestable theories. Would the p o s s i b i l i t y of rendering testable some of TGP's hypotheses by experimentation on a r e a l speaker be worth the r i s k of v i o l a t i n g the theory's s t i p u l a t i o n for an i d e a l i z e d speaker framework? The questions to be answered are what can - 3 7 -be l o s t and what can be gained i n such a step towards trans-gressing TGP's metaphysical assumptions and which i s of greater value (or for that matter, which i s of less damage) to the theory. That which would be immediately l o s t by submitting theories of TGP. to v e r i f i c a t i o n procedures external to the i d e a l i z e d framework i s consistency i n dealing with the metatheoretical assumptions of the theory. But t h i s consistency may have a l -ready dissipated. Steinberg (19 75) proposed that Chomsky, without j u s t i f i c a t i o n , maintained formalisms within his sup-posedly mentalistic framework. Botha meticulously demonstrated the varying a b i l i t i e s of hypotheses to be testable and hence empirical i n a theory system which purports to be concerned with the "doubtless empirical problem" of writing grammars. In addition, the rules and p r i n c i p l e s of TGP hypothesized by l i n g u i s t s , the hypothetical grammatical models with apparent descriptive and explanatory adequacy and eventually the meta-physical assumptions of TGP would be submitted to the scrutiny of various independent v e r i f i c a t i o n procedures and so of course could be modified or abandoned altogether. That which could be gained, on the other hand, would be an increase i n the e x p l i c i t n e s s of descriptions of a l l aspects of the theory. This would be achieved by means of independent sources of evidence r e f l e c t i n g back on the theories and forcing t h e i r e x p l i c i t modification. The e x p l i c i t n e s s would also be brought about by means of the need for continuing experimenta-ti o n to develop on the basis of test implications which could -38-only be derived from theories that were more f u l l y s p e c i f i e d . The gain would l i k e l y be a !self-propagating cycle of the development of more e x p l i c i t theory. Testing the hypotheses and modifying them s p e c i f i c a l l y would produce the opportunity to derive further test implications which might, i n turn, r e s u l t in further modification and s p e c i f i c a t i o n of the theory. Ohala points out Chomsky's admission that i t w i l l be necessary to discover conditions on theory construc-tions, coming presumably from experimental psychology or from neurology, which w i l l resolve the alternatives [that i s , alternative hypotheses] that can be arrived at by the kind of speculative theory constructions l i n g u i s t s can do on the basis of the data available to them (J. Ohala, 1970, p.10). His and Halle's most consistent p o s i t i o n , however, i s that for the foreseeable future, the study of language and mental processes w i l l have to be c a r r i e d out at such a l e v e l of abstraction [the abstraction of an i d e a l speaker-listener who instantaneously acquires language] i f i t i s to make s i g n i f i c a n t progress (Chomsky and Halle, 1968, p.332). Chomsky and Halle, however, f a i l to specify the point that TGP must reach before speculative theory constructions can be submitted to independent means of v e r i f i c a t i o n . Here i s another reason to begin t e s t i n g theory constructions, i f only to de-crease the r i s k that the mentalistic theory w i l l develop e r -roneously. 2.3 Independent v e r i f i c a t i o n procedures for phonological rules Ways of approaching the current challenge of v e r i f y i n g the psychological r e a l i t y of generative phonology are reviewed by Fischer-JjzJrgensen (1975). To her, the notion of the r e a l i t y of mentalistic claims presupposes that the speaker somehow "has command of [some equivalent of] the units and rules set up by -39-the l i n g u i s t " (p.224).. She c a l l s t h i s r e a l i t y "functional psychological r e a l i t y " and surveys the types of l i n g u i s t i c data from which inferences can be drawn for the problems of the theory's mental realism. The sources of data which she c i t e s are from various types of l i n g u i s t i c change such as diachronic l i n g u i s t i c data, accomodation of loanwords and a c q u i s i t i o n of f i r s t and l a t e r languages; speech errors, such as s l i p s of the tongue and aphasic disturbances; metrics and rhyme, phonetic puns and l i n g u i s t i c games; orthography; and d i r e c t experimenta-t i o n . Following are some cases where inferences were drawn from various types of l i n g u i s t i c behaviour on the psychological r e a l i t y of certain aspects of TGP. 2.3.1 Direct experiments as a means of v e r i f i c a t i o n of  phonological rules This means of v e r i f y i n g the psychological r e a l i t y of phono-l o g i c a l rules defies the ideal speaker-listener frame of TGP since the experiments are done on non-ideal speaker-listeners with the performance data being considered as d i r e c t l y relevant to the speakers * competence. An experiment involving the use of nonsense words i s r e -ported by Zimmer (1969) . Zimmer states that speakers may not have an awareness of the p r i n c i p l e underlying some phonological r e g u l a r i t y seen in t h e i r language i f the phonological process accounting for the r e g u l a r i t y i s not synchronically productive. Further, he suggests that i t may sometimes be impossible to determine the r e l a t i v e productivity of aspects of l i n g u i s t i c structure. He sets out to test subjects' awareness of three Turkish -40-morpheme structure conditions (MSC's) involved i n Turkish vowel harmony. The p a l a t a l i t y MSC holds that vowels i n bases of words agree i n p a l a t a l i t y (for instance, / i t i / dog and / e t i / meat ). The l a b i a l consonant MSC states that a f t e r /a/, a high vowel agrees i n l a b i a l i t y with an i n t e r v o c a l i c l a b i a l consonant (for example, /karpuz/ watermelon and /avuc/ palm of hand). Several (62) morphemes were found conforming to t h i s MSC and many of them were frequently used i n everyday speech. In addition, 13 uncommon morphemes were found to be counter-examples of the l a b i a l consonant MSC. These morphemes had the form / . . . a . . . C ^ . . . i . . . / where C^ i s a l a b i a l consonant. They can be accounted for by the general l a b i a l i t y MSC. This rule states that a no n - f i r s t vowel in the base agrees with the l a b i a l i t y of the f i r s t vowel i n the base i f both are high vowels—except i f the preceding vowel i s /a/ and a l a b i a l consonant intervenes between the two vowels. The f i r s t t est consisted of pairs of single morpheme Turkish nonsense words. Each pair was designed to test one of the three MSC's mentioned above. The l i s t of words was mailed as a questionnaire to Turkish students at Berkeley with the instructions that they were to pronounce the word pairs aloud and check which word sounded l i k e a possible Turkish word. They were permitted to check both words i f both sounded l i k e possible Turkish words. Twenty-three students r e p l i e d and t h e i r responses were considered. Five of the 16 word pairs were relevant to the l a b i a l con-sonant MSC. Only 23.5% of the responses for these 5 pairs were i n accord with t h i s rule's prediction, while 50.4% of the responses were not those which would have been predicted by the rule and 26.1% of the responses showed no preference for either word in the p a i r s . Two pairs of words were not relevant to any of the MSC's under consideration. Nine pairs were relevant either for the general l a b i a l i t y MSC or the p a l a t a l i t y MSC. Zimmer considered the results of the responses to word pairs which were relevant to these two MSC's together, he said, since t h e i r responses were what one might expect on the basis of r e g u l a r i t i e s i n the harmonic portion of the lexicon. In these 9 word pairs, 88.4% of the responses were those predicted by the MSC's, 4.4% of the responses were i n disagreement with the ru l e s ' predictions and 7.2% were in d i c a t i v e of no preference for either word i n the pai r s . The second test consisted of 18 pairs of nonsense words recorded twice by a native Turkish speaker. The time allowed subjects for producing responses was li m i t e d as the presentation of word pairs occurred with i n t e r v a l s of less than four seconds between the pairs. A second presentation of the l i s t was given. It involved the same word pairs but the order of the words in each pair was reversed. There was a f i f t e e n minute i n t e r v a l between the two tes t presentations during which the examiner and the subject talked of matters unrelated to the t e s t . Sixteen native speakers of Turkish were asked to indicate a unique preference by checking an "(a)" (for the f i r s t word) or "(b)" (for the second word) on a sheet of paper a f t e r l i s t e n i n g to each pair of words. In order to test subjects' awareness of an MSC not related -42-to vowel harmony, Zimmer included two pairs of words where one member had a consonant cluster occurring neither i n Turkish nor i n loan words accepted by Turkish. The answers were in favour of the member with the reduced consonant cluster-i only 5 responses favoured the words with unreduced cl u s t e r s . The r e s u l t s of the second test showed 84% of responses relevant to the p a l a t a l i t y MSC to be i n accord with the rule; 65.6% of responses agreed with the general l a b i a l consonant rule; and 48.1% of responses which were relevant to the medial l a b i a l consonant MSC were i n accord with that r u l e . Zimmer draws two conclusions. F i r s t , despite the great number of non-harmonic loan words i n Turkish, subjects s t i l l showed a "considerable preference" for nonsense words conform-ing to the p a l a t a l i t y and general l a b i a l i t y MSC's. Zimmer ex-plains t h i s preference by proposing that subjects need these phonological rules to use t h e i r language productively. Second, analysis of in d i v i d u a l subject's responses indicated a negative co r r e l a t i o n between the preference of words accounted for by the l a b i a l could be divided into two groups. In each group the subjects tended to ignore one MSC and more often evaluate the word forms on the basis of the other MSC. Furthermore, the two MSC's c o n f l i c t e d with each other. Zimmer believes that subjects preferred to use only one of two overgeneralized— rather than both of two complementary—MSC's concerning the aspect of l a b i a l i t y of vowels i n vowel harmony. The degree of interference between c o n f l i c t i n g MSC's was measured i n terms of the percentage of responses by a given subject where the f i r s t -43-rather than the second MSC was applied. For instance, the interference of the generalized l a b i a l i t y MSC with the l a b i a l consonant MSC ranged from 40% to 100%. To confirm the hy-pothesis of c o n f l i c t i n g MSC's Zimmer suggests t e s t i n g a large number of subjects repeatedly over an extended period of time to see i f the amount of interference of an MSC for a given sub-ject appears stable. Since the re s u l t s showed that one group of subjects more frequently used an overgeneralized l a b i a l consonant MSC while another group preferred to use an overgeneralized l a b i a l i t y MSC (that i s , the general l a b i a l i t y MSC with the "except i f " clause dropped), Zimmer concludes that his subjects had d i f f e r e n t phonological models. In addition, t h e i r phonological models were incorrect as the overgeneralized use of either rule cannot account for a l l of the word forms i n Turkish. Zimmer attributes the erroneous overgeneralizations to the p o s s i b i l i t y that un-productive phonological rules are involved in the MSC's under discussion. In t h i s case he feels that the subjects probably learn l e x i c a l items i n t h e i r f u l l y s p e c i f i e d form. The MSC's in question do not f i l l i n values for incompletely s p e c i f i e d segments. Zimmer f i n a l l y states that A precise statement of the relevant MSC's, which we might want to incorporate in a phonology of Turkish, apparently does not provide a r e a l i s t i c analog of the knowledge int e r n a l i z e d by the native speaker, and of course f a i l s completely to r e f l e c t the d i f -ferences which seem to exis t between speakers (p.320). An experiment done by M. Ohala (1974) involved unusual -44-derivations of e x i s t i n g words. Ohala was not convinced that purely s t r u c t u r a l evidence was s u f f i c i e n t proof that grammars are psychologically r e a l constructs. She attempts to provide experimental evidence for the psychological r e a l i t y of a cer-t a i n segment in underlying forms which does not always appear in surface structures. Ohala states that for a group of forms in Hindi i t i s reasonable to posit an underlying form containing a schwa. Before a medial consonant cl u s t e r the morpheme could drop i t s schwa by a.a-deletion rule of the following form: a -*• 0/VC CV. This system could account for words such as [paksr]/[pakra] (catch/caught), [ p h i s a l ] / [ p h i s l a ] ( s l i p / s l i p p e d ) , etc. Certain suffixes such as -iya (which forms adjectives from nouns) block the application of the -deletion rule and produce such forms as [ka:sar]/ [ks-.sariya] . Ohala explains that a possible alternative way to account for the forms mentioned above i s to use a[a]-epenthesis r u l e . Then the underlying forms would be /pakr/, /kas:r/, etcetera and the [a] would be inserted in the following way: f # 1 0 -*a/VC Cs +C r . Unfortunately t h i s rule would i n c o r -. +iya, r e c t l y i n s e r t schwas in a l l Hindi words with f i n a l consonant cl u s t e r s . It would be too extravagant to mark the Hindi words showing f i n a l culsters as not being receptive to the [a]-epenthesis r u l e . Ohala proposes that evidence for the presence or absence of schwa in underlying forms could be tested by observing the derivations of nonsense words. For instance, the nonsense -45-words [masak] and [mask] could be derived by using the morphemes -6 and - i y a . If the [a]-epenthesis rule was used, the addition of - iya to both stems would produce [masakiya:] for the under-l y i n g forms of both stems would be /mask/. If speakers used a a-deletion rule, the underlying forms would be /masak/ and /mask/ and the responses could be expected to be [masakiya:] (since the a-deletion rule i s blocked by -iya) and [maskiya:] respectively. After finding that one adult native speaker of Hindi gave [masko:] as the response when the oblique p l u r a l morpheme -o was added to both [masak] and [mask] and that the addition of - i y a yielded [masakiya:] and [maskiya:]., Ohala f e l t that the s-deletion rule was l i k e l y used. She sets out to test whether a schwa should be posited i n underlying forms of cert a i n mor-phemes which have clusters at the phonetic l e v e l but which have no alternating word forms containing [a]. The forms used were nouns i n common use in everyday speech: for example, [g S:sla:] (nest), [ke:kra:] (crab). The task was to add -iya to the nouns, producing non-existent but semantically reasonable forms. In a preliminary test seven college-educa-ted native speakers of Standard Hindi added -iya to f i v e nouns. They were verbally given two warm-up examples; of the two, only one stem and i t s derivation contained schwas. Five of the seven subjects consistently gave responses containing schwas ind i c a t i n g that the underlying schwa was psychologically r e a l . The other two gave forms with consonant clusters only. A second test was performed i n which 2 7 native speakers of -46-Standard Hindi added -iya to 30 common Hindi nouns. (Of the 30 words 9 were of no intere s t to the test but were inserted to prevent subjects from answering automatically in a certain pattern.) Subjects were t o l d that adding -iya might r e s u l t i n words that don't exi s t in Hindi but that nevertheless make some sense. "Occasionally" Ohala gave the meaning of a new derivation to prevent the subject from adding -iya automatically to nonsense words. Assuming that subjects would forget t h e i r responses a f t e r an intervening a c t i v i t y , Ohala assigned the subjects a b r i e f task after completing the t e s t . Then, claiming to have misre-corded some items, she asked the subjects once again to derive certain of the words. Two new words were added which have exi s t i n g forms in Hindi containing - i y a . One of the e x i s t i n g Hindi derivations with - i y a did and one did not contain a schwa. The ex i s t i n g derivations were given at the beginning of the second l i s t to see whether the speakers would be influenced by th e i r form. The responses were reocrded throughout both test presentations by noting a 0 i f subjects did not add or retain a schwa. The results showed that most subjects preferred to keep the bases unchanged. To investigate whether subjects were i n f l u -enced by the two cue words, the change i n d i s t r i b u t i o n of 0 and schwa responses was considered for the words following the cue words. The data showed a small change in the incidence of the two types of responses; for the majority of the words, the change was i n the di r e c t i o n of conformity to the cue words * -47-patterns. Chi-square tests of the significance of the d i f -ference between the observed d i s t r i b u t i o n of responses and the expected d i s t r i b u t i o n i f subjects had responded i d e n t i c a l -l y both times showed that subjects' responses were not s i g n i f i -cantly affected by cue words. With the intention of making some int e r e s t i n g points, Ohala te n t a t i v e l y assumes the sample of the responses in the test to be representative of derivations which might be made i n every-day speech. Three subjects' responses predominantly contained a schwa. She groups these subjects into a "a-dialect". Another three subjects' responses were predominatly lacking schwas and were grouped into a "jzJ-dialect". The remaining 21 subjects showed considerable v a r i a t i o n i n t h e i r responses. Ohala observes that for those subjects in the _a_-dialect, the assumption that - iya blocks the a-deletion rule i s wrong. She hypothesizes that the environment of the a-deletion rule i s s t i l l changing from the context mentioned e a r l i e r to a more general environment. The v a r i a t i o n in subjects' responses might indicate, she f e e l s , that the change has not been com-pleted. For the 21 subjects giving mixed responses, -iya sometimes does and sometimes does not block the a-deletion r u l e . There i s no clear-cut evidence for the absence or presence of an underlying schwa in the test's words. Ohala suggests that the words do perhaps have an underlying schwa since the sporadic appearance of the schwa in surface forms could be caused by the fact that the -iya s u f f i x i s being removed as a block to -48-the a-deletion r u l e . Like Zimmer, Ohala observes that "speakers who produce forms which are i d e n t i c a l phonetically may s t i l l have quite d i f f e r e n t grammars" (1974, p.223). J. Ohala (1973) reports t e s t i n g to see i f speakers use ce r t a i n sound patterns productively. Proof of the productiv-i t y of sound patterns would be, according to Ohala, the demon-strated a b i l i t y of speakers to extend the pattern to new words or to e x i s t i n g words i n a new s i t u a t i o n . I f the sound pattern was shown to be productive, Ohala feels that the next step i n the study would be to determine how i t was productive. He considers two possible means of productivity: independent phonological rules such as are hypothesized by Chomsky and Halle (1968) and analogical phonological rules which require e x p l i c i t reference to e x i s t i n g words which manifest a given sound pattern. For example, i f a speaker i s asked to derive a new word using the stem word slave and the s u f f i x - i t y he might produce [slaevlti] . Ohala believes that the speaker may have produced [slaevlti] through an analogy with the word pair sane/sanity. On the other hand the speaker may have used an w 1 independent phonological rule such as V -*• [-tense]/ CVCV which Chomsky and Halle (1968) propose. Ohala tests the productivity of six phonological rules pro-posed by Chomsky and Halle (1968) by giving subjects a word derivation task involving s u f f i x a t i o n . There were 6 3 stem-words. V was used by Ohala, and s h a l l be used i n the discussion of the present study, t o represent V [-stress]" -49-Forty-one of these were test words, that i s , words which Ohala thought would l i k e l y undergo a phonetic change i n the stem when subjects combined them with a s u f f i x . Twenty-two were " f i l l e r " words, or words which were not expected by Ohala to undergo phonetic change when a s u f f i x was added to them. The f i l l e r words were included i n the l i s t to prevent subjects from thinking that a phonetic change i n the stem of the derived word should be necessary. Subects' responses to the f i l l e r words were not included in the analysis. A t o t a l of ten suffixes were used. Some of the suffixes (such as -ment, -hood and -dom) were not expected to tr i g g e r a phonetic change in the stem of the new word. Altogether ten stem-words were to be combined with these three f i l l e r suf-fixes and the responses to t h e i r combinations were also not included i n the analysis. Therefore a t o t a l of 32 out of 62 responses in the s u f f i x a t i o n task in Ohala 1s experiment were not counted in his analysis. After the introduction of a new s u f f i x which was to be added to a following set of stem-words, two examples of the use of that s u f f i x in English were given to the subject. Most of the examples of s u f f i x a t i o n exhibited no phonetic change i n the stem (for example, odd/oddity); two example derivations demonstrated a change in vowel quality in the stem wherein tense vowels became lax (for instance, detain/detention, explain/explanatory); some examples showed two changes wherein a previously unstressed schwa was r e a l i z e d as another lax vowel which carried primary stress ( s o l i d / s o l i d i f y ) . -50-The subjects were 26 Berkeley undergraduate or graduate majors in art or architecture. They were informed that t h e i r help was needed i n constructing an extrapolated or extended dictonary of English and were reminded of the formation of new English words through s u f f i x a t i o n and given an example of i t . The subjects were directed to pronounce the word r e s u l t i n g from the addition of the s u f f i x and word supplied to them. In order to d i s t r a c t subjects' attention from the pronounciation task, subjects were also directed to say what the new word would mean and to t e l l whether they would be l i k e l y to use the word. In administering the t e s t , the subjects were interviewed i n d i v i d u a l l y . The examiner introduced a p a r t i c u l a r s u f f i x by pronouncing i t and s p e l l i n g i t aloud. Then two examples of that s u f f i x ' s use i n e x i s t i n g English words were given. F i n a l l y several stem-words (to which that s u f f i x had not been added before i n English) were presented, one at a time, per-mitting the subject to perform his tasks af t e r the presentation of each stem-word. Ohala does not mention in his report that the time allowed for responding was l i m i t e d . In fact (personal communication), there was no time l i m i t i n which subjects had to respond. The entire test was presented o r a l l y to i n d i v i d u a l sub-je c t s . The subjects' responses were tape-recorded and l a t e r phonetically transcribed. In the case where a subject gave two responses and a preference between them, the preferred form was counted. If no preference was indicated the l a s t form given -51-was counted. In general, the r e s u l t s showed the independent rules under investigation to be unproductive. In addition, as M. Ohala found in investigating new derivations i n Hindi, there was an "overwhelming" tendency for J. Ohala's subjects to keep the stem-words phonetically unchanged. The majority of responses which were relevant to the rule represented by V -»-,. [-tense]/ CVCV retained a tense vowel in the stem. To discuss whether subjects used t h e i r f i r s t r e-sponses as a model for a response that followed, Ohala con-sidered the adjacent words methane+ - i t y and sustain + - i t y as well as define + - i t y and iodine + - i t y . The tetrachoric cor-r e l a t i o n c o e f f i c i e n t measure used with responses for these words showed that the correlations of either both lax or both tense vowels in the stems of each of the above mentioned pairs of derivations were highly s i g n i f i c a n t . Therefore i t appeared to Ohala that subjects were basing t h e i r responses for j. . . sustain + - i t y and iodine + - i t y on t h e i r responses for methane + - i t y and define + - i t y respectively. Ohala concludes that subjects were l i k e l y using analogical phonological rules by using t h e i r f i r s t response as a model for the second in each of the pairs of derivations. The sound pattern of one type of vowel tensing, V -* [+tense]/ Civ (for example, mammal/mammalian) , was shown to be very unproductive with less than 4% of responses ex-h i b i t i n g the predicted tense vowel i n the stem of relevant derivations. -52-Another case of vowel tensing, V•'•*• [+ tense]/ V (for example, algebra/algebraic) showed l i t t l e evidence of produc-t i v i t y with only 13% of responses having the predicted tense vowel. The responses which occurred more frequently were the deletion of the stem-final vowel and the i n s e r t i o n of a g l o t t a l stop between the stem-final vowel and the i n i t i a l vowel of the s u f f i x . OhahaAs impression of these r e s u l t s i s that the 3/(6 alternation might be "the dominant strategy i n the English 'conspiracy' to avoid vowel c l u s t e r s " (p.43) as there are a number of e x i s t i n g English words showing t h i s pattern (such as cholera/choleric, parabola/parabolic). The 's-voicing' sound pattern, wherein .,. ; .... V /s / -»• [+ voice] /[+ tense], V (for example, gymnast/gymnasium) , [- high] was very unproductive with such e l i g i b l e stems as space and fleece where the preceding vowel was produced as a tense vowel. Only 29% of responses that could have exhibited velar d 2 softening, the pattern wherein /k /->-/s/ / I (for example, c r i t i c / c r i t i c i s m ) , did so. Ohala describes the percentage for velar softening as i n d i c a t i v e of "marginal" productivity for the rule, given the conditions of the experiment. Ohala also sought to discover whether the underlying vowel in the second s y l l a b l e s of obtain and pertain are marked for e l i g i b i l i t y for vowel s h i f t when the vowels are laxed. Ordin-a r i l y , underlying tense vowels such as those i n divine and Ohala (1973,.p.44) uses /k / to refer to what Chomsky and Halle c a l l "derivable" /k/ which i s i n opposition to ordinary /k/'s (such as that in keep which does not soften). -53-serene undergo vowel s h i f t only i f they remain tense. Oc-casionally, however, as i n words such as detain/detention and retain/retention, Chomsky and Halle (1968) find i t convenient . to hypothesize that underlying tense vowels are e l i g i b l e for vowel s h i f t even afte r being laxed (by the rule that laxes vowels appearing before consonant c l u s t e r s ) . Ohala presented the stem-words pertain and obtain on two occasions. On each occasion, a d i f f e r e n t example of an ex-i s t i n g derivation was given. The two derived examples (explain/explanatory, detain/detention) had d i f f e r e n t vowels in the second s y l l a b l e s of t h e i r stems. Subjects did not show consistent behaviour in deriving each of pertain and obtain with the suffixes -ion and -atory as they would l i k e l y have done had there been a s p e c i f i e d underlying vowel in each of these stem words. Instead, subjects appeared to make t h e i r derivations according to the pattern provided them i n the example dervations. In a l l , 9 of 26 subjects produced [fc] and [ae] as the second vowel in the new word i n accordance with the example supplied them. Ohala concludes that "It i s clear that the treatment of the vowel i n these words depends not on i t s underlying s p e c i f i c a t i o n but rather on what e x i s t i n g , sur-face alternation the derivation i s patterned a f t e r " (p.44). Perhaps a better way to v e r i f y the influence of example derivations showing appropriate surface alternations would be to give two tests to two d i f f e r e n t groups of subjects. One group would be given examples of derivations relevant to the derivations they were required to make. The other group would -54-not be exposed to these leading examples, Then a check could be made to f i n d the t o t a l number of a l l subjects' responses wherein a phonetic change was made according to the examples provided. This number could be compared with the number of responses with i d e n t i c a l phonetic changes made by subjects who were not supplied with the examples. Then some idea of the influence of subjects' exposure to leading examples of deriva-tions could be gained. From t h i s influence one might draw inferences on the possible use of analogical rules i n the derivations tasks. For instance, i f the subjects who were exposed to examples of derivations i l l u s t r a t i n g given sound patterns happened to produce more responses showing those sound patterns than the alternate group of subjects, one might conclude that the presence or absence of leading examples influenced the phonetic form of subjects' utterances. The next task would be to analyze the data to see i f the influence of those examples was caused by the examples' i l l u s t r a t i o n of surface phonetic a l -ternations or by t h e i r i l l u s t r a t i o n of abstract phonological rules. This might be done by checking to see i f subjects produce vowels of similar quality in the stems of derivations o r i g i n a l l y composed of i d e n t i c a l stems but d i f f e r e n t s u f f i x e s . The d i f f e r e n t suffixes would, of course, serve as an environ-ment in which the same independent phonological rule was hy-pothesized to apply. For example, Ohala checked to see whether the second vowel i n the stem-words obtain and pertain, (which were each administered twice to be combined once with -55-the s u f f i x -ion and once with -atory), underwent the same phonetic change i n each derivation using a d i f f e r e n t s u f f i x . He found that the responses of i n d i v i d u a l subjects for either pertain or obtain did not agree i n the quality of the second vowel. Rather, as was mentioned e a r l i e r , the vowels were i n accordance with the vowel alternation demonstrated i n the example derivations. Ohala concludes that subjects were not aware of an i d e n t i c a l underlying vowel i n either obtain or pertain. This conclusion i s based on the assumption that the same rule would be applied to the sp e c i f i e d underlying vowel in the derivations involving, for example, pertain, producing an i d e n t i c a l surface vowel i n those derivations. Ohala thinks that the treatment of the vowel i n each stem-word was not determined by an underlying vowel and a phono-l o g i c a l r u l e . He believes that subjects instead directed t h e i r attention to the surface alternation of the vowel i n .the example derivations and to the surface vowel of the stem-word. Subjects might then have produced the new derivation by analogi-c a l r u l e . Ohala suggests a possible means of operation for analogical rules: The es s e n t i a l part of the procedure i s finding a form in the lexicon of e x i s t i n g words which can serve as a model for the derivation of the input stem. The model stem and input stem must be divided into d i s s i m i l a r parts and sim i l a r parts....Then the d i s s i m i l a r part of the model stem i s subtracted from the model deriva-t i o n and replaced with the d i s s i m i l a r part of the i n -put stem, i . e . , [fckspl-] i s subtracted from [fckspl«n»bori] leaving [-aenatori] and then replaced by the d i s s i m i l a r part of the input stem [Abt-] which y i e l d s U b t x n a t o r i L (1973, p.46). -56-Ano.ther: means of operation for analogical rules which also involves the surface phonetic forms of appropriate words i s the proportional equation mentioned by Esper (1973, p.40). The proportional equation requires three words, each of which must either have a "formal" or "material" correspondence with one other, of the words in the equation. The form of the equation i s as follows: explain : explanatory :: obtain : ' , and may be read as explain i s related to explanatory in the same way as obtain i s related to . Whereas Ohala considers his hypothetical operation of analogical rules to have mentalistic s i g n i f i c a n c e , Esper believes that the proportional equation has only a f o r m a l i s t i c status. The l a t t e r states that "logicians and l i n g u i s t s may diagram analogies as proportional equations, but i t i s absurd to suppose that such operations are implied by analogical changes in language" (1973, p.153). Returning to the discussion of Ohala 1s test r e s u l t s , i t should be noted that one rule, V tl s t r e s s ] / C + a f f i x , (a part of the stress rules posited by Chomsky and Halle (1968)), was shown to be "highly" productive for certain forms. Ohala feels that the productivity of t h i s rule i n words de-r i v e d with the s u f f i x -ian may be accounted for by the examples of t h i s s u f f i x which exhibited the stress pattern. Yet, the rule was also productive in words derived with - i t y , whose example derivations did not show the stress pattern. In ad- . d i t i o n , two words showing "marginal" productivity of the rule (thermos, Thomas) contain reduced f i n a l vowels and have no -57-alternating word forms with reduced vowels. One word (human) with a reduced vowel i n the f i n a l s y l l a b l e of the stem showed great productivity for the stress rule. I t , however, does have an alternating form i n which t h i s "same "vowel i s unreduced and c a r r i e s primary stress (that i s , humanity). According to Ohala, "This suggests that i f a p a r t i c u l a r vowel i n the stem of a word i s only r e a l i z e d as an unstressed completely reduced vowel, then i t i s d i f f i c u l t for speakers to put primary stress on i t i n derivations" (p.43). Ohala concludes that the productivity of a given rule varies depending on the p a r t i c u l a r rule and on the examples of the sound pattern which are given to the subjects to i l l u s t r a t e the r u l e . He feels that i n t h i s study analogical rules were stronger than independent rules i n forming new derivations. As was mentioned above, however, t h i s hypothesis should be tested when the results of a test such as Ohala's can be com-pared with the responses of a s i m i l a r group of subjects who have not been supplied with examples on which to possibly model t h e i r derivations. Administering such a t e s t as Ohala's and omitting the leading examples might also provide the op-portunity to check for subjects using analogical rules by r e f e r r i n g not to the examiner's examples but to examples which they may have drawn from t h e i r lexicon. Ohala expects that patterns not showing high productivity in t h i s t e s t could be made to show greater productivity i f appropriate examples were supplied to subjects. F i n a l l y , Ohala concludes that the presumed underlying forms i n c e r t a i n -58-words (such as obtain and pertain) appeared to not be what certain l i n g u i s t s would have thought. Steinberg and Krohn's (1975) study attempts to f i n d e v i -dence for or against the productivity of the Vowel S h i f t Rule (VSR) proposed by Chomsky and Halle (1968) . The general nature of the rule can be i l l u s t r a t e d by considering the words extreme and extremity, whose second vowels are [I Y] and [e] respectively. The abstract representation of both vowels i s /e/. Steinberg and Krohn say that " i n the case of extreme, the underlying /e/ undergoes Chomsky and Halle's Dipthongization (e e y) and then t h e i r Vowel S h i f t Rule ( i ~ y t I Y) . In the case of extremity, the underlying /e/ undergoes a laxing rule (e -> e) " (Steinberg and Krohn, 1975, p.234). The VSR i s a general ru l e . This means that i t applies to any underlying phonological representation of a l e x i c a l item having the proper s t r u c t u r a l description, except when the item i s spe-c i a l l y marked to not undergo the r u l e . Steinberg and Krohn believe that since underlying phonological representations are posited to accommodate the application of the VSR, an extensive r e v i s i o n of Chomsky and Halle's underlying forms would be required i f the VSR's generality was proven i n v a l i d . The authors mention three experimental studies which have been done to v e r i f y the VSR and which produced evidence con-tra r y to the theory that speakers i n t e r n a l i z e t h i s r u l e . (One of these was the unpublished study done by J. Ohala (19 7 3) which was discussed above.) Like the experiments of M. Ohala (1974) and J. Ohala (1973), that of Steinberg and Krohn involved novel derivations. In -59-general, the experiment's subjects ~ were required to s e l e c t , within the context of a meaningful sentence, one appropriate s u f f i x of two given them, attach i t to a base word and pro-nounce the novel derived form. The task of selecting the appropriate s u f f i x was intended to d i s t r a c t the subjects' attention from pronouncing the new word. Five vowels which were predicted by Chomsky and Halle to undergo change (when the base word in which the vowel occurs i s combined with a suffix) were used i n the t e s t . The vowels and t h e i r alternations were: [a^] - [i] ( d i v i n e / d i v i n i t y ) , [1"^ ] - [e] (extreme/extremity) \ [e^] - [ae] (sane/sanity), , [5*] - [a] (verbose/verbosity) and [a w] - [ A] (pronounce/pro-nunciation) . Five suffixes were selected: -iC/ - i c a l , - i f y , - i t y , - i s h . A l l but - i s h are predicted by Chomsky and Halle to cause vowel alternations in derivations of the base words. The s u f f i x - i s h was used to see i f i t would t r i g g e r the a l -ternation seen i n Spain/Spanish. The base forms were 26 in number and were composed of or-dinary English words. I f the subjects chose the appropriate s u f f i x , each s u f f i x (except -ian) would be combined with f i v e base forms. An additional base form was used with the s u f f i x -ian. ^ • r • - i ir • i • . i i , . . i Steinberg and Krohn appear to use [I Y] and [T y] interchangeably for the vowel i n the second s y l l a b l e of extreme and [e] and [e] for the vowel i n the second s y l l a b l e of extremity. - 6 0 -The test directions and s t i m u l i were tape-recorded. The 26 items were arranged i n random order. Subjects were further distracted from the pronunciation task by being t o l d that the test's purpose was to investigate s u f f i x preference. Each base word and the two suffixes were introduced after a b r i e f para-graph i n which the l a s t sentence was deleted. The subject was required to repeat the l a s t sentence aloud and " F i l l the blank with the word (for example) maze plus either - i c or - i t y " . The time the subjects had i n which to respond was l i m i t e d as the experiment was pre-recorded. This time l i m i t was not s p e c i f i e d by the authors. One might conservatively estimate that the time l i m i t was probably within four seconds as the subjects, after hearing the stem and the suf f i x e s , heard the f i n a l sentence of the paragraph, a pause, the question "Ready?" and a c l i c k before they were to begin t h e i r response. Even at that point the pronunciation of the new word was delayed by the subject having to repeat the f i n a l sentence before f i l l i n g i n the blank at the end of the sentence with the pronunciation of the new derivation. Before the actual test began, subjects were required to repeat aloud a l l of the base forms and suffixes for the ex-aminer to ensure that they received the intended s t i m u l i and to determine t h e i r pronunciation of the forms. The subjects were native English speakers randomly chosen from introductory psychology courses. They were tested i n -d i v i d u a l l y and t h e i r responses were tape-recorded. Scoring of the responses was done by two people who transcribed the subjects' pronunciation of the base words, the suffixes and the derived words. In Experiment I, 12 male and 12 female subjects p a r t i c i p a t e d The experiment was conducted e n t i r e l y a u d i t o r i l y . Experiment II in addition to the auditory s t i m u l i discussed above, involved stimuli of an orthographic nature. The 8 males and 8 females p a r t i c i p a t i n g i n Experiment II were divided into two groups of 4 males and 4 females each. These groups were submitted to d i f f e r e n t test conditions and were c a l l e d Condition 1 and 2 respectively. The subjects of Condition 1 received, for each paragraph, one card on which the base word and the two suffixes were written. The subjects of Condition 2 received two cards for each paragraph. One card was i d e n t i c a l to that i n Condition 1. On the second card were written the two derived forms. The subjects of Experiment II were instructed to turn over and ( s i l e n t l y ) read the card(s) when a new s u f f i x was introduced at the end of the paragraph. Responses were discarded i f they contained odd stress, a deleted or an added s y l l a b l e , "disjuncture" between stem and s u f f i x , a fabricated s u f f i x or i f no response was given. Steinberg and Krohn do not say why the above types of responses were discarded. Granted t h e i r c r i t e r i a for d i s q u a l i f y i n g cer-t a i n responses leaves the data with a much simpler nature. It can then be analyzed into three categories: responses with phonetic changes i n the stem predicted by Chomsky and Halle, responses with phonetic changes not predicted by Chomsky and Halle, and responses with no phonetic change. The discarded -62-responses may have been considered as 'accidents' since, of the 50 discards out of 504 responses in Experiment I, 29 were attributed to 1/6 (that i s , 4) of the subjects. In Experiment I I , only 3 of the t o t a l 16 8 responses were discarded. (The t o t a l numbers of responses mentioned above do not include responses for the s u f f i x -ish.. They were not included i n the f i n a l analysis because only 1 of 120 responses in Experiment I and 1 of 80 responses in Experiment II showed a phonetic change in the stem.) It i s possible that many of these responses are not 'mistakes' but are the best answers the .subjects;could have produced. In t h i s sense, they should have been considered as v a l i d data along with the responses showing predicted and un-predicted changes and no change i n the stem. The r e s u l t s of both experiments w i l l be considered together since the authors report that the difference between the d i s t r i -bution of responses for the two Experiments was s t a t i s t i c a l l y s i g n i f i c a n t . Of the t o t a l v a l i d responses for the two Experiments, 3.5% of the responses showed vowel changes i n agreement with the VSR. An additional 6.6% of the responses showed phonetic changes not in accord with the VSR. A trend-of no change in the vowels of the derived words was most predominant — 89.9% of responses f i t in t h i s category. There were 2 7 responses predicted by the VSR and 20 of • these involved <the base vowel [a Y] . Of the 51 unpredicted vowel -63-changes, 29 were made i n response to base items containing the vowel [ T Y ] . Steinberg and Krohn report that the frequencies of the predicted responses to [a Y] and of the unpredicted responses to [ i J J are s i g n i f i c a n t l y greater than the frequencies of predicted and unpredicted responses to the other base vowels. The 2 7 predicted vowel changes occurred more frequently with the suffixes - i c (13 times) and - i c a l (5 times). The difference i n frequency between - i c and - i c a l i s not s i g n i f i c a n t ; however, the difference between -ic_ and the other suffixes (which were i n -r volved 3 times each with predicted vowel changes) i s s t a t i s t i c a l -l y s i g n i f i c a n t . The authors suggest that the s i g n i f i c a n t s u f f i x differences are not due to an e f f e c t of the - i c s u f f i x alone as most of the -ic_ responses occurred in conjunction with the base vowel C a?] . Since a l l of the -ic_ responses with [ aY] base ; vowels were responses to a single word, sapphire, they say that "the p o s s i b i l i t y remains that the observed differences are 1 due...to the e f f e c t of some id i o s y n c r a t i c feature of that par-t i c u l a r word" (p.250). Steinberg and Krohn conclude that the findings of t h e i r study indicate that vowel alternation as predicted by the VSR i s "largely non-productive." Therefore, i f one s t i l l assumes that speakers have rules to account for vowel alternations, the "VSR accounts only for exceptions, i . e . , to the creative pattern, which we state here; there i s no (productive) vowel change in derived forms" (p.252). They j u s t i f y t h i s conclu-sion i n saying that the notion of the productivity of a rule " i s e s s e n t i a l for distinguishing between generative phonology -64-(the creative generativity of l i v i n g language) and etymology." Therefore they think i t should be assigned an important r o l e in judging the r e g u l a r i t y of a r u l e . They also conclude that since vowel alternation was seldom produced and since the VSR's prerequisite rule of vowel laxing seldom appeared to apply (in the predicted t r i s y l l a b i c en-vironment and i n base forms taking the -ic_ s u f f i x ) , the v a l i d i t y of claims for the generality and the psychological r e a l i t y of the VSR i s highly dubious. Steinberg and Krohn go on to question whether alternations of sound patterns e x i s t i n g i n the language are to be accounted for at a l l by rules. They suggest that i t i s possible instead that representations of both the base and the derived form are l i s t e d i n the lexicon and that no rules are involved i n vowel alternations of e x i s t i n g related word forms. F i n a l l y , they discuss and dismiss some potential objections to the experiment and t h e i r conclusions. One objection might be that phonological rules do not operate for novel derived forms as they are meaningless. Steinberg and Krohn refute t h i s c r i t i c i s m by pointing out what they consider to be e v i -dence for the meaningful nature of the derived forms. They point out, f i r s t , that subjects generally selected the s u f f i x which was appropriate to the context of the sentence. Secondly, subjects generally assigned correct stress to the novel deriva-tions. The l a t t e r argument i s perhaps weak. Speakers might e a s i l y be shown to assign English stress patterns i n attempting to pronounce foreign phrases which were meaningless to them. -65-One might also object that a non-formative word boundary appeared between the base form and the s u f f i x when the VSR was supposed to operate, blocking the application of the r u l e . However, the authors point out that i f t h i s were the case, the Main Stress Rule would also be blocked. The Experiments' res u l t s show that i n general, stress was c o r r e c t l y assigned to the new derivations. 2.3.2 Evidence from other sources which i s relevant to the  findings of d i r e c t experimentation Zimmer (1969) and M. Ohala (1974), in t h e i r attempts to d i r e c t l y t e s t speaker's awareness of phonological rules, both found that speakers did not a l l appear to use the same phono-l o g i c a l rules. Sherzer (19 70) was writing a s t r u c t u r a l de-s c r i p t i o n of a word game played by the Cuna Indians of Panama. Like Zimmer and M. Ohala, he also concluded that speakers of a same d i a l e c t had d i f f e r e n t ;phonological models. He observed that some models might have been, from a l i n g u i s t i c point of view, "better" or "more correct" than the others. Zimmer also found that his speakers' rules for the aspect of l a b i a l i t y i n vowel harmony were not "correct." In any experiment which attempts to test subjects' awareness of rules, the examiner should not l i m i t h i s expectations to finding, as Chomsky and Halle would lead one to believe, that speakers have i d e n t i c a l sets of phonological rules and that - . these rules are as e f f i c i e n t as those, of the l i n g u i s t i n ac-counting for phonological data. Evidence which may be useful in explaining speakers' d i f f e r e n t and "incorrect" phonological -66-models i s reported by Wang, Chen and Wang, and Hsieh. Wang (1969) surveys diachronic l i n g u i s t i c data for phono-l o g i c a l change in many languages and concludes that there are "cross-currents of sound change vying for the phonological future of every word i n the lexicon" of a language (p.23). If the sound changes are not considered over a long enough time, he doubts whether one could t e l l i f the tendencies toward change are sporadic, persistent, old, new or receding. In addition Wang hypothesizes that rules for phonological change, while being phonetically abrupt, would be l e x i c a l l y d i f f u s e . In other words, the change would occur abruptly for a given word but might not simultaneously apply to a l l of the appropriate words i n the lexicon. Another inference drawn from the data he considers i s that phonological rules may often be dependent on morphological and syntactic factors. He suggests f i n a l l y that understanding the "complex dynamic situation'' of competing cross-currents of sound change w i l l require the "careful analysis each of the various interwoven/factors—the p h y s i o l o g i c a l , the str u c t u r a l , the s o c i e t a l , and yet others" (p.24). Chen and Wang (1975) consider diachronic data from Chinese, English and Swedish as well as evidence from language ac q u i s i t i o n to describe how a sound change implements i t s e l f . Their claim i s that t h i s implementation occurs through l e x i c a l d i f f u s i o n of the sound change; the change propagates gradually from morpheme to morpheme. They also comment on the "actuation" of a sound change, or why i t assumes a p a r t i c u l a r form and follows a schedule. Chen and Wang's source of evidence for t h i s problem - 6 7 -i s consonantal a t t r i t i o n in Chinese d i a l e c t s and Indo-European languages. They claim that the actuation of a phonological process i s to be sought i n the "inherent constraints of the physiological and perceptual apparatus of the language user" (p.255). They provide experimental evidence (both perceptual and physiological) i n support of t h e i r theory. Chen and Wang reveal that the aims of t h e i r study were not only to prove a certain point but to exemplify the philosophy of experimental research in l i n g u i s t i c study. They believe that t h e i r approach proved "more rewarding than the pursuit of •explanatory adequacy' i n terms of a - p r i o r i s t i c and s t e r i l e notions of s i m p l i c i t y and economy" (p.279). The "pursuit of 'explanatory adequacy'" which they mention refers of course to the research strategy of Chomsky and Halle (1968). The experi-mental research performed by Chen and Wang i s the type of a l -ternative approach l i n g u i s t s such as Ohala (1974) and Lindblom (19 71) have in mind when they c r i t i c i z e the poor methodology and explanatory power of Chomskyan theory. In addition to h i s t o r i c a l data as a source of evidence from l i n g u i s t i c change i s the data available in studying the phono-l o g i c a l development i n children acquiring language. In order to support Wang's concept that sound change i s l e x i c a l l y gradual Hsieh (19 72) c i t e s c h i l d language acqu i s i t i o n as a data source with two advantages. He believes that children's phonological development provides cases of sound change that are both com-., pleted within a period of a few years (and so are practicably observable) and that are immune to borrowing from other diaects. -68-2.4 Conclusions 1. The experimental v e r i f i c a t i o n of some of TGP's hypotheses, which requires v i o l a t i o n of the ide a l speaker-listener frame, can be j u s t i f i e d . 2. Certain phonological rules formalized by Chomsky and Halle (1968), at least in given experimental conditions, are not productive (J. Ohala, 1973; Steinberg and Krohn, 1975). The v a l i d i t y of claims for those r u l e s ' psychological r e a l i t y and generality i s therefore questionable. 3. Different speakers have d i f f e r e n t phonological models (according to evidence from d i r e c t experimentation by Zimmer (1969) and M. Ohala (1974) and from observations of a l i n g u i s t i c game by Sherzer (1970)). These phono-l o g i c a l models, furthermore, are not always "correct" from the point of view of an economical l i n g u i s t i c description. 4. Phonological change diffuses slowly across the lexicon (according_to evidence presented by Wang (1969), Chen and Wang (1975) and Hsieh (1972)). 69 Chapter III METHOD 3.1 Aims of the investigation It was mentioned i n the introduction that one of the ob-j e c t i v e s of t h i s study i s to v e r i f y the psychological r e a l i t y for a given group of speakers of some of Chomsky and Halle's ,(1968) phonological rules. This aim might be f u l f i l l e d by providing subjects with the opportunity, under d i f f e r e n t t e s t conditions, to produce novel utterances and by checking to see whether the phonetic forms of those utterances r e f l e c t the use of certain phonological rules. It was pointed out that the phrase "psychological r e a l i t y of phonological rules" i s assigned d i f f e r e n t senses by d i f f e r e n t l i n g u i s t s . In the experimental studies mentioned e a r l i e r , a rule (or sometimes an underlying phonological representation of a word) was considered to be part of the speaker's mental me-: chanisms underlying h i s speech production and perception i f the rule was used in the speaker's generation of responses i n the te s t s i t u a t i o n . The 'use' of a given rule i n past studies has been based on the assumption that the appearance of a phonetic form (which i s predicted by the application of that rule to some underlying representation) implies that the speaker em-ployed the ru l e . The studies previously discussed referred to the psychological r e a l i t y of a rule for i n d i v i d u a l speakers and also groups of speakers. The extent of the rule's use i n the te s t s i t u a t i o n , .(or in other words, the percentage of times that the rule was used with appropriate s t i m u l i ) , i s c a l l e d -69--70-the rule's productivity for the t e s t . However, a minimum amount of productivity which would q u a l i f y the rule as being psychologically r e a l either for one or for many speakers, was never s p e c i f i e d . It i s important to decide upon a s p e c i f i c amount of productivity to be demonstrated by a rule in order for i t to q u a l i f y as being psychologically r e a l . This minimum productivity i s ultimately needed to judge the v a l i d i t y of claims for including the rule i n a mentalistic grammar for a s p e c i f i c language. Some of the rules whose psychological r e a l i t y w i l l be tested in t h i s study are considered to be general rules by Chomsky and Halle (1968) and so would be expected to demonstrate a high degree of productivity before t h e i r inclusion i n a grammar i s j u s t i f i e d . Other rules which w i l l be considered i n t h i s investigation are not described by Chomsky and Halle as general rules. A s p e c i f i c c r i t e r i o n of productivity, j u s t i f y i n g the inclusion of a rule i n a grammar, might depend on such considerations as the number of speakers, for example, an i n d i v i d u a l , a small group within a d i a l e c t , a l l the members of a d i a l e c t , etc., for which the grammar was meant to be representative. In the present study, discussion w i l l center on whether the use of an "independent phonological r u l e " (such as those hy-pothesized by Chomsky and Halle) can be assumed i f the speaker produces a response predicted by the hypothetical application of that rule to some, underlying representation of the word. Also to be discussed i s whether the use of the independent phonological rule can be refuted (as was done by J. Ohala, 1973) -71-i f i t can be shown s t a t i s t i c a l l y that the forms of the speakers' responses have high correlations with the forms of adjacent responses or of recent derivations. In addition, the p o s s i b i l i t y for the application of a rule (other than that hypothesized by Chomsky and Halle) to r e s u l t i n the response predicted by them w i l l also be considered. Recall the notion discussed by Botha (1970): the fact that a theory i s able to cor-r e c t l y predict data i s not s u f f i c i e n t evidence for the proof of that theory. He believes that i t i s no more than a coincidence that at a given moment there i s only one theory that "works" and that no alternative can be proposed. Steinberg and Krohn (19 75) appear to distinguish between the "psychological r e a l i t y " and the " v a l i d i t y " of the Vowel S h i f t Rule for t h e i r group of subjects. The rule was shown to be "largely non-productive" i n the responses of th e i r subjects. Steinberg and Krohn concede .that the rule may be psychologically r e a l for t h e i r speakers. They deny, however, that the rule has "generality." They state that "C&H's [Chomsky and Halle's] claim that the VSR i s a psychologicall r e a l and general rule i s . . . h i g h l y dubious" (p.252). Judging from t h e i r concession that speakers may have such rules as the VSR to account for vowel alternations which are exceptions to the more general creative pattern of no vowel change i n derived forms, the authors probably mean that Chomsky and Halle's claim that the VSR i s psychologically r e a l as a general rule i s , "highly > dubious." At t h i s time i t i s impossible to n o n - a r b i t r a r i l y state in -72-advance, for the purpose of t h i s study, a minimum amount of productivity of a rule which would q u a l i f y that rule as being psychologically r e a l for a group of speakers and as being v a l i d l y assigned the role of a general or 'less-than-general 1 rule i n the grammar for a group of speakers. 3.2 Experimental plan In general, subjects were required to combine, without the use of pencil and paper, English stem-words with suffixes i n order to create a new word ;and to thereafter pronounce i t aloud. The subjects' spoken responses provided data on the frequency of production of c e r t a i n sound patterns i n novel utterances. The data should permit one to i n f e r whether certain types of rules were used i n the production of the new words. The purpose of the p i l o t experiments was to check several aspects of the methodology of the proposed main experiment. The problems encountered with certain methods and the conclu-sions drawn from the results of the p i l o t s w i l l eventually be discussed. In each of the two p i l o t experiments (called P i l o t I and P i l o t II> respectively) one group of subjects (Group a 1) were provided with two example derivations for each s u f f i x . The example derivations demonstrated the use of the s u f f i x i n English. Some of the examples exhibited phonetic changes in Group a s h a l l refer to Group a' and Group a" of P i l o t I and P i l o t II, respectively. Group b s h a l l likewise correspond respectively to Group b' and Group b " of P i l o t I and P i l o t I I . -73-the stem and/or i n ;the s u f f i x of the derivation and some did not. Another group of subjects (Group b) i n each of the p i l o t experiments was provided with no derivations exemplifying the use of the s u f f i x e s . The arrangement of the experiments into groups and the contents of each group's experiment i s outlined in Table I. In the main experiment, one group of subjects (Group A) were provided, for each s u f f i x , with one example derivation that showed no phonetic change, as well as one example demon-str a t i n g a possible phonetic change, i n t h e i r stems and/or su f f i x e s . Often the phonetic changes i l l u s t r a t e d by the example derivations were those predicted by Chomsky and Halle to occur when an hypothesized phonological rule was applied to an hy-pothesized underlying representation. Ohala (19 73) hypothesized that rules showing low productivity in his test could be made to show greater productivity i f ap-propriate example derivations were supplied to the subjects. Therefore some of the example derivations used by Ohala were substituted with more appropriate examples i l l u s t r a t i n g the phonetic changes predicted by Chomsky and Halle to occur with the use of t h e i r hypothesized rules. These more appropriate examples were included i n the l i s t of examples supplied to the subjects of Group A i n the main experiment. 3.3 Subjects In a l l , 41 subjects p a r t i c i p a t e d in the studies. As can be seen i n Table I, a t o t a l of 9 subjects p a r t i c i p a t e d i n the -74-Experiment: P i l o t I P i l o t II Main Group a' b' a" b " A B Contents: Practice task yes yes yes yes yes yes example, derivations with phonetic change yes yes yes yes yes yes example derivations without phonetic change yes yes yes yes yes yes Experimental task example derivations with phonetic change yes no . yes no - yes no example derivations without phonetic change yes no yes no yes yes Subjects: Number of female 0 3 1 1 13 10 Number of male 3 0 1 0 4 5 Total number 3 3 2 1 17 ; 15 Table I. Arrangement of the experiments into groups, the contents of each experiment and the number of subjects p a r t i c i -pating i n each experiment. -75-p i l o t experiments and 32 subjects took part i n the main experi-ment. Twenty-eight of the t o t a l number of subjects used i n the experiments were females and 13 were males. Table I also shows the numbers of female and male subjects p a r t i c i p a t i n g i n each experiment. It was intended that the subjects p a r t i c i p a t i n g i n the ex-periments should meet the following q u a l i f i c a t i o n s : (a) they should have been born and/or have l i v e d most of t h e i r l i v e s i n B r i t i s h Columbia; (b) English should be t h e i r native language; (c) they should have taken no l i n g u i s t i c s courses; (d) they should be working towards or have completed a Bachelor of Arts or Bachelor of Education degree at one of three u n i v e r s i t i e s i n the lower mainland of B r i t i s h Columbia. Some of the subjects did not f u l f i l l a l l of the above q u a l i f i c a t i o n s . F i r s t of a l l , with respect to q u a l i f i c a t i o n (b), three subjects reported that they f i r s t began to learn another language simultaneously with English. Before the subjects began primary school, however, English was being used more f r e -quently than the alternate language. The subjects reported at the time of the experiments that they possessed only minimal f a c i l i t y i n the l i s t e n i n g and speaking s k i l l s of the other language. Secondly, ten subjects were not students within Arts or Education F a c u l t i e s . Five subjects were studying i n a Faculty of Science, two were studying in a Faculty of Commerce, one was studying i n a Faculty of Physical Education and Recreation and two were studying in a Faculty of Fine Arts. -76-Thirty-nine of the t o t a l 41 subjects were within the age bracket 18-29 years. The remaining 2 subjects were in t h e i r mid-forties. A l l 41 subjects had been exposed to the study of a language other than English. Altogether, 20 subjects had studied a second language i n secondary school only and 21 had studied another language at university as well. In the main experiment 7 subjects i n Group A and 8 subjects in Group B had studied a second language i n secondary school only. Ten subjects i n Group A and 7 i n Group B had studied a second language i n secondary school as well as at university. F i n a l l y , i n ad- . d i t i o n to the 3 subjects mentioned e a r l i e r who had begun to ; learn English simultaneously with another language, 2 subjects in Group A of the main experiment had been exposed to a l a n -guage other than English at home some time during t h e i r l i v e s . 3.4 Composition of the word l i s t s The l i s t of word forms used:.in a l l of the experiments con-s i s t e d of English words (henceforth c a l l e d stem-words to d i f - -ferentiate- from derivations), suffixes and derivations (com-posed of stem-words and s u f f i x e s ) . The t o t a l l i s t of these forms comprised two successive s u b - l i s t s : a section of stem-words, suffixes and derivations which served as s t i m u l i for a practice task and a l i s t comprising the experimental task s t i m u l i . 3.4.1 Practice task stimuli The practice task was included i n both the p i l o t and main experiments in order to give subjects practice at deriving new -77-new words using e x i s t i n g stem-words and su f f i x e s . Ohala's format, wherein a s u f f i x and two example derivations precede the presentation of a group of stem-words, was used. The responses for the practice task were not to be included i n the analysis of the responses from the experimental task. The l i s t of word forms used as stimuli i n the practice task consisted of 9 stem-words, 3 suffixes and 6 e x i s t i n g couples of stem-words and derivations. Only 3 of the 9 stem-words were greatly expected to undergo a phonetic change in the stem when they were combined with a s u f f i x . These were located towards the end of the practice task i n an attempt to proceed from easier to more d i f f i c u l t derivations. Not a l l of the 6 example derivations showed a phonetic change i n the stem. This was arranged to prevent subjects from expecting a l l derivations to either undergo or not undergo a phonetic change. Only two of the examples exhibited any phonetic change. The two examples industry/industrious and angel/angelic both showed a change i n the location of stress and also a change in the quality of one and two vowels, respectively. Two example derivations were used with each s u f f i x . The pa r t i c u l a r stem-words to be combined with the s u f f i x - i c and -ous were expected to possibly undergo a phonetic change. The inclus i o n of stem-words l i k e l y to undergo changes when derived with suffixes i n the practice task was arranged to introduce subjects to the notion that a s u f f i x being i l l u s t r a t e d by an example showing a phonetic change in the stem did not -78-require the derivation task to necessarily involve a phonetic change. The l i s t of word forms comprising the practice task i s given i n Appendix I. 3.4.2 Experimental task s t i m u l i The l i s t of stimuli i n the two p i l o t s was i d e n t i c a l to that used by J. Ohala (1973) and i s given i n the Appendix. Although there were 60 d i f f e r e n t stem-words i n the l i s t , three of these (obtain, pertain and trade) were stated twice, making a t o t a l of 6 3 stem-words. These stem-words were arranged into twelve groups. There was one s u f f i x available to be combined with the stem-words i n each group. (Although there were only 10 d i f f e r e n t suffixes i n Ohala"s l i s t , two of the suffixes (-ic and -ity) were l i s t e d twice, making a t o t a l of 12 suffixes available for stimuli to accommodate the 12 groups of stem-words.) In addition, the l i s t contained 20 d i f f e r e n t examples of e x i s t i n g stem-words and derivations. Two of these (odd/oddity and Darwin/Darwinian) were each given twice, bring-ing the t o t a l number of example derivations to 22. This meant that 10 groups of stem-words were exemplified with two example derivations while two groups were i l l u s t r a t e d with only one example. Twenty-two of the stem-words and 3 of the suffixes (which were combined with an additional 10 stem-words) i n Ohala 1s l i s t were described by him as " f i l l e r " words and " f i l l e r " s u f fixes, respectively. Ohala did not expect derivations i n -volving these word forms to show any phonetic change and did -79-not include the responses for these stimuli i n his analysis. The purpose of including these stimuli was to prevent subjects from believing that the addition of a s u f f i x necessarily r e -quired a phonetic change i n the stem. In addition, 8 of the 20 di f f e r e n t example derivations showed a phonetic change in the stem. One example (explain/explanatory) showed only a vowel change. One showed only a consonant change (fuse/fusion). Another showed change i n a vowel and i n a consonant (detain/  detention). Five examples demonstrated a change from an un-stressed schwa to another lax vowel which was stressed (stupid/  stup i d i t y, Darwin/Darwinian, s o l i d / s o l i d i f y , a r t i s t / a r t i s t i c , r e a l i s t / r e a l i s t i c ) . One example derivation demonstrated a change from an unstressed schwa to another vowel with primary stress as well as a change in a consonant (music/musician). Although the l i s t of stimuli used i n the main experiment included the stem-words and suffixes mentioned above, the l i s t of example derivations was not i d e n t i c a l to Ohala's l i s t . In order to exemplify each s u f f i x with one derivation showing a phonetic change i n the stem and one derivation showing no change, 9 examples i n Ohala*s l i s t were replaced by d i f f e r e n t example derivations. Four of the new examples demonstrated a change i n vowel quality. The examples sane/sanity and r a t e / r a t i f y were adopted to i l l u s t r a t e vowel laxing, while mamma1/mamma1ian and algebra/algebraic were intended to exemplify a vowel tensing r u l e . One example, c r i t i c / c r i t i c i s m , demonstrated the velar softening r u l e . Four of the new examples, Kant / Kantian, f a l s e / f a l s i f y , r e b e l / r e b e l l i o n and graph/graphic were included -80-to ensure that each s u f f i x was exemplified by one example derivation exhibiting no phonetic change: A l l of the stimuli mentioned above are l i s t e d i n Appendix I. It should be noted that the combination of two stem-words and t h e i r suffixes (between + - i t y and probe + -ity ) i n the experimental task could r e s u l t i n words that ex i s t i n English. The word betweenity i s r a r e l y used. According to Oxford English Dictionary, i t i s pronounced with a tense vowel i n the second s y l l a b l e p o s i t i o n . The word probity i s c i t e d by Chomsky and Halle (1968, p.181) as an exception to the words to which the vowel laxing rule V [-tense]/ CVCV applies. These stem-words were retained in the set of forms to be combined with the s u f f i x - i t y in order to see how subjects' pronunciations of t h e i r derivations compared with the e x i s t i n g pronunciations. 3.4.3 V e r i f i c a t i o n of the appropriateness of combining the  stem-words and suffixes Studies by Nicholson (1916) and Marchand (1960) demonstrate that ex i s t i n g English derivations show trends for stem-words and suffixes to have been r e s t r i c t e d i n t h e i r combinations according to such factors as the language of t h e i r o r i g i n and t h e i r grammatical function. For instance, Marchand states that the s u f f i x - i t y forms abstract substantives from i t s combination with adjectives (most of which are of Latin origin) and with nouns. The grammatical function and the language o r i g i n of each of the stem-words and the suffixes i n both the practice task and the experimental task were examined i n preparation for th e i r -81-being retained in the l i s t s . This information was obtained from the works of Nicholson and Marchand mentioned above, from The  English Word Speculum (Dolby and Resnikoff, 1967), from the Oxford Dictionary of English Etymology (1966) and from Webster's English Dictionary. It was then possible to v e r i f y whether the stem-words' c h a r a c t e r i s t i c s were appropriate for the in d i v i d u a l d e r i vational tendencies of the suffixes with which they were to be combined. Most of the stem-words seemed to match the q u a l i -t i e s required by th e i r s u f f i x e s . Some of the stem-words, how-ever, had seemingly inappropriate c h a r a c t e r i s t i c s which could have influenced the form of th e i r derivations with the suffixes s p e c i f i c a l l y assigned to them i n the experiments. Some stem-words' grammatical functions were inappropriate for the s p e c i f i c s u f f i x . The verbs define and sustain were combined with the s u f f i x - i t y which forms substantives from nouns and adjectives. Less than 10 derivations comprised of a verb stem and - i t y were found i n Dolby and Resnikoff. The verb trample was combined with the s u f f i x - i f y which forms verbs from i t s combination with nouns and adjectives. Only two derivations formed with - i f y were found (1 i n Marchand and 1 in Dolby and Resnikoff) to contain a verb as a stem. The noun sapphire was combined with the s u f f i x -atory. This s u f f i x a c t u a l l y has the form -ory (according to Marchand and the Oxford English Dictionary of Etymology) and has been combined with verbs of Latin o r i g i n ending i n -ate. In some instances the inappropriate c h a r a c t e r i s t i c of the stem-word arose from i t s language o r i g i n . The noun Buddha, of -82-Sanskrit o r i g i n , may not be an appropriate stem for the s u f f i x -Ic since a l l the words in Dolby and Resnikoff ending i n V + - i c were of Greek or Latin o r i g i n . The noun trade, of Germanic o r i g i n , was combined with the Latin s u f f i x -ian(us). The s u f f i x -ian i s most often combined with proper nouns (Marchand) or, as was observed i n Dolby and Resnikoff, with nouns only of Greek or Latin background. The semantic features of some of the stem-words could possibly serve as a problematic source for t h e i r derivation with certain s u f f i x e s . Apple i s an inanimate noun. It was combined with the s u f f i x -hood which was observed i n Dolby and Resnikoff to be predominantly combined with animate nouns. Only one d e r i -vation with -hood, whose stem was inanimate, was l i s t e d i n Dolby and Resnikoff. The abstract noun joy was combined with the s u f f i x -dom. The s u f f i x -dom appeared in Dolby and Resnikoff to be derived only with noun stems which had the semantic qu a l i t y 'animate.' A l l of the stem-words mentioned above were retained i n the experimental task's word l i s t . The responses to these stem-words and t h e i r suffixes were observed while keeping in mind the possibly inappropriate c h a r a c t e r i s t i c s discussed. 3.5 Administration and recording of the experiments 3.5.1 The s t i m u l i The type of s t i m u l i given to each group of subjects i s outlined i n Table I. The subjects of the two p i l o t s were given the word forms of the practice task and those i n J . Ohala's l i s t of s t i m u l i . In each p i l o t , however, there were two groups of -83-subjects who were given d i f f e r e n t test conditions. One group of subjects (Group a) in each p i l o t were given a l l of the ex-ample derivations in Ohala's l i s t of tes t s t i m u l i . The other group of subjects (Group b) i n each p i l o t heard no example d e r i -vations other than those in the practice task. For them, Ohala's example derivations were omitted. As was mentioned e a r l i e r , the subjects of the main experi-ment were also members of one of two groups (either Group A or Group B). Each group was administered the entire practice task. The experimental task's word-stems and suffixes were taken from Ohala's l i s t . When i t came.to example derivations, however, Group A was given two—one example demonstrating phonetic i change and one showing no phonetic change—to i l l u s t r a t e each s u f f i x . Group B, on the other hand, received only one example derivation, showing no phonetic change, for each s u f f i x i n the experimental task. It should be noted that J. Ohala repeated only one of a su f f i x ' s example derivations upon presenting that s u f f i x to subjects for the second time. In the f i r s t condition of each of the experiments described here (that i s , in Groups a", a " , and A) where two example derivations were supplied with each s u f f i x , both of the s u f f i x ' s examples were repeated to the subjects when the s u f f i x was presented on a second occasion. This was done in an e f f o r t to keep the conditions of the experi-mental task the same for each set of stem-words with which the repeated suffixes were to be combined. -84-3.5.2 Directions to the subjects The directions to each group of subjects may be found i n Appendix I. Except for a few a l t e r a t i o n s , they are the directions c i t e d by J. Ohala (1973) which were used i n his t e s t . In b r i e f , the directions for a l l of the studies had the following points in common. Subjects were t o l d that the examiner needed t h e i r help i n forming an extended or extrapolated dictionary of English. They were reminded of a process c a l l e d " s u f f i x a t i o n " by which new words in English are often formed and were given an example of a word derived by s u f f i x a t i o n . They were then t o l d that the examiner would read aloud a s u f f i x , some examples of English words containing that s u f f i x and some English words which had not been used with the s u f f i x before. F i n a l l y , they were asked to create and then pronounce the new word formed by the combination of the s u f f i x and the English word provided them, to define the word and to say whether they would use the word. The differences in the directions given to the subjects of the d i f f e r e n t experiments mainly concerns the d i f f e r e n t words used to convey the process of s u f f i x a t i o n . For instance, both groups of subjects i n P i l o t I were t o l d that new words can be formed by the "addition" of suffixes or endings to the end of exi s t i n g words. They were also then asked to pronounce the word r e s u l t i n g from the "addition" of the s u f f i x to the stem-word. The responses of the subjects of Group b' indicated that subjects were l i k e l y following the directions e x p l i c i t l y and adding or tacking on the s u f f i x to the stem-word without making any phonetic changes regardless of any tendencies they f e l t to -85-making phonetic changes. For instance, there was a predomin-ant trend of no phonetic change i n the responses of these subjects. In addition, two of the subjects during the t e s t asked the examiner i n r h e t o r i c a l fashion whether the subject was to "just add on the ending" to the stem-word. Therefore in P i l o t II and i n the main experiment, the word "addition" was changed to "combination" in reminding subjects of the pro-cess of s u f f i x a t i o n . The subjects were also asked to take the word and the ending and "put them together somehow" i n order to create the new word. It was hoped that the implications of no or of minimal phonetic change involved i n the notion of "adding" the s u f f i x to the stem would thus be eliminated. The idea of "combining" or "somehow putting together" the word-stem and the s u f f i x was intended to broaden subjects' expecta-tions to include the p o s s i b i l i t y for phonetic change to be i n -volved in the word creation task. Another difference i n the directions consisted of reminding subjects in the main experiment that the study i n which they were p a r t i c i p a t i n g was similar to a "survey" i n that a l l of th e i r responses were v a l i d and would not be considered as "right" or "wrong." The subjects had been t o l d t h i s once before when arrangements were made for t h e i r p a r t i c i p a t i o n i n the main ex-periment. The examiner e x p l i c i t l y referred to the experiment as a "survey" for two reasons. One reason was based on a subjective impression that the subjects of both p i l o t s were concerned with the v a l i d i t y of t h e i r responses. For instance, some subjects i n the p i l o t s repeatedly apologized for th e i r -86-"poor knowledge of English" and often responded during the interview i n a hesitant manner with anxious expressions on t h e i r faces. The directions for the two groups of subjects in each of the two p i l o t s d i f f e r e d s l i g h t l y i n a way which accounted for the absence of derivation examples i n the experimental task. These differences may be seen in the directions which are l i s t e d i n Appendix I. The cutback in the number of examples given to Group B in the main experiment was not mentioned i n the directions to the subjects as each s u f f i x i n the experimental task was exemplified once. 3.5.3 Order of presentation i n the experiments The contents of each experiment (except P i l o t II b " ) were presented i n the following order. F i r s t the directions were read to the subject. Then the examiner presented the word forms of the practice task and then those of the experimental task. There was no break between the practice task and the experimental task. Subjects were not.informed that t h e i r responses for the f i r s t nine stem-words were only practice derivations that would not be counted i n the r e s u l t s . The order of presentation within both the practice and experimental tasks was the following. F i r s t the s u f f i x and then one or two example derivations were presented. Then a word-stem was given. F i n a l l y the subject made his/her responses for the combination of the word-stem and s u f f i x . That i s , he/she pronounced the new word, attempted to define i t and then said whether he/she might use i t . The next stem-word was -87-presented, the subject made his responses, and so on.. This order of presentation of the s u f f i x , the example derivation(s) and the stem-words i s i d e n t i c a l to that used by Ohala i n his t e s t . In P i l o t II b", the examiner's instructions, and the sub-ject's opportunity, for defining the new words and for saying whether the subject would use the new words were delayed. When the subject had pronounced a l l of the words created from the stimuli i n both the practice and experimental tasks' l i s t s , she was informed that the examiner would give the s u f f i x , the example derivation, the stem-word and then the new words the subject had suggested. (The new words were pronounced by the examiner according to the phonetic t r a n s c r i p t i o n i n which the words had been recorded.) F i n a l l y the subject was directed to t e l l what the new word would mean and whether or not she would use the new word. The delay in defining and commenting on the use of the new words was interposed i n an attempt to see i f the predominant trend of no phonetic change i n the responses of P i l o t l b ' 's subjects had been influenced by the possibly d i f f i c u l t job of having to define and comment on the subject's use of the new word immediately aft e r pronouncing the word. Phonetic change did occur more frequently i n the responses of the single sub-ject i n Group b " than in the responses of Group b' 's sub-jec t s . However, t h i s separation of the pronunciation task from the other two required responses almost doubled the time needed to complete the interview. Therefore the method was not -88-adopted for use i n the main experiment. 3.5.4 Method of presentation With one exception, the studies were presented o r a l l y by the examiner. The directions and the word forms were read aloud by the examiner from a sheet on which the subject's responses were to be recorded. In the p i l o t experiments the suff i x e s , in addition to being pronounced aloud, were spelt aloud. The suffixes were presented once at the beginning of the l i s t of stem-words which were to be combined with that s u f f i x . The one exception to the o r a l presentation occurred in the main experiment. In t h i s case, the suffixes were not pronounced and spelt aloud. Rather, each s u f f i x was presented i n printed form on a 3 inch by 5 inch card. The subject retained the card i n front of him throughout the time needed to respond to the combination of the s u f f i x with each of the stem-words. The card was turned face-down before the card with the next s u f f i x was presented. The suffixes were presented i n printed form i n the main experiment because i t was found that subjects i n the p i l o t experiments had trouble remembering the s u f f i x when i t was spoken and spelt only once at the beginning of a series of stem-words. In addition, the fact that a given s u f f i x i s pro-nounced d i f f e r e n t l y i n d i f f e r e n t derivations made i t d i f f i c u l t to decide on how the examiner should pronounce the s u f f i x i n i s l o a t i o n without biasing the subjects' pronunciation of the s u f f i x i n d i f f e r e n t new words. A l l the subjects were interviewed i n d i v i d u a l l y . As was done by J. Ohala, i n the case where a subject produced more than one -89-pronunciation of a new word without indicating h i s preference, the examiner asked the subject which pronunciation he preferred. 3.5.5 Recording the data The subjects responses were recorded i n two ways: by a tape recording and by phonetic t r a n s c r i p t i o n . Tape recordings were made of the examiner's presentation of and the subjects' responses to the practice task and to the experimental task s t i m u l i . The d e f i n i t i o n s of the new words and the comments on the use of these words were also tape recorded to d i s t r a c t subjects' attention from the pronunciation task. The recording equipment used was an Ampex Micro 10 tape recorder powered either by batteries or by a P h i l i p s CP 9140 A power source. The Ampex Micro 10 microphone was clipped onto the subject's clothing just below the neck. Most of the subjects (in a l l the experiments except P i l o t II b " ) required about 40 minutes to complete the experimental session. Subjects* pronunciations of each new word were recorded i n broad phonetic t r a n s c r i p t i o n on a response form during the interview. The symbols of the International Phonetic Associa- t i o n (19 72) were used i n t h i s t r a n s c r i p t i o n . The tr a n s c r i p t i o n was broad i n the sense that i n addition to the segmental sym-bols, only the signs for the extended length of a sound [:], for the s y l l a b i c i t y of a contoid [,] and for primary stress f a l l i n g on a s y l l a b l e ['syllable] were used. To further d i s t r a c t subjects from the pronunciation task, comments made by the subject i n defining the new words and the i r decisions about possibly using the new words were ran-domly noted on the response form. Chapter IV ANALYSIS AND RESULTS OF THE DATA 4.1 Checking the r e l i a b i l i t y of the transcriptions Since the phonetic transcriptions of responses were done by the experimenter during the interview, an attempt was made to informally check the r e l i a b i l i t y of the transcriptions of the main experiment. The tape recorded responses of 3 subjects each from Group A and from Group B were played on a high qu a l i t y reproduction system and were transcribed by a second person. This person was directed to transcribe, as the experimenter had done during the interview, the d i f f e r e n t l y pronounced responses for the given stem-suffix sets and to note the pronunciation preferred by the subject. Only the responses to the experimental task whose stimulus was relevant to one of the rules, and whose form was preferred by the subject, were considered. Disagreements between the two transcriptions of responses were counted. The disagreements involved: (a) a difference in the location of primary stress; (b) a difference i n the number of s y l l a b l e s i n the response; (c) a difference i n a consonant and/or a vowel which was pre-dicted (by Chomsky and Halle, 1968) to undergo phonetic change; (d) a difference in which response was preferred by the subject. The two transcriptions showed disagreement i n an average of 3-4 responses for the t o t a l of 39 items which were considered in each of the six interviews. Two interviews (one from each group) were a r b i t r a r i l y chosen -90--91-to be transcribed a second time by the experimenter and the second transcriber i n order to check the r e l i a b i l i t y of each person's tr a n s c r i p t i o n s . There were three disagreements be-tween the two second transcriptions of one interview and four disagreements between the two second transcriptions of the other interview. These disagreements represented an increase of one disagreement for each interview over those found aft e r the f i r s t t r i a l of tra n s c r i p t i o n s . The differences between the second transcriptions did not involve many of the same re-sponses which produced disagreement in the f i r s t t r i a l . This meant that each transcriber was, on occasion, inconsistent i n transcribing some responses over the two t r i a l s . It was not anticipated that the transcriptions by each per-son would become consistent a f t e r further t r i a l s of l i s t e n i n g to and transcribing the tape recorded responses. In the case of the two interviews which were transcribed a t o t a l of four times, disagreements were resolved when one ^phonetic form for a given stem-suffix set occurred once and the other form occurred three times. In the two interviews, a t o t a l of 5 disagreements in responses were resolved i n t h i s way. The o r i g i n a l transcriptions of the examiner were retained for the remaining disagreements. 4.2 Estimations of the lengths of response latencies Word derivation experiments conducted by Zimmer (1969) and Steinberg and Krohn (1975) involved time l i m i t s for subjects' to make t h e i r responses. Since the present study did not im-pose a l i m i t on the time which subjects had to pronounce the -92-new word, an estimate of the range in the length of response latencies and an estimate of median and upper q u a r t i l e response latencies were made in two interviews (one l a s t i n g the greatest and one l a s t i n g the least amount of time) from each of Group A and Group B. In each interview the time latency between the end of the examiner's presentation of the stem-word and the beginning of the subject's preferred response was measured for each item in the practice and i n the experimental task . Considering the shortest and the longest interviews does not of course guarantee that the range of response latencies or the median or the upper q u a r t i l e response latencies found i n those interviews w i l l be respectively least and greatest for a group of subjects. It should, however,give at least a f a i r idea of the range to be encountered. The results can be seen in Table II Group A Group B o v e r a l l duration of interview 30 min. (shortest) 42 min. (longest) i 2 8 min. (shortest) i 4 5 min. (longest) response laten-cies (in sec.) range .7-6.2 1.0-48.8 .2-24.5 1.2-33.7 median 1.3 3.8 .8 3.3 upper q u a r t i l e 1.8 8.5 1.3 6.3 Table II. Estimated range, median and upper q u a r t i l e values o response latencies. -93-The r e s u l t s show a considerable range of response latencies for the subjects within each group. The median response latency values indicate that half of the preferred responses for the subjects of both groups were probably made within a period of about 4 seconds. This period of time i s sim i l a r to the time l i m i t s (mentioned e a r l i e r ) which were afforded the subjects i n the experiments of Zimmer (1969) and Steinberg and Krohn (1975). The upper q u a r t i l e values, however, vary considerably between the shortest and longest interviews of each group. 4.3 The "productivity" of the phonological rules 4.3.1 Introduction Authors such as Zimmer (1969), Krohn (1972), J. Ohala (1973) and Steinberg and Krohn (19 75) refer to the "productivity" of phonological processes accounting for the phonological regulars' i t i e s in language. These authors tend f i r s t to refer to the "productivity" of a phonological r e g u l a r i t y , of a sound pattern or of a vowel alternation. Only afterwards do they discuss the "productivity" of a formal rule which has been hypothesized to describe the r e g u l a r i t i e s i n sound patterns observed at the surface structure l e v e l . They do t h i s because, as was men-tioned in Chapter 3, they equate the "productivity" of a rule with the frequency of speakers' presumed 'use' of that rule, which they in turn i n f e r from the "productivity" or occurrence of a given sound pattern. In the present study also, the f r e -quency of occurrence of given sound patterns w i l l be discussed before tr y i n g to inf e r the subjects' 'use' of certa i n phonologi-c a l rules. -94-Krohn states: "By productivity, we are r e f e r r i n g to the a b i l i t y of a rule to apply to new items that have been added to a grammar" (1972, p.17). It seems peculiar to use the words "productivity" as well as " a b i l i t y " with reference to non-animate nouns such as "phonological r e g u l a r i t y , " "sound pattern" and "rule." Therefore the word "productivity" w i l l only be used in the present discussion with reference to the a b i l i t y of in d i v i d u a l subjects or groups of subjects to produce a sound pattern or to apply a rule to new items. Furthermore, i t w i l l be s p e c i f i e d whether the a b i l i t y of subjects involved i n t h i s "productivity" refers to the a b i l i t y only to produce a sound pattern by unspecified means or to the a b i l i t y to produce a sound pattern by using a c e r t a i n ru l e . Otherwise, reference to a sound pattern w i l l concern the notion of that pattern's f r e -quency of occurrence. Reference to a rule w i l l concern the sound patterns' implications for the corresponding frequency of use of that ru l e . 4.3.2 The occurrence of sound patterns i n each group of  subjects" One purpose for considering the d i f f e r e n t sound patterns and the number of times that they occurred in the responses of each group was to determine i f the d i f f e r e n t experimental conditions experienced by each group influenced the responses. The stem-suffix sets whose responses were considered for the occurrence of d i f f e r e n t sound patterns were those considered by Ohala (1973). In addition, some sets which had not been considered by Ohala happened i n t h i s study to e l i c i t responses exhib i t i n g the predicted sound pattern and were therefore -95-included i n the analysis. A l l of the stem-suffix sets men-tioned above are given i n Appendix II and comprise what s h a l l be referred to as the 'long' l i s t . Most of the stem-suffix sets considered in the analysis were preceded i n the conditions of Main Experiment A by an example derivation demonstrating the predicted sound pattern. (In order for an example derivation to qual i f y as 'preceding' a stem-suffix set, the example had to be presented at the beginning of the group of sets i n which the set under con-sideration occurred.) Some sets (noted i n Appendix II) were not preceded by an appropriate example derivation. The purpose of considering the occurrence of sound patterns was to determine i f the responses were influenced by subjects 1 ex-posure or lack of exposure to appropriate example derivations. Therefore i t seemed important to consider as well the occurrence of sound patterns i n a shorter l i s t of stem-suffix sets which included only those sets which had been preceded by an appro-priate example derivation i n Main Experiment A. This shorter l i s t w i l l be c a l l e d the 'short' l i s t . It i s possible that (a) the variety of sound patterns and (b) the percentage of occurrence of a given sound pattern i n the t o t a l responses considered w i l l vary under the influence of the number of responses considered i n the analysis. Steinberg and Krohn (1975) eliminated several forms of responses (see Chapter 2) from the data which was considered i n t h e i r analysis. Unlike the subjects p a r t i c i p a t i n g in Steinberg and Krohn's experiments, the subjects i n the present study were not -96-systematically limited i n the time which they had to make th e i r responses, ..and so probably always produced the response they intended to make. If they did make a s l i p of the tongue and were conscious of i t , they had the time to correct t h e i r response. Therefore a l l of the responses made by the subjects i n the present study were analyzed. It was believed that con-sidering a l l of the responses would more f a i r l y r e f l e c t sub-j e c t s ' a b i l i t i e s to produce predicted (and unpredicted) sound patterns and therefore to use predicted (and unpredicted) rules i n novel utterances. Another variable which can influence the vari e t y of sound patterns and t h e i r frequency of occurrence i s the choice and the number of categories to which the d i f f e r e n t responses are assigned. The categories were based upon the d i f f e r e n t types of responses produced by the subjects for the stem-suffix sets relevant to one rule . The responses d i f f e r e d according to the vowel or consonant i n the stem of the derivation which had been predicted to undergo phonetic change. For instance, methane + - i t y e l i c i t e d responses containing d i f f e r e n t vowels such as are seen i n [rns&aenati ], [ma'8elnati ], [ mal'B. mat M and toxic + -ism resulted i n the production of d i f f e r e n t con-sonants such as can be observed in [/ta ksa slz-m] and ['taksak-izm]* Some responses (such as sustain + - i t y [s/\sta'nlnad;i ] exhibited an additional segment or segments and were counted i n an .'Addition1' category. Responses were counted i n a 'Substitution 1 category when the stem of the response d i f f e r e d from that of the stimulus yet appeared to closely resemble another e x i s t i n g -97-stera. For example, [bu'distik] (derived from Buddha + - i c but resembling Buddhist + -ic) and [ ae'd 'va-t L z.m] (derived from adverse + -ism but resembling advert(ise) + -ism) were counted in a 'Substitution' category. A 'Deletion' category was used for responses such as ['sowfiz-m] (derived from sofa + -ism) and [ baVo. I a CI3 1 k 3 (from b i o l o g i s t + -ic) which appeared to drop one or more segments. Some responses (e.g. [ bo.11 QI a 1 s 11 k] (from b i o l o g i s t + -ic) and ['6'a-masij'a'n ] (from thermos + -ian) ) showed, according to the primary stress rule"*" tested i n t h i s study, an incorrect placement of primary stress. It may seem presumptuous to cate-gorize responses being considered as relevant to certain rules, according to another rule which i s i t s e l f being investigated i n t h i s study. However, those vowels which are predicted to undergo a phonetic change (according to a l l of the rules predicting a phonetic change in a vowel, with the exception of the rule V [+tense]/ V) happen to be those vowels which are located in the s y l l a b l e immediately preceding the s u f f i x . These vowels are therefore those which, according to the primary stress rule, re-2 ceive primary stress when derived with.certain suffixes . In order to avoid u n f a i r l y counting responses i n one or another vowel category when the vowel was not predicted to change unless i t was stressed, the primary stress rule was given the benefit of the doubt,- that i s , responses with an-.unpredicted stress -'-The formalized primary stress rule considered here i s V •*• [1 stress] / . C + a f f i x . Chomsky and Halle l i s t some a f f i x e s which are exceptions to t h i s rule, one of which, -ism, i s combined with cert a i n stem-words in t h i s study. -98-pattern were placed i n a separate category. The separate category was f e l t to be warranted by the fact that Groups A and B each produced the predicted stress pattern i n about 67% of t h e i r responses (to the long l i s t of stimuli) which were hypothetically e l i g i b l e for a change i n the stress l o c a t i o n . The 'Wrong Stress ' category was not mutually exclusive from each of the other categories. For instance, a response could contain a predicted (e.g. tense or lax) vowel yet show 'incor-r e c t ' stress l o c a t i o n : for example, [ bi'doian] from bed + -ian and [ ' s a e f L J a t D j i i ] (from sapphire + -atory) . The response [ m 3 0a'n L n a t i ] (from methane + -ity) could be counted i n either of the 'Addition' or the 'Wrong Stress' categories while [ baValad"} ik] (from b i o l o g i s t + -ic) could be counted i n either of the 'Deletion' or 'Wrong Stress' categories. Such responses were counted i n the 'Addition' and 'Deletion' categories r e r -spectively. In the 'Addition' case, one could not t e l l i f the subject considered the stem of the derivation to be [SASt3n-] or [ S A S t a n c n - ] and therefore could not say whether the stress rule was c o r r e c t l y applied. In the 'Deletion' case, on the other hand, one could not know i f the location of stress was assigned before or aft e r the s y l l a b l e was dropped and therefore one could not f a i r l y judge the f i n a l stress assignment. The responses to the stem-suffix sets which were considered for the primary stress rule were counted i n categories with stress f a l l i n g on d i f f e r e n t s p e c i f i e d s y l l a b l e s . For example, [ 'Barmaaijan] (from thermos + -ian) was counted i n the category where stress f e l l upon the i n i t i a l s y l l a b l e . Two responses with -99-the form ['tami jasan] (derived from Thomas + -ian) necessitated the adoption of the category c a l l e d 'Metathesis' since one could not t e l l whether the assignment of stress occurred before or afte r the transposition of segments i n the stem and the s u f f i x . (This category was used for data relevant to the primary stress rule as well as to the rule which tensed vowels preceding CiV.) Two responses of the form Paed'varSizm] (from adverse + -ism) required the adoption of a category c a l l e d 'Equal Stress 1 i n which two s y l l a b l e s contained heavy stress of equal strength. Each group's responses to the stimuli which were relevant to a given rule were f i r s t categorized in matrices which con-sidered the responses to one stem-suffix set. The columns of the matrix contained the d i f f e r e n t categories of responses to the set while the rows referred to the responses of the subjects in the group. These matrices served as the basis for a l l l a t e r measures of the occurrence of sound patterns. The cumulative number of predicted responses i n each group Each subject's predicted responses for a given rule were counted for the long l i s t of s t i m u l i . 'Predicted- responses or sound patterns s h a l l henceforth refer to those sound patterns predicted by the phonological rules under discussion which were hypothesized by Chomsky and Halle (1968). For each ru l e , the subjects in each group were assigned an average rank according to the number of predicted responses they produced. However, since a d i f f e r e n t number of stem-suffix sets was con-sidered for each rule, finding the unweighted mean rank of the seven (average) ranks i n j u s t l y gave equal weight to each -100-(average) rank. Therefore each ru l e was assigned a weight equal to the number of stem-suffix sets considered relevant to the ru l e . The number of sets considered for the /s/-voicing rule varied for each subject according to the number of e l i g i b l e responses (containing [. V + s + V. ])' he produced. The T+tense] /s/-voicing rule was assigned a weight of 3 i n both groups even though the average number of e l i g i b l e responses per subject was 2.5 in Group A and 2.3 i n Group B. A weighted mean rank was 2 found for each subject . The unweighted and weighted mean ranks of each subject were then compared. The two ranks never d i f f e r e d by more than 5.0 mean ranks in either group of subjects. Often the two mean ranks of a subject were equal. Another informal check on the two mean ranks was made by grouping subjects according to a combination of the 'statuses' ( i . e . , low, medium, high) of the average ranks which subjects achieved for each r u l e . When the unweighted and weighted mean ranks were than considered for these grouped subjects, both ranks corresponded f a i r l y well to the status of the grouping, with only a few cases where, 2 The formula used for the weighted mean rank was: i=7 £ w. • r. 1=1 1 1 weighted mean rank = E w. i=l 1 where w. re f e r s to the weight of a given rule and r ^ refers to a subject's average rank for that r u l e . -101-according to either mean rank, the subject should have been placed i n another grouping. Therefore one might conclude that the fact that a varying number of stem-suffix sets were con-sidered for the d i f f e r e n t rules did not greatly influence sub-j e c t s ' mean ranks i n the group's productivity of a l l of the predicted sound patterns. The difference in the two groups' o v e r a l l productivity of predicted sound patterns can be shown by comparing the cumula-t i v e histogram of each group's t o t a l number of predicted sound patterns. The responses to the short l i s t of s t i m u l i were used in order to take the opportunity to att r i b u t e the difference i n the two groups' cumulative number of predicted responses to the di f f e r e n t experimental conditions. The subjects were ordered according to t h e i r mean unweighted ranks. The subjects with the lowest and the highest ranks i n Group A were eliminated i n order to keep the number of subjects i n the two groups equal. The r e s u l t s can be seen i n Figure 2. The subjects of each group are plotted along the abscissa with the subject with the •worst' rank ( i . e . with the smallest number of predicted re-sponses) at the far l e f t and the subject with the 'best' rank (i . e . the greatest number of predicted responses) at the far r i g h t . The cumulative number of predicted responses i s plotted along the ordinate. There c l e a r l y i s a difference i n each group's production of predicted responses. In order to at-tri b u t e the difference to the difference experimental conditions, one needs to know that the two groups of subjects were equally -102-Figure 2. Cumulative number (N) of predicted"responses (short l i s t ) for each group. -103-Vadept' i n the p r a c t i c a l matter of producing novel derivations. Such an assumption could be v e r i f i e d s t a t i s t i c a l l y i f several groups of subjects p a r t i c i p a t e d (in d i f f e r e n t orders) i n the two experiments on two d i f f e r e n t occasions. This type of check was beyond the scope of t h i s study. However, a less formal check on the difference in the word-derivation s k i l l of each group can be made by comparing each group's cumulative h i s t o -gram of percentage of predicted changes versus subject rank. The data from the short l i s t of stem-suffix sets were used. (The data from the long l i s t of stimuli produced curves which are very similar to those shown for the short l i s t . ) The sub-jects of each group were ranked according to the t o t a l number of predicted responses they made for the six rules under con-sideration i n the short l i s t . Once again, the subjects i n Group A with the highest and lowest ranks were eliminated. For each group, the cumulative percentage of predicted responses were plotted on the ordinate. The subjects were plotted i n rank order on the abscissa (with the worst ranking subject on the far l e f t and the best ranking subject on the far r i g h t ) . The curves are shown i n Figure 3. The curves f i t very clos e l y , suggesting that the t o t a l number of predicted responses was d i s t r i b u t e d s i m i l a r l y among the mem-bers in each group. This d i s t r i b u t i o n which i s seen i n the two groups i s l i k e l y what would be expected i f the groups repre-sented a normal sampling of the population—that i s , a few subjects produced a large number of predicted responses, a few produced an intermediate number and a few produced a small -104-% predicted responses 100 T •Worst' 'Best' Figure 3, Cumulative percentage of predicted responses (short l i s t ) for Group A ( ) and Group B (---105-number. One could l i k e l y conclude that the two groups were si m i l a r i n t h e i r a b i l i t y to perform the experimental task. In ad-d i t i o n , each group of subjects was l i k e l y representative of the population of young university students from which they were drawn. So, i n turn, one could probably conclude that the d i f -ference in the cumulative number of predicted responses i n the two groups was a r e s u l t of the d i f f e r e n t experimental condi-ti o n s : namely, the presence of absence of example derivations showing relevant sound patterns. Whereas Steinberg and Krohn (19 75) found that a small number of subjects were responsible for the production of pre-dicted responses, i n the present study 13 out of 15 subjects i n each group were responsible for approximately 80% of t h e i r group's predicted responses. The data used to construct Figure 3 showed that 13 subjects i n Group A produced 80.4% of that group's t o t a l number (short l i s t ) of predicted responses while 13 subjects in Group B produced 77.3% of that group's predicted responses. The various sound patterns and t h e i r frequency of occurrence  in each group The responses counted i n each category for a given stem-s u f f i x set were t o t a l l e d and entered in a matrix which con-tained the categorized responses to a l l of the s t i m u l i relevent to a given ru l e . The columns of the matrix referred to the categories of responses and the rows referred to the stem-s u f f i x sets. The percentage of occurrence of each given cate-gory of response with respect to the t o t a l number of responses -106-(considered for a rule) was found. The r e s u l t s , seen i n Tables III through IX, show for each group of subjects the percentage of occurrence of given categories of responses for each r u l e . Results for the short l i s t of s t i m u l i are shown on the l e f t side of each column while the re s u l t s for the long l i s t of s t i m u l i are shown on the right side of each column i n the tables. The data which w i l l be discussed i n the text be-low with reference to each rule w i l l be that derived from the short l i s t . The vowel laxing rule (henceforth VLR^) : V -> [-tense] CVCV As can be seen i n Table I I I , the sound pattern predicted by VLR^ (e.g. ['spaesati ] from space + -ity) occurred in a greater percentage of responses i n Group A (38.1%) than i n the responses of Group B (20.4%). The 'tense vowel' sound pattern (e.g. [ ' b j i d f t b a t i ] from bribe +. - i t y . and [ sup'jimafal] from supreme + - i f y ) occurred most often in the t o t a l responses of each of the two groups. However, t h i s pattern occurred more often in the responses of Group B (74.9%) than i n those of Group A (57.2%). Both groups produced a few responses cate-gorized as 'Additions' (e.g. [f l i ' s i d a t i 1 from fleece + - i t y ) . The segments which were added in the new derivations were [-err] , [-id-], [-el-] and [-acs-] . They always appeared between the stem and the s u f f i x - i t y . The occurrence of the responses containing [-in-], [-id-] and [-ei -J might i n part be accounted for by the number of e x i s t i n g English derivations containing one of these s y l l a b l e s and the s u f f i x - i t y (e.g. v a l i d i t y , t i m i d i t y ; -107-TOTAL LAX TENSE SUBSTI-RESPONSES VOWEL VOWEL ADDITION DELETION TUTION 1 GROUP A no. 283 300 108 115 162 170 6 6 5 5 2 4 Q. •& 100 100 38.1 38.3 57.2 56.6 2.1 2 1.8 1.7 0.7 1.3 GR DUP B no. 255 270 52 57 191 201 6 6 3 3 2 2 Q. 100 100 20.4 21.1 74.9 74.4 2.3 2.2 1.2 1.1 0.7 0.7 Table I I I . V -*• [-tense]/ CVCV. The number of responses in each category and the percentage of occurrence of each category for the l i s t s of stimuli relevant to the rule mentioned. The data for the short l i s t of stimuli are on the l e f t side of each column while the data for the long l i s t are on the ri g h t side. -108-v i c i n i t y , v i r g i n i t y , femininity; omneity, spontaneity, deity and homogeneity). Subjects may have analogized t h e i r responses a f t e r the surface structure of words such as these. Or perhaps they used an independent phonological r u l e , by which those e x i s t i n g words were p a r t l y generated, to produce such responses. The insertion of the four s y l l a b l e s i n i t i a l l y l i s t e d above might be due to a strategy which c a l l e d for segment or s y l l a b l e reduplication. This follows from observing, f i r s t , that [-In-] was inserted a f t e r a sequence of [-V+n-] in [ ma 8annuit\i3JtLJ , [SASta'ntn a t i ] . and [ bitw"i1 n u n a t i ] . Secondly, [-ej-] was i n -serted after a s y l l a b l e containing a s i m i l a r l y tense diphthong in [ t u c u ' t e i a t i ]. Thirdly, [-ae+s-] occurred aft e r the se-quence of [-ej+s-] in [ s p e t ' s a e s a t i ] . Both groups also produced a few responses in the 'Deletion' category (e.g. ['saefato^i ] from sapphire + -atory) . The dele-; t i o n of a s y l l a b l e might be due to subjects' conscious or unconscious desire to make one or more phonetic changes i n the second s y l l a b l e of the stem but being unable to make a change which would be 'satisfactory' to them. They might therefore drop the 'awkward' s y l l a b l e . On the other hand, they may have had a 'distaste' for words containing more than four s y l l a b l e s . Group A produced responses in the 'Substitution' category. In t h i s case, two subjects responded with [ b i ' t w i K s a d i ] for between + - i t y . This response may have occurred because sub-ject s wished to lax the vowel preceding the s u f f i x . On the other hand, they may have produced [- ks-] i n the new derivation by considering the surface structure of betwixt, a word that i s -109-close i n sound structure and meaning to the stem between. In addition, the two stems occur together in the phrase, "betwixt and between," and so might have some col l o c a t i n g association. One subject i n Group B placed stress on a wrong s y l l a b l e i n just one response. The primary stress rule (henceforth PSR) :  V -> [1 s t r e s s ] / C + a f f i x The sound pattern predicted by the PSR (e.g. [ alow dmati] from iodine + -ity)occurred i n an almost equal percentage of the responses of both groups: 64% i n Group A and 61.4% i n Group B (cf. Table IV). Furthermore, t h i s sound pattern was the most frequent one in both groups' responses. The response category which ranked second i n the number of responses was the 'Deletion' category (e.g. [ bdl'oiladj ik] from b i o l o g i s t + -ic) . The per-centages of occurrence of 'Deletion' responses were also quite close for the two groups, being 22.8% i n Group A and 19.2% i n Group B. One might mention and t r y to explain the occurrence of other sound patterns produced for the stimuli under considera-tion for the PSR. However, such explanation, as i n the cases of the explanations for the unpredicted categories of responses for VLR^, i s not exhaustive and would only be speculative. The vowel tensing rule (henceforth VTR^ : V-> [+ tense]/ V Table V shows that the 'tense vowel' sound pattern (e.g. [zi'bjeiak] from zebra + -ic_) predicted by t h i s rule occurred most often (41.1%) in the responses of Group A while the 'Deletion' category of response (e.g. [Vibj»i.k]) occurred most . often in Group B (4 3.3%). A clue to the common occurrence of TOTAL SSSS+SS SSSS+SS SSSS+SS SSSS+SS SSSS+SS DELETION ADDITION SUBSTITUTION META- EQUAL RESPONSES THESIS STRESS GROUP A no. 153 221 4 13 : l 1 0 0 98 149 5 5 35 40 2 3 8 9 0 0 0 1 % 100 100 2.6 6.3 0.7 0.4 0 0 64 67.4 3.3 2.2 22.8 18 1-3 1.3 5.2 . 4.1 0 0 0 0.7 GROUP B no. 135 195 10 18 1 1 1 1 83 132 8 8 26 30 1 1 4 4 1 1 0 1 % 100 100 7.4 9.2 0.7 0.5 0.7 0.5 61.4 67.6 5.9 4.1 19.2 15.4 0.7 0.5 2.9 2 0.7 0.5 0 0.5 TABLE IV. V ->• [1 s t r e s s ] / C + a f f i x . (See Table I l l ' s legend.) The 'S' i n SSSS+SS i n the categories above stands f o r ' s y l l a b l e . ' -111-TOTAL TENSE RESPONSES VOWEL LAX VOWEL ADDITION DELETION SUBSTI-TUTION no. no. .34 51 100 100 30 45 100 100 14 16 41.1 31.1 12 13 40 28.9 GROUP A 0 0 0 0 2.9 11.8 GROUP B 1 2 3.3 4.4 3.3 20 12 22 35.2 43.1 13 18 43.3 40 7 7 20.6 13.7 10 6.6 TABLE V. V [+ tense]/ V. (See Table I l l ' s legend.) -112-t h i s sound pattern i n both groups might be the fact that (as was mentioned by Ohala) there are several words e x i s t i n g in English that demonstrate t h i s sound pattern, e.g. s a l i v a / s a l i v i c , g u e r r i l l a / g u e r r i l l i c , cholera/choleric, oedema/oedemic and vulva/vulvic. The percentage of occurrence of the 'tense vowel' response in Group B (40%) was, however, almost as great as i t s percentage i n Group A. In addition, the predicted 'tense vowel' responses in Group B occurred almost as often as the most common response (the 'Deletion' category of response, 43.3%) in that group. The 'Substitution' responses (20.6% in Group A and 10% in Group B) a l l had the form [ b.u'dList a kJ (from Buddha + -ic) . One possible explanation of t h i s sound pattern i s that subjects were considering the stem-word to be Buddhist. This may not be surprising, as a l l three English derivations of Buddha (e.g. Buddhist, Buddhistic, Buddhistical) contain the stem Buddhist. The 'Addition' responses involved the i n s e r t i o n of [?] or [kt ] between the stem-word Buddha and the s u f f i x - i c . This strategy was probably used to separate the sequence of the two vowels r e s u l t i n g from the combination of Buddha + - i c . One might also note that there are a number of English derivations ending i n [-aakttk] (e.g. d i d a c t i c , prophylactic, climactic) whose sound patterns the subjects may have r e c a l l e d and extended by some means. The vowel tensing rule (henceforth VTR 2): V + [+ tense]/ CiV "[-high] Although the predicted tense vowel sound pattern (e.g. [ 8»'mo>wsijanj from thermos + -ian) occurred i n a higher -113-percentage of the responses of Group A (23.5%) than i n Group B (8.3%), i t was not, by f a r , the most common category of response in either group (cf. Table VI). The 'tense vowel' response had the second highest percentage of occurrence i n Group A and only the t h i r d highest percentage in Group B. A sound pattern con-taining a lax vowel (e.g. [ Bzr'm a s i ja n] from thermos + -ian and [ia'mus ] from Thomas + -ian) occurred most often in each groups' responses, in fact three to four times as often as the second ranking pattern. The 'lax vowel' category of response occurred almost as often i n each group (69.9% i n Group A and 70.8% in Group B). A speculative explanation of the 'Deletion' responses (e.g. [VJUSmi ja n ] from Christmas + -ian) might be that subjects were dropping what they f e l t were 'awkward' s y l l a b l e s . In doing so, they could eliminate the 'predicted' need (a) for a change in the location of primary stress in a l l of the two-syllable stem-words and (b) for a change to a tense vowel i n Christmas + -ian and thermos + -ian (due to the double consonant context). In Group B, responses counted in the 'Wrong Stress' category (15%) numbered almost twice as many as those i n the predicted 'tense vowel' category. Six of. eight 'Wrong Stress' responses made by Group A and six of eighteen made by Group B involved primary stress being located on the f i r s t vowel of the s u f f i x -ian (e.g. [ t am a'si j an] , [ faa 'd i jan] and [ uow ba'ti j an] ) . One speculative explanation i s that subjects may have been extending, by some means, the sound pattern seen in English derivations -114-TOTAL TENSE LAX META- WRONG RESPONSES VOWEL VOWEL ADDITION DELETION THESIS STRESS no. 136 32 87 GROUP A 1 8 0 8 Q. 100 23.5 63.9 0.7 GROUP B 5.8 0 5.8 no. 120 10 85 0 5 2 18 o. 100 8.3 70.8 0 4.1. 1.6 15 Table VI. V -»••[+• tense]/. CiV. The number of [-high] responses in each category and the percentage of occurrence of each category for the single l i s t of stimuli relevant to the rule mentioned. -115-such as plebian, European and Jacobean, whose stems end in a consonant. This sound pattern also occurs i n many ex i s t i n g derivations whose stems end in a vowel (such as Chaldean and Galilean ). The velar softening rule (henceforth VSOR) : /k / -»• s/_ I The predicted 'velar softening ' sound pattern (e.g. [dow'mtstastzm] from domestic + -ism) showed the highest per-centage of occurrence in Group A's responses (52.9%) but only the second highest percentage of occurrence i n Group B's responses (42.2%) (cf. Table VII). Conversely, the category of response which retained the [k] of the stem (e.g. [ dow'mesia kcT.m ] from domestic + -ism) ranked second in the percentage of occurrence (33.3%) in Group A's responses but f i r s t i n percentage of occurrence (48.8%) i n the responses of Group B. The underlying vowel in the second s y l l a b l e of obtain and  pertain \ Chomsky and Halle state that ordinary underlying vowels undergo the vowel s h i f t rule (VSR) only i f the vowel i s [+ tense]. They propose that some vowels (such as those i n the underlying forms of retain and detain) remain e l i g i b l e for the application of the VSR even i f the vowel has been laxed in the environment d i r e c t l y preceding a double consonant. Ohala believes that i f the underlying vowel i n the second s y l l a b l e of obtain and pertain are marked in e l i g i b i l i t y for vowel s h i f t on laxed vowels, then one would expect obtain + ion to possibly be rendered [AbtCnSan ]/ but obtain + atory should then likewise be rendered [A bt tnatoai ] (1973, p.44). -116-TOTAL RESPONSES /s/ [k]; ADDITION DELETION no. 51 68 27 30 GROUP A 17 20 1 2 6 16 o, *5 100 100 52.9 44.1 33.3 29.4 GROUP B 1.9 2.9 11.7 23.5 no. 45 60 19 20 22 28 1 1 3 11 % 100 100 42.2 33.3 48.8 47 2.2 1.7 6.6 18.3 TABLE VII. A d / - s/__I (See Table III s legend.) -117-If the underlying vowel i s not marked to undergo the VSR a f t e r being laxed, then [ ae ] might be expected to appear in the second s y l l a b l e of the surface structure of obtain + -ion and obtain + -atory. The s u f f i x -atory i s predicted to cause the vowel preceding i t to become lax according to VLR^. The s u f f i x vion i s predicted to lax the vowel preceding i t a f t e r the epenthesis of a dental consonant (between a stem such as r e t a i n , ending in a dental nasal, and a suffix) which creates a consonant c l u s t e r . The rule which predicts the lax vowel i s V ->- [-tense]/ CC, henceforth c a l l e d VLR 2. If the subjects were unable to, or did not wish to, extend the sound pattern in which the vowel i s laxed in the environments preceding CVGV or CC, then presumably the tense vowel [ e l ] i n the stem-word would be retained i n the subjects' responses. The appearance of [et] in Group B's responses to s t i m u l i combining with the s u f f i x -ion could lead to two conclusions. F i r s t , the sub-jects may not have laxed the vowel because they could not or did not extend the sound pattern wherein a vowel i s laxed before a consonant c l u s t e r . On the other hand, the subjects may not have extended a consonant cluster sound pattern in the derivation since they were given no leading example derivations and since the written form of the s u f f i x which they were given did not include a " t " . For the present, discussion of the responses to pertain and obtain w i l l be l i m i t e d to the occurrence of lax or tense vowels which would r e f l e c t on the predictions for a 'lax vowel' sound pattern as well as the opportunity to in s e r t a [t] or -118-a [§] before the s u f f i x -ion. It w i l l be assumed that sub-jects can extend the double consonant sound pattern. In other words, responses without the double consonant pattern w i l l not be excluded from the analysis for tense or lax vowels. Com-ments on the underlying vowel in question i n these stem-words w i l l be discussed i n a l a t e r section. Table VIII shows that the predicted 'lax vowel' sound pattern (e.g. [ pa*'t E n a to J 1 ] from pertain + -atory and [ftb'ttn^n] from obtain + -ion) occurred most often (70.5%) i n the responses of Group A. One 'lax vowel' response in Group A, [ pa-'ttn jan ] , occurred i n the environment preceding CiV ... which i s predicted by Chomsky and Halle to e l i c i t a tense vowel. Twenty percent of the responses of Group B were counted i n the 'lax vowel' category. A l l of the lax vowel responses made by Group B to the sti m u l i combining with the s u f f i x -ion occurred d i r e c t l y preceding a consonant cluster environment [ ri ^ n ] which i s hypothesized to e l i c i t a lax vowel. The 'tense vowel' response (e.g. [ Ab't el n j a n ] and [ pa- 'tel n at 3 J i ]) occurred most often (76.6%) i n the responses of Group B. Some of Group A's responses (27.9%) contained tense vowels. Both groups of subjects also produced a few 'Deletion' responses such as ['pa-tatoai 1. The /s/-voicing rule (henceforth /s/VR):  /s/ •* [+ vo i c e ] / V _ V [+ tense] ' The sound pattern predicted by t h i s rule cannot s t r i c t l y be included in the analysis which considers the e f f e c t s of the di f f e r e n t experimental conditions. None of the sti m u l i -119-TOTAL RESPONSES ..LAX VOWEL TENSE VOWEL DELETION 68 100 60 100 GROUP A 48 70.5 19 27.9 GROUP B 12 20 46 76.6 1.5 3.3 Table VIII, V -* [- tense]/ JCVCVI . A b i l i t y ICC } to extend the double consonant sound pattern i n the context -ion was assumed. (See Table VI's legend.) - u n -considered for t h i s rule were preceded in Main Experiment A by example derivations demonstrating the 'voiced /s/' sound pattern. Some idea of the subjects' a b i l i t y to extend the sound pattern to novel utterances might be gained from considering the numbers in Table IX. In new derivations in which [s] was preceded by a tense vowel, Group A produced the predicted sound pattern (e.g. [ 8 v 'mow s"» jan] from thermos + -ian) in 7% of th e i r responses while Group B f a i l e d to produce the sound pattern. One might speculate that the reason that even a few of Group A's e l i g i b l e responses exhibited the 'voiced-/s/' sound pattern was that Group A had been exposed to example derivations showing phonetic change and were, on the whole, producing more responses involving phonetic changes. Perhaps i n a li m i t e d sense, therefore, the responses e l i g i b l e to be analyzed for the occurrence of the voiced / s / can be included in the comparison of each group's responses. Further evidence for the sound pattern occurred i n responses to prose + - i f y wherein a few subjects i n each group laxed the vowel i n the stem of the derivation. One of these four 'lax vowel' responses made by Group A and one i n three pro-duced by Group B devoiced the [z] and produced [ p j a s a f a t l . Summary The sound patterns predicted by each of the rules occurred more often in the responses of Group A whose subjects were exposed to example derivations which demonstrated a l l but one -121-TOTAL ELIGIBLE RESPONSES [-VOICE] [+VOICE] GROUP A no. 42 39 3 % 100 93 7 GROUP B no. 31 31 0 Q, "6 100 100 0 Table IX. s -* [+voice]/ V V. [+tense] (See Table VI's legend.) -122-of the predicted sound patterns. Since the two groups were equally adept i n handling the experimental task, the greater percentage of occurrence of predicted sound patterns in the responses of Group A can l i k e l y be attributed to that Group's exposure to leading example derivations. This hypothesis w i l l be further investigated l a t e r i n r e l a t i o n to the means by which subjects extended given sound patterns. If the results of the /s/VR are a r e l i a b l e i n d i c a t i o n for each group's tendency to make predicted phonetic changes, the trend towards or away from making phonetic changes (predicted by some rules) might also bear an influence on the occurrence of predicted sound patterns for another r u l e , independent from the others. One might also consider the r e l a t i o n of the t o t a l number of phonetic changes (whether predicted or unpredicted) to the number of predicted changes made. It could be that as the number of responses involving a phonetic change increases or decreases, so does the number of predicted phonetic changes. Finding the t o t a l number of responses involving a phonetic change i n the stem of the derivation would be very d i f f i c u l t . It would be desirable to categorize responses according to the number of phonetic changes made. For example, should not [maBa'ni-nati ] (from methane + -ity) count as having been sub-mitted to more phonetic changes than [ rna'B el n a 11 ] ? It i s d i f f i c u l t to t e l l which phonetic changes are discrete from one another and so assign to them a unit value. With stimuli such as sustain + - i t y and pertain + -ion i t would also be impossible to t e l l whether a subject l e f t the primary stress on the -123-hypothetically appropriate s y l l a b l e because he was oblivious to the stress pattern or because he f e l t that the stress should remain i n the same location i n the new derivation. Despite the apparent trend for predicted sound patterns to occur more often when subjects are exposed to example derivations i l l u s t r a t i n g the sound patterns, one cannot specify whether the sound patterns are produced through the use of analogical phonological rules, the independent phonological rules under investigation or through some other means. 4.4 Attempts to determine the means by which a sound pattern was produced 4.4.1 The influence of example derivations on 'similar' stem-suffix sets A further check to see i f the more frequent occurrence of predicted responses i n Group A was due to the presence of r e l e -vant example derivations was attempted by considering each group's chi-square value for the number of predicted and un-predicted responses for ce r t a i n stem-suffix sets. The stem-word of each set was similar to the stem of the preceding example (with the relevant sound pattern) i n that i t contained the same strategic vowel or consonant as that i n the example. It was expected that the difference i n the number of predicted responses might also have some implications for the subjects of Group A having used a given example derivation as a surface structure form on which to model t h e i r responses to a given stem-suffix set. That i s , i f the occurrence of the predicted sound pattern was s i g n i f i c a n t l y greater i n Group A's responses to a given stem-suffix set, there would be a p o s s i b i l i t y that the -124-difference was caused by subjects' exposure to an example derivation i l l u s t r a t i n g the predicted sound pattern with the vowel or consonant which was common to the example and the stimulus. The chi-square measure was taken on several stem-suffix sets by entering the number of predicted responses and the number of remaining responses made by one group to a given stem-s u f f i x set i n one row of a 2 x 2 matrix. The matrix i s exempli-f i e d i n Table X using the data from the two groups' responses to slave + - i f y . No. Predicted Responses No. Remaining Responses Tot. 8 9 17 4 11 15 12 20 32 Group A Group B Tot. Table X. The matrix used for the chi-square measure of both groups' responses to slave + - i f y . Due to the small number of each group's responses to a stem-s u f f i x set many of the matrices contained entries with a small number. The questions as to how small expected frequencies could be before using cert a i n chi-square correction formulae and before abandoning the use of the chi-square measure altogether were answered d i f f e r e n t l y in d i f f e r e n t sources. Therefore two correction formulae were used, in addition to the formula for the tetrachoric c o r r e l a t i o n c o e f f i c i e n t (see fn. 3, p.126), to -125-determine whether the values obtained for a given set consis-te n t l y remained s t a t i s t i c a l l y s i g n i f i c a n t with respect to confidence l e v e l . The sets whose values did remain consistent are mentioned below. The l e v e l of significance mentioned for each set i s the most conservative one obtained for i t . There were s t a t i s t i c a l l y s i g n i f i c a n t differences i n the number of predicted responses for only a few sets. Only one of those preceded by mammal/mammalian (Thomas + -ian (p < .05)) showed a s i g n i f i c a n t difference. A possible explanation for obtain + -ion (p '< .001) , pertain + -ion (p < .05), obtain + -atory (p < .05) and pertain + -atory (p < .005) a l l having reached s i g n i f i c a n t l e v e l s i s , f i r s t , that the stem-words of the stimuli and those of the examples ended i n t-eln] ; secondly, they were composed of stems and prefixes. Likewise, a possible explanation for methane + - i t y arid sustain + - i t y , which were preceded by sane/sanity, having reached lev e l s of significance (both at p < .005) i s that t h e i r stems contained the same V + C sequence as the example's stem-word. Predicted responses occurred s i g n i f i c a n t l y more often in Group A's responses to the stem-suffix sets which were similar to t h e i r preceding example derivations. This conclusion might at f i r s t glance appear as evidence that the subjects of Group A were using the surface structure of the example derivations ( p a r t i c u l a r l y of those whose stems contained [-eln] ) as models on which to base analogous responses. This idea i s reinforced by the fact that many of the stem-suffix sets whose numbers of -126-predicted responses did not reach s i g n i f i c a n t l e v e l s had only a vowel in common with t h e i r preceding example derivation's surface structures. However, the p o s s i b i l i t y s t i l l remains that the example derivations served to i l l u s t r a t e the indepen-dent phonological rule formalized by Chomsky and Halle. For instance, sane/sanity could have i l l u s t r a t e d VLR^ more e f f e c t i v e l y for pairs such as methane + - i t y and sustain + - i t y than for pairs whose stem-words did not contain [el]. 4.4.2 The influence of a previous response on a l a t e r response J. Ohala sought to check whether analogical phonological rules were the means by which a sound pattern was productive by considering the cor r e l a t i o n of responses to successive or nearby stem-suffix sets. The s t a t i s t i c a l measure that he used and that was used i n t h i s study i s the tetrachoric c o r r e l a t i o n 3 c o e f f i c i e n t . In the present study, the responses considered were those that were counted i n two sound patterns that (a) were common to both groups' responses to. the two stem-suffix sets and (b) most frequently occurred in both groups' responses to the two sets. For each group, the data for a pair of stem-suffix sets were arranged i n a matrix such as the following (which uses data from methane + - i t y and sustain + - i t y ) : The tetrachoric c o r r e l a t i o n c o e f f i c i e n t measures the related-ness of events t a l l i e d in a 2 x 2 matrix, using the formula: IT /ad - /be r^ = sin t 2 /ad + /be The d i s t r i b u t i o n of r t was obtained by generating (2500 times) random 2 x 2 matrices, for each value of N (sum of the entries) from 10 to 17 and from 30 to 32. The levels of confidence were inferr e d from these d i s t r i b u t i o n s . -127-Group A (methane + -ity) . tense, vowel . . lax vowel tense vowel (sustain + - i t y ) lax vowel Table XI. Example of a matrix arrangement for an ind i v i d u a l group's' responses to a pair of stem-suffix sets. The pairs of stem-suffix sets whose responses showed a s i g n i f i c a n t p o s i t i v e c o r r e l a t i o n involve s i g n i f i c a n t l y more pairs of responses which are either both sound pattern 'x' or both sound pattern *y'. Those pairs whose responses showed a s i g n i f i c a n t l y negative correlation involved s i g n i f i c a n t l y more pairs of responses which were comprised of two d i f f e r e n t sound patterns. When the co r r e l a t i o n c o e f f i c i e n t was non-s i g n i f i c a n t , the responses to the pairs were usually comprised of more equal numbers of 'both the same' responses and 'different' responses. It i s possible, in the cases of the stem-suffix sets ( p a r t i c u l a r l y successive sets) whose responses showed a s i g n i f i -cant pos i t i v e c o r r e l a t i o n , that the form of the f i r s t response influenced the form of the following response. This could support both the hypothesis of an independent phonological rule or that of a model whose surface structure was the object of analogy. A l t e r n a t i v e l y , the influence of the e a r l i e r response a = 1 b = 1 c = 3 d = 10 -128-could have occurred in s t i l l some other way. However, a high s t a t i s t i c a l c o r r e l a t i o n does not imply a causal r e l a t i o n be-tween the two events considered. Therefore, checking the cor-r e l a t i o n of s i m i l a r responses in pairs of stem-suffix sets cannot even prove a cause and e f f e c t r e l a t i o n s h i p between the sets, l e t alone i n f e r the means by which the r e l a t i o n s h i p took place. The pairs of stem-suffix sets, t h e i r positions ( r e l a t i v e to each other) i n the experimental task, the sound pattern cate-gories ( i . e . , parameters) considered and the results are l i s t e d in Table XII. The r e s u l t s indicate whether the values for the tetrachoric c o r r e l a t i o n c o e f f i c i e n t and for the chi-square mea-sure both remained at a confidence l e v e l better than p = .05. (The chi-square measure's values were taken into consideration as well since the tetrachoric c o r r e l a t i o n c o e f f i c i e n t reaches i t s maximum or minimum values of 1.0 or -1.0 when one entry in a matrix i s 0.) No conclusions can be drawn i n these data concerning the r e l a t i o n of the l e v e l s of significance of the pairs of stem-suffix sets to t h e i r location ( r e l a t i v e to each other) and to t h e i r common c h a r a c t e r i s t i c s . 4.4.3 The influence of d i f f e r e n t example derivations on  responses involving one stem-word J. Ohala (1973) exposed his subjects to two example d e r i -vations whose sound patterns d i f f e r e d i n such a way as to sug-gest that the VSR had been used in the derivation of one example (detain/detention) but not i n the other (explain/ explanatory). Several of his subjects' responses to d i f f e r e n t CO EH W CO w > H CO CO w u u CO EH S CO W d, •« HD w o PH O H Q H H g H CO EH EH CO W CO CO EH H CO >H w PAIRS OF STEM-SUFFIX SETS PARAMETERS GROUP BOTH % AND r VALUES REACH t .05 LEVEL OF SIGNIFICANCE (i d e n t i c a l vowel) define + - i t y / i o d i n e + - i t y tense vowel vs lax vowel A no it i i B no methane + - i t y / s u s t a i n + - i t y II A no i i H B no trade + -ity/space + - i t y II A no II II B yes sublime + - i f y / p r i d e + - i f y II A no II n B no (different vowel) iodine + -ity/methane + - i t y II A no B no ( i d e n t i c a l Vowel of consonant) bed + -ian/Fred + -ian II A yes II n B no Christmas + -ian/thermos + -ian II II A TD no toxic + -ism/public + -ism Ik] vs [s] E> A yes II •I B no zebra + -ic/Buddha + - i c tense vowel vs Deletion A no II II B yes (different vowel) define + -ity/methane + - i t y tense vowel vs lax vowel A yes n II B no (id e n t i c a l vowel) obtain + -ion/obtain + -atory II A yes II n B yes pertain + -ion/pertain + -atory II A no n ti B no probe + -ity/prose + - i f y i i A no it II B no Table XII. x and r. level s of significance of di f f e r e n t pairs of stem-suffix sets. -130-derivations (containing one stem-word) followed the sound pat-terns of the c o n f l i c t i n g example derivations. Since the sub-je c t s ' responses provided contradictory evidence for the use of a single i d e n t i c a l underlying form for a given stem-word, Ohala rules out the use of the phonological rules hypothesized by Chomsky and Halle. He does not consider the p o s s i b i l i t y that some other independent phonological rules were used. Instead, he turns to another means of production, that of analogical phonological rules, and concludes that t h i s was the means in use since the subjects' responses supported t h i s hypothesis. In the present study Group A's responses "followed" the lead-ing example derivations quite frequently. Nine of 17 subjects derived obtain and 10 of 17 subjects derived pertain by producing the lax vowel [£] when the preceding example was detain/detention and [ae] when the example was explain/explanatory. The fact that the vowels in the responses to the pairs of stem-suffix sets were not the same involved two implications for Chomsky and Halle's phonological theory. The responses could indicate e i t h e r (a) that Chomsky and Halle's hypothesized single underlying form was not used or (b) that the underlying form was used but t h e i r hy-pothesized rules were not applied to i t . However, r e j e c -t i o n of Chomsky and Halle's theory does not serve, as Ohala would have i t , as an unconditional warrant for accepting the theory of analogical rules. Such a "how-else" argument would preclude other p o s s i b i l i t i e s for the production of the sound patterns to have been influenced by the example derivations. One asks i f the subjects of Group A possibly f e l t obliged to -131-follow the example derivations, and i f so, one wonders what sound patterns they would have produced had they not f e l t t h i s way. An answer might be found by considering the responses of Group A which were not similar to the examples 1 sound patterns as well as the responses of Group B, whose subjects were not exposed to the examples. i One finds that the two groups derived each of the two pairs of stem-suffix sets inconsistently (that i s , without the same vowel in each member of the pair of derivations) in other ways than that mentioned above. Most often the pair of derivations involving a given stem contained a tense vowel i n one response and a lax vowel i n the other. This occurred i n 7 of the t o t a l 34 pairs of stem-s u f f i x sets i n Group A and i n 8 of the 30 pairs i n Group B. In only 3 of the 7 pairs i n Group A and i n 5 of the 8 pairs i n Group B did these inconsistent derivations agree with the sound patterns predicted by Chomsky and Halle. For instance, a tense vowel would appear i n the environment preceding CiV (e.g. [ par'telnja n] from pertain + ion) and a lax vowel would appear i n the environment be-fore CVCV (e.g. [ pe/te.natxn ] from pertain + -atory) . On the other hand, 4 of the 7 pairs i n Group A and 3 of 8 pairs i n Group B i n -volved 'tense' or 'lax vowel' responses (e.g. [ pa-'t eln j a n ] from pertain + -ion and [pa,'teinat:>jti ] from pertain + -atory i n hypotheti-c a l l y appropriate environments. Other sets of inconsistent responses included one derivation containing a tense vowel and the other derivation with a deleted s y l l a b l e i n Group A while i n Group B, 2 pairs of sets were comprised of one derivation with a lax vowel and the other with a deleted.syllable. A few subjects i n each group consistently used the same tense or lax vowel i n t h e i r responses to a pair of stem-suffix sets. In response to the pair of sets involving obtain, 3 -132-subjects in Group A produced 'tense vowel' sound patterns and 1 produced 'lax vowel' sound patterns (containing the vowel [£,]). In response to the pairs of derivations involving pertain, 3 subjects i n Group A and 8 i n Group B produced 'tense vowel* sound patterns. However, most (25 of 26) of the consistent pairs of responses mentioned above contain one member response which contradicts the sound patterns predicted by Chomsky and Halle (e.g. a tense vowel preceding CVCV or a lax vowel preceding CiV). While the occurrence of i d e n t i c a l vowels might have seemed to support the hypothesis for i d e n t i c a l underlying forms, the occurrence of unpredicted sound patterns in one member of 2 6 consistent pairs of responses showed contradictory evidence for the hypothesized phonological rules (VLR^ and VTR 2). Ohala (19 73, p.45) uses the low frequency of occurrence of the predicted sound patterns in his study as evidence that subjects infrequently used the phonological rules i n question. He believes, on the basis of the r e s u l t s of the tetrachoric c o r r e l a t i o n c o e f f i c i e n t measure of successive and nearby responses as well as the r e s u l t s of responses to obtain and pertain, that subjects were using analogical phonological rules. Ohala f i n a l l y concludes that "analogical phonological rules seem to be stronger in determining the form of a new derivation than are independent phonological rules" (1973, p.45). Recall, however, that i t was shown in t h i s study that no conclusions could be made on the p a r t i c u l a r means of sound pattern production by considering the occurrence of d i f f e r e n t sound patterns and the existence of high s t a t i s t i c a l c o rrelations. In other words, -133-Ohala's conclusion precludes the p o s s i b i l i t y that independent phonological rules (other than those formalized by Chomsky and Halle) or that some other means of sound pattern production, not yet proposed in l i n g u i s t i c theory, were being used by the subjects. 4.5 Further observations 4.5.1 Subjects' productivity with sound patterns and with phonological rules Other authors investigating the psychological r e a l i t y of phonological rules discuss whether the rules under investigation were frequently used by subjects and so were shown to be "productive. 1 1 In essence, they are saying that a hypothesized phonological rule, was used i f the sound pattern i t predicted occurred in subjects' responses. This form of argument which they use to make positive statements about the rule's use and i t ' s productivity commits a l o g i c a l f a l l a c y (discussed in Chapter 2) known as "the f a l l a c y of affirming the consequent." As pointed out by Caws (1965) , t h i s argument cannot be used to v e r i f y hypotheses. Therefore one cannot discuss whether the subjects are productive with the rules under inves t i g a t i o n . A p o s i t i v e statement about subjects' "productivity" w i l l have to be r e s t r i c t e d to t h e i r productivity with p a r t i c u l a r sound patterns. For example, one could discuss whether any of the predicted sound patterns have a 'special status' i n so far as they were produced, and could i n similar situations be expected to be produced, more often than other sound patterns. Then i f any -134-did have a 'special status,' i t would be interesting in further study to check for the frequent extension of these sound patterns in other situations to see i f subjects more generally have a special "awareness," whatever that might be, of these sound patterns. If, furthermore, subjects did extend these sound patterns i n other situations, then i t would be i n -teres t i n g to t r y to discover why they did. In order to assign a 'special status 1 to a sound pattern produced by subjects, i t i s necessary to decide on a minimum percentage of occurrence for the sound pattern. This c r i t e r i o n frequency of occurrence should depend on the probable d i s t r i b u -t i o n of the possible sound patterns for a given stimulus. However, i t i s not possible at t h i s time to know t h e i r probabil-i t y of occurrence since the number of sound patterns and t h e i r frequencies of occurrence depend on such factors as the pre-sence or absence of example derivations, on the various stem-s u f f i x sets, t h e i r number, as well as the time permitted for responding, the duration of the experiment, memory factors, etcetera. A more informal way to approach the problem would be to consider how many subjects produced the sound pattern pre-dicted by a given rule i n at least 2/3 of his/her responses. Two-thirds i s a severe c r i t e r i o n that would l i k e l y insure that the special status of a sound pattern was not overestimated. Two-thirds was chosen as the c r i t e r i o n since one-half would be just at chance l e v e l i n the case where only two sound -135-patterns were possible. Using t h i s c r i t e r i o n , the responses of Group B were considered as t h i s group seemed equally 'adept' at the production of novel derivations as Group A. In addition, they were not exposed to leading example derivations and so t h e i r responses may have been closer to the responses which would be produced by speakers i n a natural s i t u a t i o n when there i s a need for a novel derivation. Group B's responses to the long l i s t of stimuli were considered in t h i s case since i t i s preferable to know t h e i r production of sound patterns in response to the greatest number of s t i m u l i possible. Data for VLR^ and V T R 2 were derived from e l i g i b l e responses to obtain and pertain as well as from each rule's separate l i s t of stem-suffix sets. In t h i s case, the relevant data from both l i s t s of s t i m u l i were combined for each r u l e . Nine subjects reached the c r i t e r i o n of 6 7% in producing the sound pattern predicted by the PSR. Five subjects produced the 'tense vowel' sound pattern predicted by VTR^ in over 2/3 of t h e i r responses. Four subjects reached the c r i t e r i o n percentage in producing the velar softening sound pattern predicted by VSOR. No subjects produced the 'lax vowel' sound pattern predicted by VLR^, the 'tense vowel' pattern predicted by VTR2 nor the voiced-/s/ pattern predicted by /s/VR in 2/3 or more of his/her responses. To be safe, one could conclude that only that sound pattern which a majority of the subjects extended (that i s , the sound pattern predicted by PSR), by unknown means, in 2/3 or more of t h e i r responses could be given 'special status* -136-for the subjects of Group B. In addition, however, four subjects produced 6 responses to sets involving obtain and pertain that were relevant to VLF^. A l l 6 of the responses were the predicted 'lax vowel' sound pattern. It i s possible, then, that t h i s sound pattern may also be e l i g i b l e for special 'status' for the subjects of Group B. A v a l i d argument may be used in the case where the conse-quences are not what the hypothesis says they ought to be. In Caw's words, "while a hypothesis cannot be v e r i f i e d (that i s , shown to be true) by reference to i t s consequences, i t may be f a l s i f i e d conclusively i f the consequences f a i l to occur" (1965, p.112). So examining the responses in which the con-sequence ( i . e . , the predicted sound pattern) f a i l e d to occur w i l l indicate how often the hypothesized rule was not used. Therefore one may say how often subjects were 'unproductive' with a given ru l e . The percentage of responses (from the long l i s t ) which showed unpredicted sound patterns ( i . e . , any sound pattern other than the predicted one) i s l i s t e d for each rule i n Table XIII. Only the responses of Group B were considered as t h i s group was not submitted to the influence of leading example derivations. The data for each o£ VLR^ and VT^' from e l i g i b l e responses to pertain and obtain and from each rule's separate l i s t of s t i m u l i , were once again combined for each of those rules. The r e s u l t s show that, as a group, the subjects of Group B did not use the following rules i n a clear majority of the responses considered: VLR^ VTR2, VSOR and /s/VR. -137-RULE TOTAL NO. ELIGIBLE RESPONSES UNPREDICTED RESPONSES % UNPREDICTED RESPONSES VLR r 300 237 79 PSR 195 63 32.3 VTRj^ 45 32 71.1 VTR2 144 110 76.4 VSOR 60 40 66.7 VLR 2 6 0 0 /s/VR 31 31 100 Table XIII. The percentages of occurrence of unpredicted sound patterns i n the responses (long l i s t s ) of Group B which were relevant to each rul e . -138-The occurrence of unpredicted sound patterns was also con-sidered for each subject i n Group B to determine i f some sub-jects f a i l e d to use a given rule more often than other subjects. For Group B, the range in the percentages of occurrence of un-predicted sound patterns and the mean percentage are l i s t e d for each rule i n Table XIV. In t h i s case also, the data from the responses to obtain and pertain and that from each rule's separate l i s t of s t i m u l i were combined for each of VLR^ and VTR^. The ranges are quite broad for a l l rules except /s/VR and VLR 2. This suggests, f i r s t , that the subjects did vary in the number of responses i n which they f a i l e d to use a given r u l e . Secondly, since the subjects' lack of productivity with each rule varies, i t i s possible that t h e i r strategies for novel word derivations and hence t h e i r phonological models are not exactly the same. 4.5.2 Stem-suffix sets most and least often involved i n  predicted sound patterns I t was observed i n t h i s study that the stem-suffix set (within the group of those sets considered for each rule) which most often underwent the predicted phonetic change tended to be the same for each group of subjects. The stem-suffix sets which most often underwent the predicted phonetic changes (in each group) and the percentages of times that the sets underwent the changes are l i s t e d i n Table XV. One can probably conclude from the r e s u l t s l i s t e d i n the table that since the stem-suffix set (which most often showed the sound pattern predicted by a given rule) frequently remained the same in both groups' responses, the d i f f e r e n t -139-RULE % OCCURRENCE OF UNPREDICTED RESPONSES RANGE MEAN highest lowest 100 35 79 PSR 61..6 7.7 33.3 VTRX 100 33.3 71.1 VTR2 100 60 76.1 VSOR 100 25 66.7 VLR 2 0 0 0 /s/VR 100 100 100 AVERAGE DEVIATION 13.6 15.9 27.2 8.5 28.9 0 0 Table XIV. Ranges i n the percentages of unpredicted responses, the mean percentages and the average deviations from the means for Group B (long l i s t ) . RULE GROUP A GROUP B Short L i s t Long L i s t Short L i s t t Long L i s t stem-suffix set % stem-suffix set % stem-suffix set % stem-suffix set % :VLRX define + - i t y 76.4 76.4 60 60 — - — . -> PSR robot + - i a n 88.2 methane + - i t y iodine + - i t y robot + - i a n 100 94.1 88.2 robot + - i a n 100 methane + - i t y iodine + - i t y robot + - i a n 100 100 100 VTR 1 zebra + - i c 64.7 64.7. , / 6.6 — — ,— ^ 46.6 — — > VTR 2 47 64.7 64.7 47 64.7 64.7 20 6.6 40 ^ 20 cop + - i a n domestic + -ism p u b l i c + -ism — — > , / 46.6 VKQK > 40 — > VLR X* obtain + -io n 82 82 p e r t a i n + -ion 26 26 * • > VLR 2 Table XV. Stem-suffix sets whose responses most often showed the pre d i c t e d sound pattern. The % i s based on the number of times (out of the t o t a l number of responses to that set) that the set e l i c i t e d a p r e d i c t e d response. Note th a t /s/VR i s not l i s t e d since Group B d i d not produce any response p r e d i c t e d by i t . -141-experimental conditions did not play a strong role in deter-mining these sets. Rather, the given stem-suffix sets must have been determined as a re s u l t of the sound pattern for which they were considered, the stem-word or the s u f f i x com-p r i s i n g the set or a combination of these factors. For example, the set define + - i t y may have demonstrated the predicted 'lax vowel' sound pattern so often as a resu l t of the sound pattern being demonstrated i n the e x i s t i n g English derivation d i v i n i t y . This derivation i s comprised of a stem, which i s i n minimal phonemic contrast with define, and of the same s u f f i x . So the stem and the s u f f i x may have played a role i n inducing the lax vowel sound pattern. Another p o s s i b i l i t y i s that many of the ex i s t i n g English derivations in which the stem-word define i s involved contain a lax vowel in the second s y l l a b l e of the stem: e.g. d e f i n i t i o n , d e f i n i t e , d e f i n i t e l y , definiteness, d e f i n i t i v e . In t h i s case the stem i t s e l f may have played an important role in e l i c i t i n g the 'lax vowel' sound pattern since that stem con-tains a lax vowel when derived with several d i f f e r e n t s u f f i x e s . In the data from the short and the long l i s t s which were considered for each group of subjects, the stem-suffix set which lea s t often underwent the predicted phonetic change remained the same only in the case of two rules. In the sets considered for VLRX, the stem-suffix set bribe + - i t y showed the predicted change least often i n both groups' data from both the short and long l i s t s of st i m u l i . In the sets considered for the PSR, the stem-suffix set i n a l l four l i s t s of stimuli was -142-Buddha + - i c . .. This set also showed the least occurrence of phonetic change i n the short l i s t s ' data comprised from Group A and B's responses for VTR^. As in the case of the sets showing the most frequent oc-currence of predicted phonetic change, the sets consistently showing the predicted changes the least often may be determined from a combination of such factors as the c h a r a c t e r i s t i c s of a given set's stem-word and s u f f i x . For instance, the low frequency of Buddha + - i c undergoing the predicted change in primary stress and the predicted tense vowel change might be explained by the fact that the stem-word i s of Sanskrit o r i g i n while the s u f f i x combines with proper nouns usually only of Greek or Latin o r i g i n . Another p o s s i b i l i t y i s that none of the English derivations involving Buddha (e.g. Buddhist, Buddhistic, Buddhistical) r e t a i n and a l t e r the f i n a l vowel of the stem-word. 4.5.3 Suffixes and vowels most often involved i n predicted sound patterns" Steinberg and Krohn (1975) s p e c i f i e d the suffixes i n t h e i r study which were most often involved in derivations showing the predicted sound pattern. It was possible for them to do t h i s since i n t h e i r experiments an equal number of stem-words was derived an equal number of times with each s u f f i x . A l l of the derivations were considered for the. occurrence of a given sound pattern predicted by the VSR. The conditions of the present study prevent the suffixes most often involved i n predicted sound patterns from being s p e c i f i e d . Each s u f f i x was used a d i f f e r e n t number of times with the stem-words -143-considered relevant to each r u l e . None of the stem-words was paired an equal number of times with each s u f f i x . Steinberg and Krohn's experimental design also permitted them to i d e n t i f y which vowel of the stem-words most often underwent a predicted phonetic change. In t h e i r design, the given vowels were a l l (with one exception) represented an equal number of times i n the corpus of stem-words and were a l l i n -tended to be paired with each s u f f i x an equal number of times. In the present study, however, a l l of the given vowels oc-^ curring i n the l i s t of stem-words did not occur an equal number of times i n the experimental task l i s t nor i n those stem-words considered for each rul e . CHAPTER V DISCUSSION AND CONCLUSIONS Three secondary objectives for t h i s study, mentioned at the end of Chapter I, were f u l f i l l e d by the l i t e r a t u r e review and by the r e s u l t s discussed above in an unforeseen manner. F i r s t , in addition to stating "how,11 i t was stated "why" experiments on non-ideal speaker-listeners could apply to theories on ideal speaker-listeners. "How" such experiments could apply to theories on i d e a l speaker-listeners was l i m i t e d to a discussion of the experiment undertaken i n the present study and s h a l l be further discussed below. The reason "why" such experiments should be conducted i n r e l a t i o n to theories on ideal speaker-listeners was said to be e s s e n t i a l l y the lack of s c i e n t i f i c methodology i n generative l i n g u i s t i c theory and the consequently limited value of generative l i n g u i s t i c s to other s c i e n t i f i c f i e l d s of study. One might s t i l l f e e l uneasy about te s t i n g Chomsky and Halle's phonological theory outside the i d e a l i z e d speaker frame for the basic reason that t h i s procedure draws on some of the very notions (that i s , "behavioural" notions) which Chomsky set out to disprove. Any residual uneasiness might be d i s p e l l e d by considering Kuhn's (1962) b e l i e f that arguments for p a r t i c u l a r s c i e n t i f i c -144--145-paradigms are necessarily c i r c u l a r : When paradigms enter, as they must, into a debate about paradigm choice, t h e i r role i s necessarily c i r c u l a r . Each group uses i t ' s own paradigm to argue i n that paradigm's defense. The re s u l t i n g c i r c u l a r i t y does not make arguments wrong or i n e f f e c t u a l . The paradigms' arguments w i l l show what s c i e n t i f i c practice w i l l be l i k e for those adopting the paradigm (p.93). If one s c i e n t i f i c paradigm has reached the " c r i s i s " stage (wherein the e x p l i c i t and fundamental generalizations of the paradigm are questioned and other conditions, mentioned by Kuhn, occur), then a new paradigm—whether an e n t i r e l y new one or a compromise of former ones—should be adopted. The t e s t i n g of Chomsky and Halle's phonological theory i n the present i n v e s t i -gation i s based on the presumption,that, as i n c r i s e s of other s c i e n t i f i c paradigms, the present c r i s i s s i t u a t i o n i n v e r i f y i n g generative phonological theories (by st r u c t u r a l evidence) c a l l s for at least some compromise of the opposing paradigms. Kuhn suggests that "paradigm choice can never be unequivocally s e t t l e d by l o g i c and experiment alone" (1962, p.93). According to him, i n the end the question of paradigm choice i s answered in terms of c r i t e r i a outside of normal science by the assent of the relevant community who dictates which problems are more s i g n i f i c a n t to have solved. ''"Kuhn uses "paradigm" i n the sense of study patterns used, by s c i e n t i s t s i n a p a r t i c u l a r f i e l d of research. A paradigm i s characterized by the unprecedented achievement of s c i e n t i s t s employing the study patterns which "attracts an enduring group of adherents away from competing modes of s c i e n t i f i c a c t i v i t y " and by the open-ended qua l i t y of the study patterns which leaves a l l sorts of problems to be resolved (Kuhn, 1962, p.10). -146-Secondly, the s p e c i f i c a t i o n of e x p l i c i t c r i t e r i a which could serve to confirm or disconfirm the mentalistic claims of the phonological rules was discussed. While evidence was speci f i e d which could serve to disconfirm the use of phono-l o g i c a l rules, none could be given to confirm the use of the rules. This was due to the nature of the v a l i d l o g i c a l argu-ment, modus t o l l e n s , used i n s c i e n t i f i c investigation, which i s only able to disconfirm and thus eliminate incorrect hypotheses. I t was shown that confirmation and disconfirmation of the use of phonological rules were not conclusions which could stand i n a complementary relationship. In Other, words, in the case where the hypothesized use of a rule was not d i s -conf irmed, i t ' s use was not consequently confirmed. Thirdly, the l o g i c a l i m p o s s i b i l i t y of specifying evidence which would confirm the use of independent phonological rules (or any other means of sound pattern production) prevented the recognition, and hence the explanation, of the means of production which were used. The few attempts to explain the occurrence of certain sound patterns were, indeed, speculative and inexhaustive. Thus one of the main objectives of t h i s study, which was to draw conclusions on the v a l i d i t y of the experimental paradigm as a confirmation or disconfirmation procedure, was f u l f i l l e d . It was pointed out that the form of the l o g i c a l argument used in the investigation could at best be used to disconfirm given hypotheses. The other main objective, which was to decide -147-whether ce r t a i n phonological rules were psychologically r e a l for a given group of speakers, was more d i f f i c u l t to f u l f i l l . Three problems currently stand in the way of confirming or disconfirming the psychological r e a l i t y of the rules under investigation. One problem concerns the unknown maximum frequency with which a phonological rule can f a i l to be used i n the responses of a group of speakers (or in those of an i n d i v i d u a l speaker) and s t i l l be described as psychologically r e a l . This problem might be resolved i f generative phonologists could specify the extent to which hypothesized rules are used. (Chomsky and Halle (1968) merely state that t h e i r rules are applied to a l l prerequisite underlying structures and then give what i s pre-sumably an inexhaustive l i s t of exceptions to the r u l e s 1 application.) Before phonologists specify the exact extent to which phonological rules are used, the adoption of a maximum frequency for the non-use of a rule can only be ad hoc and, due to the problem to be discussed next, could cause one to draw fals e conclusions. The second problem i s that even i f the rule's extent of use was c l e a r l y s p e c i f i e d , a conclusion on the psychological r e a l i t y of the rule could not always be made. Indeed, the exact number of times that a given rule was not used i n a group of responses remains unknown when the predicted sound pattern occurs at least once in those responses. Therefore a s i t u a t i o n could occur i n which the rule's use was disconfirmed i n a number of responses x% short of the c r i t e r i o n percentage -148-and the predicted sound pattern (and the unknown means of production) occurred i n at least x% of the responses. For example, i f the c r i t e r i o n percentage of 'non-use' at which a rule was disconfirmed as psychologically r e a l was 80%, 77% of the responses could possibly contain unpredicted sound patterns while 23% could possibly show the predicted sound pattern. In the case, on the other hand, wherein the stimuli considered relevant to a rule e l i c i t e d unpredicted sound patterns i n 100% of the responses, one could state the exact number of times that the rule was not used by the group. In t h i s study t h i s oc-curred i n the case of the responses which were considered for the /s/VR. However, there was some poten t i a l evidence for the extension of the predicted sound pattern when one subject laxed a tense vowel preceding [z] and devoiced the [z] as well. This evidence could possibly weaken the certainty that the rule f a i l e d to be used at a l l by the group. A t h i r d problem involves the fact that the percentages of unpredicted sound patterns for the rules were derived from d i f f e r e n t numbers of responses. It would not be f a i r to con-sider the psychological r e a l i t y of the rules when the decisions for or against the psychological r e a l i t y were based on a d i f -ferent number of responses relevant to the rules and con-sequently on percentages (of unpredicted sound patterns) that did not vary in a s i m i l a r l y discrete way. S t i l l another d i f f i c u l t y which might be considered i n the attempt to determine rules* psychological r e a l i t y i s that the l o g i c a l argument modus t o l i e n s accepted for t h i s investigation -149-has a potential flaw which, i f Chomsky and Halle's i d e a l i z e d framework were not the frame under consideration, could allow the wrong conclusion to be drawn. Consider Caw's (1965) exemplification of the modus t o l i e n s argument. Someone hy-pothesizes that i f a man takes arsenic, he w i l l die. "Suppose the man...does not die: we may then conclude with certainty that he did not take arsenic, or at least not enough" (Caw, 1965, p.112). At least two conclusions are possible: either the man did not take arsenic or else he did not take enough arsenic to k i l l him. S i m i l a r l y , more than one conclusion could be drawn from the negative consequence of hypothesized phonological rules i f the rules were removed from Chomsky and Halle's i d e a l i z e d framework. For instance, suppose that the following points i n Chomsky and Halle's theory were abandoned: (a) that a l l speakers of an hypothesized homogenous speech community possess i d e n t i c a l phonological models; (b) that the use of a phono-l o g i c a l rule implies the use of a s p e c i f i c prerequisite under-lyi n g form; and (c) that a given rule i s applied to a l l underlying forms with prerequisite structures. Then i n the case where the predicted sound pattern did not occur, the argument modus t o l l e n s would be incapable of permitting a single conclusion disconfirming the use of the ru l e . A possible conclusion could be, as was suggested by Sherzer (19 70) and Zimmer (1969), that speakers of the same d i a l e c t have d i f f e r e n t phonological models. Therefore d i f f e r e n t rules or other strategies of sound pattern production could be used by -150-d i f f e r e n t speakers i n response to the same stimulus. This p o s s i b i l i t y might be explained by the idea put forth by Wang (1969), Hsieh (1972) and Chen and Wang (1975) that phonetic changes (and hence phonological rules) diffuse gradually across speakers' lexicons. Therefore speakers might not apply a given phonological rule to a l l of the word forms with the prerequisite underlying form. Nor might a l l speakers apply the same phono-l o g i c a l rule, or use the same strategy for sound pattern pro-duction, .in response to the same word form. In order to derive one conclusion (that i s , disconfirmation of the hypothesis) from negative consequences, the statement of the hypothesis should l i k e l y be quantified. For example, Caw's example of an hypothesis could be stated: If a man takes enough arsenic, he w i l l die. The hypothesis tested in t h i s investigation might be stated: I f and only i f a given phonological rule i s used, then a p a r t i c u l a r sound pattern w i l l be produced. The present investigation might be improved, f i r s t , by l i m i t i n g the time for subjects to respond in order to prevent them from repeating one form of a response to a stimulus several times, thus shortening the duration of the interview. A time l i m i t which might be generous enough to permit subjects to respond without making " s l i p s of the tongue" might be 4 seconds, the estimated period within which half of the responses were made by the two groups in the present study. Secondly, subjects' responses could be transcribed a f t e r , not during, the interview so that the transcriber might have a better -151-opportunity to judge how the phonetic forms of responses d i f -fered. Thirdly, those stem-words and suffixes which i n Chapter III were mentioned to have c h a r a c t e r i s t i c s inappropriate for t h e i r combination i n a derivation might be replaced by stem- . s u f f i x sets whose members did have appropriate c h a r a c t e r i s t i c s . Fourthly, the example derivation Kant/Kantian, showing no phonetic change i n i t s stem, was unfamiliar to most subjects and could be replaced with Marx/Marxian. One r e s u l t of t h i s investigation which warrants further study i s that both groups produced a predicted sound pattern (in those responses relevant to PSR and also in those relevant to VTR^) almost equally as often. One could check to see i f t h i s r e s u l t remained when d i f f e r e n t groups of subjects and/or longer l i s t s of s t i m u l i were involved. If the occurrence of predicted sound patterns continues to be very close for the two groups, one might suspect that the presence or absence of leading example derivations i s not influencing the groups' numbers of predicted responses. To investigate t h i s hypothesis, the leading example derivations preceding Group A's l i s t of s t i m u l i could be replaced to see i f other leading examples make a difference to the group's numbers of predicted responses. In addition, more than two groups could be submitted to the l i s t s of s timuli for each rule while exposing each group to a d i f f e r e n t number of leading examples. The r e s u l t s might suggest that each group's production of predicted sound patterns i s l imited in that t h i s production would not increase after a certain number of leading examples had been supplied them. -152-The trend for the stem-suffix set most often involved i n a predicted response to be the same in each group's l i s t s of data could be further investigated to determine i f such a trend occurs (a) when more subjects p a r t i c i p a t e i n the study and/or (b) when the l i s t s contain more stem-suffix sets. If such a trend did remain in the two conditions mentioned above, a few stem-suffix sets in a l i s t of s t i m u l i of given length could be replaced, one at a time, with other sets. This might serve as a check to see whether the number of times that sets were involved i n predicted responses remained f a i r l y constant. Such a check might also have implications for the p o s s i b i l i t i e s that the stem-suffix set most often involved i n predicted responses i s somehow self-determined or i s somehow determined by other sets in the group of s t i m u l i . On another occasion, the corpus of stem-suffix sets could be controlled for the number of sets which were relevant to each rule, the numbers of d i f f e r e n t strategic vowels represented i n the stem-words and the number of stem-words, as well as the number of stem-words containing given s t r a t e g i c vowels, assigned to each s u f f i x . This should help to obtain equal numbers of responses which are to be considered for the various rules and should f a c i l i t a t e the i d e n t i f i c a t i o n of the s u f f i x and of the strategic vowel which are most often involved i n predicted sound patterns. Another method which could be used to check for the ex-tension of predicted sound patterns would be to present sub-jects with e x i s t i n g or novel derivations and require them to -153-retrieve the component stem-word and s u f f i x . This method was used by Hsieh (1975) with the purpose of comparing i t s r e s u l t s with those of the novel derivation task. Although the consequences of the novel derivation task cannot be used to confirm any hypothesized means of sound pattern production, the consequences might be valuable as evidence for which sound patterns speakers are able to, or wish to, extend in certain phonetic contexts. Linguists might be interested i n knowing which sound patterns are or are not extended for the purpose of assigning d i f f e r e n t statuses to the phonological r e g u l a r i t i e s of a language which they describe. Such information could be used i n a grammar which i s at some intermediate position between ones which are purely f o r m a l i s t i c and ones which are mentalistic. It has been pointed out that the v a l i d i t y of the modus  to l l e n s argument, employed when the consequences of an hypothesis are negative, l i m i t s the conclusions that may be drawn i n the ontological experiment p a r t i c u l a r to t h i s study. The strategies of sound pattern production used by the subjects could not be determined. It was only possible to determine, through negative consequences, when the hypothesized rule was not used. Thus, when some of the consequences were p o s i t i v e , i t was impossible to determine the exact number of times that the hypothesis ( i . e . , the use of a given rule) was disconfirmed. It has been suggested above that one cannot specify a maxi-mum frequency for the non-use of a phonological rule, (above which that rule w i l l no longer be considered to be -154-psychologically r e a l ) , u n t i l the hypothesized extension of rules i s made clear by generative phonologists. Since a c r i t e r i o n frequency of non-use of a rule could not at t h i s time have been well motivated, the psychological r e a l i t y of given phonological rules for an i n d i v i d u a l or for a group of speakers could not be determined. If the modus to11ens argument were to be used with phono-l o g i c a l theories which do not employ certain ideals of Chomsky and Halle's (196 8) th e o r e t i c a l framework, then the unquantified statement of the argument's hypothesis would be compatible with the existence of more than one possible conclusion. With respect to the p a r t i c u l a r experimental design employed i n t h i s study, the similar d i s t r i b u t i o n of numbers of predicted responses i n each group showed that the two groups were sim i l a r i n t h e i r a b i l i t y to perform the experimental task, and were therefore representative samples of the population under study. Within each group, the o v e r a l l production of predicted sound patterns was not a t t r i b u t a b l e to just a few subjects. The o v e r a l l occurrence of predicted sound patterns was shown to be greater for the group of subjects (Group A) who were ex-posed to leading example derivations. Due to the two groups' similar a b i l i t i e s in handling the experimental task, the greater occurrence of the predicted responses in the t o t a l of Group A's responses was attributed to that group's exposure to leading example derivations. However, in the group's responses which were considered for just one rule at a time, the occurrence of predicted sound -155-patterns was not, in the case of each rule, much greater for Group A. The reason for the closeness in the two groups' numbers of sound patterns which were predicted by each of PSR and VTR^ requires further investigation. The trend for the same stem-suffix set (in both groups' long and short l i s t s of stimuli) to be most often involved in a predicted phonetic change could only be explained speculatively. Much further investigation would be required to determine whether such a trend would always exi s t and i f so, to successively eliminate some of the hypotheses proposed to account for the trend. It appeared that a small v a r i a t i o n i n the number of stem-s u f f i x sets which were considered for a rule did not greatly influence subjects' mean ranks for the o v e r a l l production of predicted responses. Since the p r o b a b i l i t y of occurrence of sound patterns could not be determined, the procedure used to assign a 'special status ' to certain predicted sound patterns (those predicted by PSR and possibly also that predicted by VLR„) was ad hoc. -156-BIBLIOGRAPHY Botha, R. 196 8. The Function of the Lexicon i n Transformational Generative Grammar. (Janua Linguarum, Series Maior, Nr. 38.) The Hague: Mouton & Co. Botha, R. 19 70. The Methodological Status of Grammatical Argumentation" (Janua Linguarum, Series Minor, Nr. TO5.) The Hague: Mouton & Co. Botha, R. 1971. Methodological Aspects of Transformational  Generative Phonology. (Janua Linguarum, Series Minor, Nr. 112.) The Hague: Mouton & Co. Carterette, E. and M. Friedman, eds. 1974. Handbook of Perception, Vol. One. H i s t o r i c a l and Philosophical Roots  of Perception. New York: Academic Press. Caws, P. 1965. The Philosophy of Science, a systematic account. Princeton: Van Nostrand. Chen, M. and W. Wang. 1975. 'Sound change: a c t i v a t i o n and implementation.' Language 51, 2 (June): 255-281. Chomsky, N. 1964. Current Issues i n L i n g u i s t i c Theory. (Janua Linguarum^ Series Minor, Nr.38.) The Hague: Mouton & Co. Chomsky, N. 1965. Aspects of the Theory of Syntax. Massachusetts: M.I.T. Press. Chomsky, N. 196 8. Language arid Mind. New York: Harcourt, Brace and World. Chomsky, N. and M. Halle. 1968. The Sound Pattern of English. New York: Harper and Row. Dolby, J. and H. Resnikoff. 11967. The English Word Speculum. Vol. Three. The Hague: Mouton & Co. Esper, E. 1973. Analogy and Association i n L i n g u i s t i c s and Psychology. Athens: University of Georgia Press. Fischer-J0rgensen, E. 19 75. 'Perspectives in phonology.' Annual Report, Institute of Phonetics. ARIPUC 9: 215-236. University of Copenhagen. Greene, J. 1972. Psycholinguistics, Chomsky and Psychology. England: Penguin Books, Ltd. Hockett, C. 1968. The State of the Art. (Janua Linguarum, Series Minor, Nr~ 73.) The Hague: Mouton & Co. Hsieh, .H.-I. 1972. 'Lexical d i f f u s i o n : evidence from c h i l d language a c q u i s i t i o n . ' Glossa 6, 1: 89-103. -157-Hsieh, H.-I. 19 75. 'How generative i s phonology?' The Transformational-Generative Paradigm and Modern L i n g u i s t i c  Theory, ed. E. Koerner, et a l . Amsterdam: John Benjamins B.V. Krohn, R. 1972. 'The vowel s h i f t rule and i t s productivity.' Language Sciences 20: 16-18. Kuhn, T. 1962. The Structure Of S c i e n t i f i c Revolutions. Chicago: University of Chicago Press. Ladefoged, P. 19 70. 'The phonetic framework of generative phonology. 1 Working Papers i n Phonetics-UCLA 14 (March): 25-32. Lindblom, B. 1971. Numerical Models i n the Study of Speech  Production and Speech Perception. Some Phonological  Implications. Department of Speech Communication, Royal Institute of Technology, 10044 Stockholm 70, Sweden and Institute of L i n g u i s t i c s , Department of Phonetics, University of Stockholm. L i n e l l , P. 1974. Problems of Psychological Reality i n Generative Phonology. A C r i t i c a l Assessment. Reports from Uppsala University Department of L i n g u i s t i c s (RUUL), Nr. 4. Lyons, J. 1970. Chomsky. Great B r i t a i n : Wm. C o l l i n s & Co. Ltd. Marchand, H. 1960. The Categories and Types of Present-Day  English Word-Formation. Wiesbaden: Otto Harrassowitz. Nicholson, G. 1916. English Words with Native Roots and with  Greek, Latin or Romance Suffixes. Chicago: University of Chicago Press. Ohala, J. 19 70. 'Aspects of the control and production of speech.' Working Papers i n Phonetics-UCLA 15. Ohala, J . 1973. 'On the design of phonological experiments I I . ' Mimeographed version of an unpublished paper read at the Winter 1972 L i n g u i s t i c Society of America meeting in Atlanta. Ohala, J. 1974. 'Phonetic explanation in phonology.' Paper delivered at the Natural Phonology Parasession, Chicago L i n g u i s t i c Society, 18 A p r i l , 1974. Ohala, M. 1974. 'The abstractness controversy: experimental input from Hindi.' Language 50, 2 (June): 225-235. Onions, C , ed. 1966. Oxford English Dictionary of Etymology. London: Oxford University Press. -158-Th e P r i n c i p l e s of the International Phonetic Association. 1949 tReprinted LlTf£) . international Phonetic Association, University College, Gower Street, London WC1E 6BT. Sherzer, J. 1970. 'Talking backwards i n Cuna: the socio-l o g i c a l r e a l i t y of phonological descriptions.' Southwestern  Journal of Anthropology 26: 343-353. Steinberg, D. 1975. 'Chomsky: from formalism to mentalism and psychological i n v a l i d i t y . ' Glossa 9, 2: 218-252. Steinberg, D. and R. Krohn. 1975. 'The psychological v a l i d i t y of Chomsky and Halle's vowel s h i f t r u l e . ' The Transforma-tional-Generative Paradigm and Modern L i n g u i s t i c Theory, ed. E. Koerner, et a l . Amsterdam: John Benjamins B.V. Wang, W. 1968. 'Vowel features, paired variables, and the English vowel s h i f t . ' Language 44, 4 (December): 695-708. Wang, W. 1969. 'Competing changes as a cause of residue." Language 45, 1 (March): 9-26. Zimmer, K. 1969. 'Psychological correlates of some Turkish morpheme structure conditions.' Language 45, 2 (June): 309-321. APPENDIX I Pilot I DIRECTIONS AND LIST 0? WORD FORMS FOR EACH EXPERIMENT Pilot II Main Directions I would like your help in a project of mine which is to construct an extended or • extrapolated dictionary of English. A 3 you know, English allows the formation of new words by the addition of suffiexes or endings such as noss, - i t y and so on to the end of existing words. For example, odd can have -i t y added to i t to give oddity. We'll proceed as follows: I ' l l give a particular suffix, then give some examples of i ts use with existing words, and then" give a number or words which that suffix lias not been added to before. (Later on, I ' l l give just the suffix and then words to which the suffix has never been added.)! Would you please pronounce the word that would result from the addition of that suf- . fix, t e l l me what you think this new word might mean, and then t e l l me whether or not you would be likely to use this new word. I would like your help in a project of mine which is to construct an extended or extrapolated dictionary of English. As you know, English allows the formation, of hew words by the combination of suffixes or endings with existing word3. For example, the word odd combined with the ending -ity gives oddity. We'll proceed as follows: to begin with, I ' l l give a particular ending, then give some examples of its use with existing words, and then give a number or words which that ending has never been combined with before. Later on, though, I ' l l give just the ending and the words which the ending is to be com-bined with.)2 So, to help me in my survey, would you please take the word and the end-ing, put them together somehow and pronounce the new word. (Then would you t e l l me what y.ou think the new word would mean and lastly t e l l me whether or not you would be likely to use this new word.)' Kow let's return to the new words you proposed. I ' l l give the ending, the word, and the new word you suggested. Thi3 time, to help me in my survey, would you t e l l me what you think this new word;would mean and then t e l l me whether or not you would be likely to use this new word. May I remind you that this exper-iment is like a survey in that the res-ponses that you give are neither right nor wrong; rather, they are a l l valid. I would like your help in a project of mine which is to construct an ex-tended or extrapolated dictionary of English. As you know, English allows the formation of new words by the com-bination of suffixes or endings with existing words. For example, the word odd combined with the ending - i t y gives oddity. We'll proceed as follows: I ' l l give a particular ending, then give a number of words which that ending has never been combined with before. To help rae in my survey, would you please take the word and the ending, put them together somehow and pronounce the new word. Then would you t e l l me what you think the new word would mean, and lastly t e l l me whether or not you would be likely to use this new word. 1 The directions to Groups a' and b' in Pilot I were identical except that the sentence enclosed^in parentheses was included in the directions to Group b* only. 2 The directions to Group b " in Pilot II included the sentence in parentheses. The directions to Group a'' did not. 3 The directions to Croup b " in Pilot II ommitted the sentence in parentheses and included the next paragraph. Pilot I Word Lists Practice Task Experimental. Task *-y' burglar master rider mover fa l l e r ' - 0 U 3 1 pore industry worm bubble o i l • - i c ' scene rhapsody-ovary celery, nunnery '-hood' brother-father Pilot II burglary mastery (the same) porrous industrious (the same) scenic rhapsodic (the Same) (the same) brotherhood fatherhood (the same) apple scholar normal Main burglar master, burglary mastery rider mover faller pore industry porrous industrious O I worm bubble o i l scene angel scenic angelic ovary celery nunnery brother. brotherhood father fatherhood apple scholar normal Pilot I '-ity' odd stupid shred fad between probe rod , trade space bribe mop fleece '-dom' king |(serf) bird man house joy ••"dan' Darwin music trade crime bed Thomas prude robot Fred I human oddity stupidity^ kingdom _ (serfdom) Darwinian musician Pilot II (the sane) (the same) (the same) (the same) (the same) (the same) odd (sane) shred fad between probe rod trade space bribe mop fleece king (serf) bird man house joy < Cant (mammal) trade crime bed Thoma3 prude robot Fred human Main, oddity c (sanity) kingdom (serfdom) Kantian (mammalian) 4 Recall that no example derivations were given to group b in the two pilots. 5 The example derivations enclosed in parentheses were witheld from Group B in the main experiment. Pilot I Pilot II rify» beauty solid slave sublime pride supreme prose trample ?'-ion' fuse detain obtain cruise pertain pact -ism' fad radical school buck sofa • domestic ' brain toxic adverse public ' -ment' amend (agree) face rate flinch beautify solidify fusion detention faddism radicalism amendment (agreement) (the same) (the same) (the - same), (the same) (the same) (the same) (the same) (the same) Main false falsify (rate) (ratify) slave sublime pride supreme prose trample rebel rebellion (detain) (detention) obtain cruise pertain pact fad faddism (critic) (criticism) school buck sofa domestic brain toxic adverse public (the same) face rate flinch Pilot.II Pilot II '-ian' Darwin music thermos artistic cop Christmas -i c ' artist realist dust zebra biologist Buddha incite '-atory' conserve (explain) obtain sapphire pertain '-ity' odd stupid define iodine methane sustain trite Darwinian musician ar t i s t i c realistic conservatory (explanatory) oddity stupidity •(the same) (the same) (the same) (the same) (the same) (the same) (the same) (the same) Kain Kant .'Kantian (mammal) (mammalian) thermos artistic cop Christmas graph graphic (algebra) (algebraic) dust zebra biologist Buddha incite conserve conservatory (explain) (explanatory) obtain sapphire pertain odd (sane) define iodine methane sustain trite oddity (sanity) -164-APPENDIX II STEM-SUFFIX SETS CONSIDERED FOR EACH RULE Rule: V -> [-tense]/ CVCV between trade + space + bribe + fleece + define + iodine + methane sustain t r i t e + probe + * i n c i t e + sapphire slave + sublime prose + pride + + - i t y - i t y - i t y - i t y - i t y - i t y - i t y + - i t y + - i t y - i t y - i t y - i c + -atory - i f y + - i f y - i f y - i f y Rule: V -> [1 Stres s ] / C + a f f i x Thomas + -ian robot + -ian human + -ian thermos + -ian a r t i s t i c + -ian Christmas + -ian *adverse + -ism zebra + - i c b i o l o g i s t + - i c Buddha + - i c *sapphire + -atory *iodine + - i t y *methane + - i t y Rule: V [ttense]/ V Rule: *sofa + -ism zebra + - i c Buddha + - i c Rule : /k Q/ -» s / _ I domestic + -ism toxic + -ism public + -ism * a r t i s t i c + -ian Rule; Underlying vowels i n pertain and obtain ~~ obtain + -ion pertain + -ion obtain + -atory pertain + -atory V [-high] [+tense]/ CiV bed + -ian Thomas + -ian robot + -ian Fred + -ian human + -ian cop + -ian thermos + -ian Christmas + -ian /s/ [+voice]/ V V [ttense] *space + - i t y *fleece + - i t y *Thomas + -ian *thermos + -ian *Christmas + -ian * a r t i s t i c + -ian "marks those stem-suffix sets which were not preceded i n Main Experiment A by an example derivation demonstrating the sound pattern predicted by the rule for which the sets were considered. o 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0093706/manifest

Comment

Related Items