UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A phonological description of contemporary literary ukrainian Andersen, Henning 1962

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

Item Metadata


831-UBC_1962_A8 A5 P4.pdf [ 7.33MB ]
JSON: 831-1.0105733.json
JSON-LD: 831-1.0105733-ld.json
RDF/XML (Pretty): 831-1.0105733-rdf.xml
RDF/JSON: 831-1.0105733-rdf.json
Turtle: 831-1.0105733-turtle.txt
N-Triples: 831-1.0105733-rdf-ntriples.txt
Original Record: 831-1.0105733-source.json
Full Text

Full Text

A PHONOLOGICAL DESCRIPTION OP CONTEMPORARY LITERARY UKRAINIAN by KENNING ANDERSEN B.A., Uni v e r s i t y of B r i t i s h Columbia, 1961 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS i n the Department of Slavonic Studies We accept t h i s t h e s i s as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September, 1962 In presenting t h i s thesis i n p a r t i a l f u l f i l m e n t of the requirements for an advanced degree at the University of B r i t i s h Columbia, I agree that the Library s h a l l make i t f r e e l y available for reference and study. I further agree that permission for extensive copying of t h i s thesis f o r scholarly purposes may be granted by the Head of my Department or by his representatives. It i s understood that copying or publication of th i s thesis f o r f i n a n c i a l , gain s h a l l not be allowed without my written permission. Department of Slavonic Studies The University of B r i t i s h Columbia, Vancouver 8, Canada. Date September 14th, 1962 ABSTRACT The present t h e s i s presents the r e s u l t s of a pre-liminary i n v e s t i g a t i o n of the phonology of contemporary l i t e r a r y Ukrainian (CLU). Based on a morphological analysis of the material contained i n H. Holoskevyc, Pravopysny rj  slovnyk, i t describes the orthoepic standard set f o r t h i n 0. Synjavs'kyj, Normy ukrajins'ko.ji l i t e r a t u r n o j i movy (with due consideration given to the current Soviet norms) within the framework of the theory of phonology formulated by Morris Halle (The Sound Pattern of Russian). Chapter II contains the d e s c r i p t i v e statements under three headings: Segments and Boundaries, Morpheme Structure Rules, and Phonological Rules. The segments are defined i n terms of d i s t i n c t i v e features as follows: j i i y y n u e e a a o 6 d d t | n n 3 5 c c z z s § b p v f m 3 C z s g k h x r r ^ Vocalic -++++++++++++ ++++ Compact o ++++++ + + + + + + + + 0 0 0 0 P l a t o ++ + + 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Grave o — + + 0 0 — + + 0 0 — + + + + + + + + + 0 0 0 0 Strident 0 0 0 0 0 0 0 0 0 0 0 0 0 + + + + + + + + 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Nasal 0 0 0 0 0 0 0 0 0 0 0 0 0 + + 0 0 0 0 0 0 0 0 + 0 0 0 0 0 0 0 0 0 0 0 0 Continuous 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + + + + — + + 0 — + + — + + — + + Tense 0 0 0 0 0 0 0 0 0 0 0 0 0 — + + 0 0 — + + — + + - + - + 0 - + - + - + - + 0 0 0 0 Sharped 0 0 0 0 0 0 0 0 0 0 0 0 0 - + - + - + - + - + - + - + 0 0 0 0 0 0 0 0 0 0 0 0 0 - + - + Accented 0 - + - + - + - + - + - + 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Chapter III presents a discussion of the i n d i v i d u a l morphonemes and boundaries. Among the problems discus'sed are: the morphonemic representation of geminates and of i i i distinctively sharped labial and palatal consonants: the distribution of sharped and plain consonants before / i / in different varieties of CLU; the distribution of / i / and /y/ word i n i t i a l l y and after / j / ; earlier attempts to reduce the inventory of vowel phonemes to five; the synchronic and dia-chronic status of the feature tense vs. lax; the status of the marginal phonemes / f / , /"$/, /j>/* /s/» Chapter IV surveys and discusses the material on which the morpheme structure rules are based and offers a few incidental comments on problems connected with their formulation. Chapter V primarily illustrates the operation of the phonological rules and discusses their order. A concluding section contains a brief discussion of the optional phono-logical rules which describe deviations from Synjavs'kyj»s Normy. ACKNOWLEDGEMENTS In the preparation of the present paper I have received assistance from many parts, and I am pleased to acknowledge my debt to them. I wish to thank, the many members of the Ukrainian community of Vancouver who have helped me i n various respects. I am p a r t i c u l a r l y g r a t e f u l to Dr. M. L. Huculak, and to Mr. and Mrs. Ivan Reshetnyk, from whom valuable text material was c o l l e c t e d . I am g r a t e f u l to the Ukrainian Professional and Business Men's Club of Vancouver f o r the generous f i n a n c i a l assistance I received, and I am e s p e c i a l l y g r a t e f u l to the president of t h i s organization, Dr. John Yak, who took a personal i n t e r e s t i n my work and helped me i n every way. To the s t a f f members of the Uni v e r s i t y L i b r a r y who went out of t h e i r way to a s s i s t me I o f f e r my warmest thanks. I am most deeply g r a t e f u l to my teacher, Dr. J . 0. St. C l a i r - S o b e l l , who introduced me to the f i e l d of S l a v i c l i n g u i s t i c s and f i r s t awakened my i n t e r e s t i n the Ukrainian language. September 1962 Henning Andersen CONTENTS CHAPTER PAGE I. INTRODUCTORY REMARKS . . . . . . 1 A. Contemporary Literary Ukrainian 1 B. Theoretical Background . . . . . 2 C. Limitations 5 D. Organization 7 E. Graphic Conventions 8 II. DESCRIPTIVE STATEMENTS 11 A. Segments and Boundaries 11 B. Morpheme Structure Rules . . . . . 13 C. Phonological Rules 16 III. COMMENTS ON THE SEGMENTS AND BOUNDARIES 21 A. Comments on the Segments 21 1.0 The Glide 21 2.0 The Vowels 38 3.0 The Consonants 61 3.1 The Labial Consonants 72 3.2 The Dental Consonants . 79 3.3 The Palatal Consonants 87 3.4 The Velar Consonants 89 4.0 The Liquids 96 5.0 The Distinctive Features . 97 B. Comments on the Boundaries 100 CHAPTER PAGE IT. COMMENTS ON THE MORPHEME STRUCTURE RULES . . . . 104 V. COMMENTS ON THE PHONOLOGICAL RULES . . . . . . . 117 BIBLIOGRAPHY 137 LIST OF FIGURES FIGURE PAGE I. Branching Diagram Representing the OLU Morphonemes 11 II. Matrix Representing the Morphonemes of CLU . . . 12 III. The Six Accented Towels of CLU in Unchecked Position 39 IT. The Occurrence and Non-occurrence of Towels Following Non-vowels and Pause, and of Non-vowels Preceding Towels and Pause . . . . . 47 CHAPTER I INTRODUCTORY REMARKS The present paper presents some of the results of a preliminary investigation into the phonology of contemporary literary Ukrainian. A full-scale study of contemporary literary Ukrainian was, for practical reasons, impossible within the scope of this investigation, and a number of limitations had to be imposed at various points. These limitations are discussed below (Chapter I-C). First, however, i t is necessary to clarify the sense in which the terms "contemporary literary Ukrainian" and "phonological description" are used in the present paper. A. CONTEMPORARY LITERARY UKRAINIAN Contemporary literary Ukrainian (hereafter referred to as CLU) is the of f i c i a l language of the Ukrainian Soviet Socialist Republic and the literary language of Ukrainians outside the Soviet Union. The present description of CLU is essentially based on the orthoepic norms described by Oleksa Synjavs'kyj in his Normy ukrajins*koji literaturnoji movy.^  These norms are part of the literary standard adhered to by Ukrainians 1 Second edition, L'viv, Ukrajins'ke£,Vydavnyctvo, 1941, pp. 5-37 and 168-176. 2 p outside the Soviet Union, but have been found i r r e c o n c i l -able with the objectives of Marxist-Leninist educational policieis and, consequently, relinquished i n the Soviet Ukraine.^ As a r e s u l t of t h i s , i t i s necessary to reckon with several v a r i e t i e s of CLU. The d i f f e r e n c e s among these are d i f f e r e n c e s i n the d i s t r i b u t i o n of phonemes and are, within the t h e o r e t i c a l framework adopted f o r the present d e s c r i p -t i o n , r e f l e c t e d i n the existence of a l t e r n a t i v e phonological r u l e s . For the sake of c l a r i t y , only the r u l e s describing the orthoepic norms set f o r t h by Synjavs'kyj have been included i n Chapter II-C (The Phonological Rules). The a l t e r n a t i v e r u l e s are given i n the course of the discussion of the phonological r u l e s i n Chapter V. B. THEORETICAL BACKGROUND The present d e s c r i p t i o n i s an attempt to apply to the 2 See, f o r example, G. Y. Sheyelov (pseud. J u r i j Serex), Narys sucasiioji ukra.jins'ko.ji l i t e r a t u r n o j i movy, Munich, Molode Z y t t j a , 1951, p. 12. 3 The,standard d e s c r i p t i o n of the current Soviet norms, Kurs sucasnoji ukra.jins'ko.ji l i t e r a t u r n o j i movy, ed. L. A. Bulaxovs'kyj (Kyjiv, 1955), unfortunately was not a v a i l a b l e . The Soviet norms are^ however, described i n such textbooks as M. A. Zovtobrjux, Sucasna ukrajins'ka l i t e r a t u r n a mova, K y j i v , Radjans'ka Skola, 1961; and 0. M. Parxomenko, U k r a j i n -s'ka mova, I, K y j i v , Radjans'ka Skola, 1961; as w e l l as i n recent a r t i c l e s on the subject ( c f . , f o r instance, the a r t i c l e s by F. T. Zylko, l i s t e d i n the b i b l i o g r a p h y ) . data of CLU the theory of phonology formulated by Morris Halle in his The Sound Pattern of Russian.^" This theory of" phonology, which has been integrated into the theory of 5 language^ proposed by Noam Chomsky, is embodied in six for-mal conditions which phonological descriptions must satisfy. They are summarized in the following. Condition 1. Speech events are represented as sequences of segments, characterized by phonetic properties, and boundaries, characterized by their effects on the former. Condition 2. The phonetic properties that charac-terize segments belong to a restricted set of such proper-ties, the binary distinctive features. Condition 3. The phonological description, provides a method for deriving the utterance symbolized from every phonological representation without recourse to information not contained in the phonological representation. Condition 4. The phonological description is inte-grated into the grammar of the language in such a way that i t facilitates simple statements of a l l grammatical operations. 4 "A Linguistic and" Acoustical Investigation, 's-Graven-hage, Mouton & Co., 1959, pp. 19-44. 5 Syntactic Structures (Vol. IV of Janua Linguarum, ed. C. H. van Schooneveld), 's-Gravenhage, Mouton & Co., 1957. Cf-7 also the review by R. B. Lees, Language, XXXIII (1957), pp. 375-408, and, for a more recent selected bibliography, Condition 5. In phonological representations the number of s p e c i f i e d features i s c o n s i s t e n t l y reduced to a minimum compatible with s a t i s f y i n g Conditions 3 and 4. Condition 6. A l l phonological boundaries correspond to • morpheme boundaries. The theory s p e c i f i c a l l y r e j e c t s a condition which has played an important r o l e i n t r a d i t i o n a l American l i n g u i s t i c s , v i z that the phonological d e s c r i p t i o n provide i n s t r u c t i o n s f o r i n f e r r i n g the correct phonological representation of any speech event without recourse to information not contained i n the phy s i c a l s i g n a l . This condition i s r e f e r r e d to as 6 Condition 3a. An exposition of the considerations that make these s i x conditions necessary, as w e l l as a discussion of t h e i r main implications, i s given i n Sound Pattern (pp. 19-44) and w i l l not be attempted here. Where appropriate i n the discu s s i o n below (Chapter I I I to V), the relevance of the conditions f o r the solutions adopted i n the present des-c r i p t i o n w i l l be mentioned, and a few i n c i d e n t a l comments the l a t t e r ' s "Gto takoe transformacija," Yoprosy jazyko-znanija, X:3 (1961), pp. 69-77. 6 The conditions l i s t e d above are introduced i n the d i s -cussion i n Sound Pattern as follows': 1 and 2, on p. 19; 3 and 3a, on p. 21; 4, on p. 24; 5, on p. 29 f . j 6, on p. 41 • 5 on some of t h e i r implications w i l l be offered. C. LIMITATIONS The t h e o r e t i c a l framework adopted f o r the present d e s c r i p t i o n requires that the phonological d e s c r i p t i o n be based on a f u l l - s c a l e a n a lysis of the language i n question. This i s necessary, f o r instance, i n order that morpheme structure r u l e s of a s u f f i c i e n t degree of ge n e r a l i t y can be formulated. I t i s also necessary i n order to s a t i s f y Con-d i t i o n 4. A f u l l - s c a l e study of CLU was c l e a r l y excluded by the scope of the present i n v e s t i g a t i o n . As a consequence a number of l i m i t a t i o n s of the present d e s c r i p t i o n have to be mentioned. To make i t possible f o r the phonological d e s c r i p t i o n to s a t i s f y Condition 4, a preliminary survey of CLU i n f l e c -t i o n and d e r i v a t i o n was made. The sample analysed was H. Holoskevyc's orthographic d i c t i o n a r y , which, with i t s some 40,000 e n t r i e s , constitutes the r i c h e s t e a s i l y acces-7 s i b l e source of information on CLU morphology. It i s on t h i s sample that the morpheme structure r u l e s (Chapter II-B) have been based. It i s po s s i b l e , though not probable, that an expansion of the corpus w i l l necessitate a r e v i s i o n of these r u l e s . 7 Pravopysnyj slovnyk, 8-e vyd., New York, Knyhospilka, 1955. 6 The non-phonemic d i s t r i b u t i o n of prominence i n the unaccented s y l l a b l e s of the Ukrainian phonological phrase has, to t h i s writer's knowledge, not been investigated since T. Lehr-Splawiiiski's study "Z fonetyki maioruskie j " i n 1916, based on the speech of eight informants from various parts of eastern G a l i c i a . I t i s not c e r t a i n that the r e s u l t s of. t h i s study, which are summarized by Jan Z i l y i i s k i i n h i s o chapter on Ukrainian s t r e s s , are representative of CLU. A new study of Ukrainian s t r e s s , based on the current norms of CLU, i s very d e s i r a b l e , but such a study could not be under-taken within the present i n v e s t i g a t i o n . Condition 2 requires that a l l segments be defined i n q terms of the d i s t i n c t i v e features. I d e a l l y , the feature composition of the CLU morphonemes would be established on the basis of measurements of t h e i r acoustic properties. While t h i s could not be done i n the present i n v e s t i g a t i o n , i t i s hoped that the d e f i n i t i o n s proposed i n t h i s d e s c r i p -t i o n w i l l serve as a valuable point of departure f o r a future acoustic study of the CLU segments. 8 Opis fonetyczny jezyka ukrairiskiego. Krakow, Gebethner i Wolff, 1932. 9 Cf. Roman Jakobson, C. Gunnar M. Fant, and Morris H a l l e , P r e l i m i n a r i e s to Speech Analysis; M»I.- T» Acoustics Laboratory  Technical Report N Q . 15, 2nd p r i n t i n g , Cambridge (Mass.), Massachusetts I n s t i t u t e of Technology, 1952; and Roman Jakobson and Morris H a l l e , Fundamentals of Language ( V o l . I of Janua Linguarum, ed. C. H. van Schooneveld^, 's-Graven-hage, Mouton & Co., 1956. 7 D. ORGANIZATION A phonological d e s c r i p t i o n based on Morris Halle's theory of phonology consists of d e s c r i p t i v e statements of two s o r t s : an inventory of the segments and boundaries i n terms of which any utterance i n the language i n question can be represented (Condition l ) ; and a number of r u l e s by means of which any utterance (conveniently i n a phonetic t r a n s c r i p -tion) can be derived from i t s phonological representation (Condition 3 ) . In conformity with Condition 5, any feature i n a given segment which can be i n f e r r e d from other features i n the same segment, or i n other segments i n the context j should, ^ e l e f t unspecified i n the phonological representa-t i o n and assigned by a r u l e . In order that t h i s p r i n c i p l e can be applied c o n s i s t e n t l y to ensure maximal economy with-out Condition 4 being v i o l a t e d , i t i s necessary that the ru l e s be divided into two sets, the morpheme structure r u l e s and the phonological r u l e s . The l a t t e r , which describe the sequential constraints that hold f o r segment sequences i n general, are applied a f t e r the morphological r u l e s . The former, which describe the automatic d i s t r i b u t i o n of features within i n d i v i d u a l morphemes, are applied before the morphological r u l e s , i . e . as soon as the l e x i c a l morphemes have been selected from the d i c t i o n a r y . 8 The resulting tripartition of the descriptive state-ments (A. segments and boundaries; B. morpheme structure rules; C. phonological rules) is reflected in Chapter II of the present description. The three subdivisions of Chapter II are commented on in Chapters III, IV and V respectively. For the purpose of facilitating unavoidable cross-references, these chapters have been subdivided into sections, to which decimal numbers have been assigned. This system of numbering follows the generally accepted conventions (e.g. Sec. 1.32 follows See. 1.3134; Sees. 1.321, 1.322, and 1.323 can be referred to en bloc as Sec. 1.32). E. GRAPHIC CONTENTIONS The phonetic transcription used in the present paper is an adaptation of that used by Olaf Broch. 1 0 It differs from Broch's and/or the alphabet of the International Phonetic Association in the following respects: The symbols a, e_, 1, o, u, j_ have the values des-cribed in Chapter III-A, Sec. 2.11. Raising and lowering are denoted by adscript diacritics (thus [y"] is a loweredf [y*] a raised variety of [y]). Moderate and strong fronting ..10 Slavische Phonetik (Sammlung slavischer Lehr- und Hand-bucher, herausg. v. A. Leskien u. E. Berneker), Heidelberg, Carl Winter, 1911. 9 are denoted by a superscript dot and diaeresis respec-tively (e.g. [a], [a]). Accented vowels are distinguished from corresponding unaccented vowels by an acute accent. The dental affricates are denoted by the letters c (tense) and 3, (lax). The palatal consonants are symbolized by the letters _c, j[, s, z. Palatalization is indicated by a Polish hook (e.g. d represents a palatalized lax dental stop). The symbols fe and £ represent the tex bilabial and velar continuants respectively. Phonetic transcription is enclosed in square brackets, phonemic representations (satisfying Conditions;' 3 and 3a), in slants, and morphonemic representations (not satisfying Condition 3a), in braces. Since i t is convenient to distinguish between different stages of the sp edification of non-phonemic features., segment sequences to which the morpheme structure rules have not been applied are distin-guished from segment sequences containing only fully and in-completely specified morphonemes by underlining. Incom-pletely specified morphonemes are marked with a superscript ring to distinguish them from the corresponding fully specified morphonemes. The asterisk marks phonological entities (phonemes, morphonemes) with phonetic properties or distributional characteristics differing from those of similar phonological entities in CLU. CHAPTER II DESCRIPTIVE STATEMENTS A. SEGMENTS AND BOUNDARIES The Segments The feature composition of the f o r t y - f o u r morpho-nemes of CLU i s displayed i n the branching diagram on the following page (Figure I ) . The numbers at the d i f f e r e n t l e v e l s of the diagram r e f e r to the following eleven d i s -t i n c t i v e - features: 1 . Consonantal vs. non-consonantal. 2 . V o c a l i c vs. non-vocalic. 3. Compact vs. d i f f u s e . 4. F l a t vs. n a t u r a l . 5. Grave vs. acute. 6. Strident vs. mellow. 7. Nasal vs. non-nasal. 8. Continuous vs. interrupted. 9. Tense vs. l a x . 1 0 . Sharped vs. p l a i n . 1 1 . Accented vs. unaccented. The statements of the branching diagram are summarized i n the d i s t i n c t i v e feature matrix below (Figure I I ) . The segments w i l l be discussed i n d e t a i l i n Chapter II-A. 1 2 3 i i y f u e e a a 0 f 0 d t n 3 •5 c Consonantal + + + + + + + + + Vocalic — + + + + + + + + + + + + Compact o + + + + + + Plat 0 — — — — + + — — — — + + 0 0 0 0 0 0 0 0 0 Grave 0 — — + + 0 o — — + + 0 o Strident 0 0 0 o 0 0 0 0 0 0 0 0 0 - - - - - - + + + Nasal 0 0 o 0 0 0 0 0 0 0 0 0 0 - - - - + + 0 0 0 Continuous 0 o 0 O 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - -Tense 0 0 0 0 0 0 0 0 0 0 0 0 0 - - + + 0 0 - - + Sharped 0 0 0 0 0 0 0 0 0 0 0 0 0 - + - + - + - + -Accented 0 — + - + — + — + — + — + 0 0 0 0 0 0 0 0 c z z s s f c p v f m ^ S z s g k h x r r l l Consonantal + + + + + + + + + + + + + + + + + + + + + + Vocalic _ _ _ _ _ _ _ _ _ + + + + Compact _ _ _ _ _ _ _ _ _ _ + + + + + + + -».oooo Plat o o o o o o o o o o o o o o o o o o o o o o Grave - - - - - + + + + + - - - - + + + + 0 0 0 0 Strident + + + + + 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Nasal 0 0 0 0 0 - - - - + 0 0 0 0 0 0 0 0 0 0 0 0 Continuous - + + + + - - + + 0 - - + + - - + + - - + + Tense + _ _ + + _ + _ + o - + - + - + - + o o o o Sharped + - + - + 0 0 0 0 0 0 0 0 0 0 0 0 0 - + - + Accented o o o o o o o o o o o o o o o o o o o o o o Figure II. Matrix representing the morphonemes of CLU. The Boundaries The following five morpheme boundaries are postulated for CLU: 1 . A phrase boundary, designated by the symbol OlQ. 2 . A word boundary, designated by the symbol |2[j. 3. A prefix boundary, designated by the symbol l 3 l . 4. A suffix boundary, designated by the symbol U4-0 -13 5. A compound boundary, designated by the symbol |5l. The effects of these five boundaries are discussed in Chapter II-B. B. MORPHEME STRUCTURE RULES The morpheme structure rules are partially ordered. Each rule has been assigned a number and a letter, of which only the number refers to the order in which the rule must be applied. The section numbers correspond to the sections in Chapter IV in which the rules are discussed. 1.0. The following rules assign the proper values to a l l segments in which the features consonantal vs. non-consonantal and vocalic vs. non-vocalic are unspecified. 1.1. The rules in this section apply to segment-sequences immediately following the & marker. Rule MS l a . If the f i r s t segment is a glide, the second segment is a vowel. Rule MS lb. If the f i r s t segment is a liquid, a consonantal second segment is non-vocalic, and a non-con-sonantal second segment is vocalic. Rule MS l c . If the f i r s t segment is a liquid, and the second segment is consonantal, the third segment is a vowel. Rule MS 2. If the f i r s t three segments are 14 consonants, the fourth segment is a vowel. Rule MS 3. If an i n i t i a l cluster contains a non-i n i t i a l glide or liquid, the glide or liquid is followed by a vowel. Rule MS 4. If the morpheme contains no vowel, or i f the vowel which follows the i n i t i a l cluster is followed by a & marker, rules MS 5 to 11 do not apply. 1.2. The rules in this section apply to segment sequences immediately preceding the & marker. Rule MS 5a. If the last segment is a glide, a con-sonantal-penultimate segment is non-vocalic, and a non-con-sonantal penultimate segment is vocalic. Rule MS 5b. If the last segment is a glide, and the penultimate segment is consonantal, the antepenultimate segment is a vowel. Rule MS 5c. If the last segment is a liquid, and the penultimate segment is a consonant, a consonantal antepen-ultimate segment is non-vocalic, and a non-consonantal antepenultimate segment is vocalic. Rule MS 5d. If the last three segments are conson-ants, the preceding segment is vocalic. Rule MS 6. If the last segment is a liquid, and both the penultimate and the antepenultimate segments are conson-ants, the preceding segment is a vowel. Rule MS 7. If in a final cluster a segment other than the last is a glide or a liquid, i t is preceded by a vowel. Rule MS 8. If the morpheme contains only one vowelr rules MS 9 to 11 do not apply. 1.5. The rules in this section apply to segment sequences between the f i r s t and the last vowel of a mor-pheme. The rules are applied f i r s t to the segment sequence immediately following the f i r s t vowel of the morpheme. Rule MS 9a. If the f i r s t segment is a glide, a non-consonantal second segment is vocalic. Rule MS 9b. If the f i r s t segment is a glide, and the second segment is consonantal, the third segment is a vowel. Rule MS 9c. If the f i r s t segment is consonantal, and the second segment is a consonant, the third segment is vocalic. Rule MS 10. If in a medial cluster a segment other than the f i r s t is a glide or a liquid, i t is followed by a vowel• Rule MS 11. If the vowel following the medial cluster is not the last vowel of the morpheme, rules MS 9 to 11 are reapplied. 2.0. The following rules apply to segments in which the values of the features compact vs. diffuse, grave vs. 16 acute, and continuous vs. interrupted can "be inferred from other features in the same morpheme. Rule MS 12a. If a grave compact consonant is followed "by a consonant, the latter is diffuse. Rule MS 12b. If a glide is preceded by a consonant, the latter is grave and diffuse. Rule MS 12c. If a glide is followed by a non-flat diffuse vowel, the latter is acute. Rule MS 12d. In a sequence of liquids, the f i r s t is interrupted, and the second is continuous. The material on which the above rules are based is surveyed and discussed in Chapter IV. C. THE PHONOLOGICAL RULES The phonological rules are partially ordered. Each rule has been assigned a number and a letter, of which only the former is significant for the order in which a given rule is to be applied. The section numbers correspond to those of Chapter V, where the rules are discussed and exem-plified. 1.1. The following rules hold within a phonological phrase. Rule P I . If in a word i n i t i a l obstruent cluster the 17 last segment is tense, a l l obstruents other than {v} preced-ing the tense segment become tense; except that a lax ob-struent followed by a prefix boundary may remain lax i f i t is immediately preceded by a vowel and no pause intervenes. Rule P 2a. If an obstruent cluster contains a lax segment other than {v}, a l l obstruents in the cluster which precede the lax segment become lax unless a compound boundary intervenes. 1.2. Unless otherwise indicated, the following rules hold within a phonological word only. Rule P 2b. Non-nasal mellow acute diffuse consonants are interrupted. Rule P 2c. Before {j}, grave compact consonants become acute. Rule P 3a. Before acute compact consonants, non-nasal mellow and strident acute diffuse consonants become compact. Rule P 3b. Acute compact consonants are strident. Rule P 4a. Before strident acute diffuse consonants, acute compact consonants become diffuse. Rule P 4b. Before strident acute diffuse consonants, mellow non-nasal acute diffuse consonants become strident. 1.3. Rule P 5a. Grave diffuse consonants are sharped before {1} and, when preceded by a consonant with no 18 intervening boundary, before {j}; they are plain elsewhere before non-consonantal segments. 'Rule P 5b. Acute diffuse consonants and continuous liquids are sharped before {j} when no boundary intervenes. Rule P 5c. Acute compact consonants are sharped before {i} and {j} and plain elsewhere before non-consonant-al segments. Rule P 5d. {r} preceded by a consonant is sharped before {j}; otherwise i t is plain before {j}. Rule P 6. Unless a boundary intervenes, acute con-sonants and continuous liquids are geminated before a se-quence of glide followed by vowel. Rule P 7. After a sharp consonantal segment, or after any consonantal segment and before a word boundary, {q} is eliminated. Rule P 8a. Grave diffuse consonants are plain when not followed by a non-consonantal segment. Rule P 8b. Except for mellow and continuous strident acute diffuse consonants followed by a suffix boundary, a l l acute diffuse consonants are plain before {y} and {e}. Rule P 8c. Except across a prefix boundary, strident acute diffuse consonants are sharped before Rule P 8d. Grave compact consonants are sharped before {1} and plain elsewhere. Rule P 8e. Acute compact consonants are plain when 19 not followed by a non-consonantal segment. Rule P 8f. {r} is sharped before {i}, but plain before {y} and {e} and when not followed by a non-consonant-al segment. Rule P 8g. Except when part of a gemination immmedi-ately preceded by a word or prefix boundary, or when a suf-fix boundary intervenes, {1} is plain before {y} and {e}. Rule P 9a. In clusters of acute diffuse consonants, a l l segments are sharped i f the last segment is sharped, and a l l segments are plain i f the last segment is plain, unless a boundary intervenes. Rule P 9b. Acute diffuse consonants are plain before {1} and {r}. Rule P 9c. When no prefix boundary intervenes, s t r i -dent acute diffuse consonants are sharped before sharped grave diffuse consonants. Rule P 9d. Acute diffuse consonants are sharped before {1} when no boundary intervenes. Rule P 9e. {1} is sharped before sharped acute diffuse consonants. 1.4. Rule P 9f• When no prefix boundary inter-venes, a continuous strident acute consonant becomes inter-rupted in position after an interrupted strident acute consonant. 20 Rule P 9g. {j} is acute and sharped. Rule P 10a. In position after a strident acute dif-fuse consonant and before {n}, an interrupted acute conson-ant is eliminated. Rule P 10b. Geminates are simplified when contiguous to a consonantal segment. Rule P 10c. When not followed by a vowel, postvoe-alic {v} becomes non-consonantal. Rule P 11a. A sequence of identical consonantal seg-ments is articulated with a single onset or closure, and a single terminal transition or release, separated by a pro-longed tenure. 1.5. Rule P l i b . Flat vowels are grave. Rule P 12. In position after a sharped segment, grave vowels tend to become acute. Rule P 13. In position before a sharped segment, grave vowels have [i]-like off-glides. Rule P 14. Unaccented grave flat and acute natural compact vowels tend to become diffuse. 2. Rule P 15. In a phonological phrase containing two accented vowels, the f i r s t accented vowel is less pro-minent than the second. Rule P 16. Unaccented vowels are less prominent than accented vowels. CHAPTER I I I COMMENTS ON THE SEGMENTS AND BOUNDARIES A. THE SEGMENTS The fo l l o w i n g pages o f f e r some comments on the mor-phonemes of CLU under the following headings: 1. the g l i d e ; 2. the vowels; 3*1. the l a b i a l , 3.2. the dental, 3.3. the p a l a t a l , and 3.4. the v e l a r consonants; 4. the l i q u i d s . Each of these sections discusses the feature composition of the segments as proposed i n Chapter II-A as well as the problems of analysis connected with the i n d i v i d u a l morpho-nemes. Where the solutions adopted i n the present d e s c r i p -t i o n d i f f e r from those of e a r l i e r d e s c r i p t i o n s , the l a t t e r are reviewed. The concluding Section 5 contains a few remarks on the degree of economy achieved i n the branching diagram i n Chapter II ( P i g . I ) . The Glide The d i s t i n c t i v e features consonantal vs. non-conson-antal and v o c a l i c vs. non-vocalic define four classes of segments i n CLU, g l i d e s , vowels, consonants, and l i q u i d s . This i s i n conformity with the d e f i n i t i o n s i n Fundamentals  of Language and with Morris Halle's treatment of Russian. 1 1 Roman Jakobson and Morris H a l l e , Fundamentals of Lan-guage (Vol. I of Janua Linguarum, ed. C. H. van Schooneveld), 22 Apparently the definitions of vowels, consonants, and liquids 2 are well founded. But glides seem problematic. In his review of the definitions of the distinctive features which are proposed in Fundamentals, Fant discusses the difficulties raised by glides in general, and by sounds like those of Russian {j} in particular. Specifically, he suggests the possibility of defining the latter as non-voc-alic and consonantal ( i f its friction be considered distinc-tive), or as vocalic and consonantal ( i f the friction be considered unimportant). On the whole, i t would seem, sounds like those of Russian {j} are not optimally described as non-vocalic and non-consonantal, a description that best f i t s pure glides of the [h]-type. Fant does concede, how-ever, that "a weak non-fricative [j] connected smoothly to a following vowel may be labeled a glide.^ It i s on the basis of this statement that the decision to define CLU {j} as non-consonantal and non-vocalic was made. For where {j} is represented by [j] , i t 's-Gravenhage, Mouton & Co., 1956, p. 29; and Morris Halle, The Sound Pattern of Russian; A Linguistic and Acoustical  Investigation, 's-Gravenhage, Mouton & Co., 1959, p. 52 f. 2 Gunnar Fant, Acoustic Theory of Speech Production, •s-Gravenhage, Mouton & Co., I960, p. 215 f. ~ 3 Pant, p. 216. 23 answers Pant's description of a [j]-glide very well. Ziiyriski describes i t in part as follows: . . . Tylko w emfazie w pozycji akcentowanej, osobliwie przed i , otrzymuje i w jgzyku ukrainskim charakterystyczny dia spdlgloltek frykatywnych szmer, zblizajacy je do dzwigcznego spirantu j , jednakowoz odeieri ten i rdzni sie, bardzo znacznie od wlas"ciwego j[ niemieckitgo slabszym szmerem. . . .5 1.2. With one exception, a l l the descriptions of CLU known to this writer recognize [j] as constituting a phoneme, / j / of the language. The one dissenter is Karol Dejna, who in his "Fonologiczny system jgzyka ukrainskiego" states that f. "wariantem fonemu i jest 3.." Dejna does not elaborate this statement, but presumably he means that [i] and [j] do not contrast and, consequently, can be considered representa-7 tions of one and the same phoneme. This is not true, however, [i] and [j] do contrast, and, where they do, their order cannot be inferred from the phonetic context. This is clearly shown by pairs of words 4 It may be noted here that {j} is not uniquely realized as [ j ] . Cf. Sec. 1.3. 5 Jan Ziiyriski, Opis fonetyczny jgzyka ukrainskiego (Polska Akademja Umie je, tnosci, Prace Komisji Jfzykowej, Ir 19), Krakow, Gebethner i Wolff, 1932. 6 Biuletyn Polskiego Towarzystwa Jgzykoznawstwa, X (1950), p. 151. Dejna's phoneme */±/ will be discussed again below, See. 2.32. 7 Cf. IST. S. Troubetzkoy, Principes de phonologie, trans. J. Cantineau, Paris, C. Klincksieck, 1957, 52 f. 24 like [obijdiisa] 'I shall do without' and [objid^juga] 'I am gorging myself, or [bezzubij] 'toothless' (dat.sg.fem.) and [bezzubji] 'toothlessness' (loc.sg.). Dejna's phonemic solution would render the two last mentioned words */bezzubii/ and */bezzubii/. Since i t f a i l s to account for the phonetic facts, i t must be rejected. 1.3* It was mentioned above (Sec. 1.1, n. 5) that {j} is not uniquely realized as [ j ] . Two distinct problems are involved, the matter of long, or geminate, consonants, and the question of distinctive sharping of generally only non-distinctively sharped consonants. Although the former problem involves more than the multiple realizations of {j}, this seems the most convenient place to survey both questions in f u l l . q 1.31. The long, or geminate, consonants^ have been treated in widely divergent ways by different writers. It may be of some interest to compare the different treatments by Zilynski, Dejna, and George Y. Shevelov. 8 The absence of accentuation in these examples is in accordance with Dejna's failure to confer phonemic status on the CLU stress. Cf. his remarks on p. 151. In the article, stress is marked only in the phonetic transcriptions. 9 The CLU geminates are described variously as long, lengthened, and geminate (both phonetically and phonemically). The terminological confusion has not been relieved by Trou-betzkoy's term "correlation de gemination" for the feature of consonant length. Cf. Principes, p. 184 f. 25 1.5111* Zilynski, who treats geminates in his chap-ter on quantity, calls them "dlugie (wzdiuzone), but points out that in intervocalic position geminates are articulated with diminuendo-crescendo intensity and, consequently, give the impression of identical consonants separated by a syllable boundary.1^ He does not give an exhaustive account of gemination in the chapter on quantity, but geminates are commented on passim a l l through the book. Prom the table of consonants, where "dzwi§ki niesamoistne" (i.e. positional variants) are included in brackets, i t is apparent that Ziiyiiski does not consider geminates separate phonemes.11 1.3J12. Dejna's very tersely formulated statement on geminates can be quoted in f u l l . 5. stopien przezwyoiezenia; Korelacja podwojeniat Na skutek tego, Se w dawnyeh poiaczeniach 17, n^, t*, d*, s*. z l , £l» J l , z* + 1 j po zaniku ^  zidentyfiko-walo sie, j| z poprzednia zmiekczona spdigloska, a po-laczenia l ' l * , t*t* . . . znalazly sie, w obr^bie jed-nego morfemu, mamy w j § zyku ukr. korelac j § podwojenia. Ponemy ^ IL, I^L, l H , J^sl, JV^, 21* 21* o wyraznym poczatku i koiicu pozostaja do niepodwojonych 1*, nj_, tj_, d7, &2_r z^ ., c, _s, z, gdzie nastapilo fo-nologiczne zlanie sie, poczatku i konca, w opozycji 10 Thus these consonants are properly speaking geminates from a phonetic point of view. Of* J» Marouzeau, Lexique de l a terminologie linguistique. 3-e Id., Paris, Geuthner, 19bl. 11 Zily&ski, p. 158 f., p. 6, and passim. 26 12 jednowymiarowej proporcjonalnej prywatywnej. Dejna adds that in the Volhynian dialects, where the third person marker {$} has fused with the i n i t i a l of the reflex-ive enclitic {§a}, a geminate "'c'" M. . . z mozliwego 13 staje sie, fonemem realnym . . . 1.3113* Shevelov, who mentions Dejna1s article in his bibliography, 1^ does not share Dejna1s views. He rejects the term "podvojennja" (gemination) as referring properly only to the orthographic rendering of these consonants and holds that articulatorily there is only "podovzennja" (lengthening). After stating that tj_, cP, s_^ , z7, cj_, 1*, n', b, v, m, d, jz, _z, n, ss, j_ can occur, lengthened, and that the normally plain z, a, 6 can be sharped and length-ened, Shevelov surveys the contexts in which geminates are found. He defines five distinct contexts: A. At the juncture between prefix and radical, between radical (or base) and suffix or desinence, between word and enclitic, or between words. B. In stem final position in neuter nouns 12 Op;, cit.., p. 151. 13 0p_. c i t . , p. 152. It is unclear why such a phoneme should"be limited to the Volhynian dialects. The phenomenon is wide-spread in the eastern part of the Ukraine. Cf. the texts numbered 1, 2, 3, 4, 6, 7 (East Polesian); 8 (Central Polesian); 3 7, 38, 39 (Middle Dnieper); 41, 43, 44 (Sloboda); 45, 47, 48 (Steppe); in P. T. Zylko, Narysy z  dialektolohiftukrajins'koji movy. Kyjiv, Radjans'ka Skola, 1955. Cf. also Zilyhski, p. 89. 14 George Y. Shevelov (pseud. Jurij Slerex), Harys sucas-27 in -ja, and in some masculine and feminine nouns in -ja. C. Before the in.sg. desinence -ju of feminine nouns of the third declension. D. Before the reflexive enclitic {§a}, where the person markers {s} '2nd pers. sg.' and {$} '3rd pers.' have been assimilated. E. In the adjectival suffixes -enn(yj) and - a n n ( y j ) S h e v e l o v considers the geminates separate phonemes, but he is quick to point out that . . . vaha podovzennja pryholosnyx u sueasnij literaturnij movi ne velyka, bo vono maje zvycajno til''ky morfolohicnu rolju, buvsy zv'jazane z rozmezuvannjam morfolohicnyx castyn slova abo z pevnymy morfolohicnymy katehorijamy. Only in a few unclassifiable instances does gemination "characterize the radical of a word," as seen in examples like panna, ssaty, 11ju — l l j e s , where " . . . podovzennja . . . ne tvoryt' jakojis' systemy, a zberihajet'sja cysto 16 tradycijno . . . ." j.312. In the following pages, Dejnate and Shevelov's phonemicizations will be confronted with the material they describe, and at the same time i t will be shown how the noji ukrajins'koji literaturnoji movy, Munich, Molode Zyttja, 1951, p. 396. 15 These categories wi l l be referred to repeatedly in the discussion below, Sec. 1.313. 16 Shevelov, Narys, pp. 379-81. 28 morphonemes of the present description, which includes no distinctive feature of "consonant length" or "gemination", are related to the geminate consonants of CLU. First, how-ever, a few remarks of a general nature seem to he in order. In his chapter on gemination, Troubetzkoy makes the following introductory remarks, which are quite pertinent to the problem in CLU . . . Or, i l y a dans beaucoup de langues des con-sonnes dites ge'mine'es. Elles se distinguent des consonnes non ge'mine'es par leur duree plus longue et la plupart du temps aussi par une articulation plus Inergique . . . . Mais en position intervocalique les consonnes ge'mine'es sont reparties entre deux syllabes, leur implosion appartenant a l a syllabe prece'dente et leur explosion a l a syllabe suivante. En outre ces consonnes geminees nfapparaissent que dans les positions ou l a langue en question admet des groupes de consonnes; elles . . . sont traitles en general exactement comme les groupes de consonnes. Toutes ces marques indiquent une valeur polyphone-matique, c'est-a-dire invitent a interpreter les consonnes ge'mine'es comme des groupes formes de deux consonnes identiques.17 The following features are characteristic of the CLU geminates. In intervocalic position, implosion and explosion are separated by a syllable boundary (cf. Ziiynski's descrip-tion, summarized above, Sec. 1.3111). Geminates do not occur in positions where consonant clusters are not admitted. Geminates are subject to certain constraints that seem to be 1 8 general for a l l sequences of non-vowels. It is thus 17 Principes, p. 184 f. 18 This statement is elaborated below, Sec. 1.3134. perfectly possible, from a theoretical point of view, to interpret the CLU geminates as clusters. Since their inter-pretation as functional units requires the introduction of several additional segments, and of one extra distinctive feature, in the description, this interpretation should not be accepted unless strong arguments in its favour are found. Shevelov remarks that ". . . movna svidomist' spryj-maje joho [podovzenyj pryholosnyj] jak podvojenyj i nezelez-19 no vid pys'ma . . . ." When he nevertheless chooses to interpret geminates as lengthened consonants, he appears to prefer a counter-intuitive solution to one that seems more natural. He adduces only one reason for this step, viz that articulatorily there is only lengthening: ". . . u slovi 'cyslennyj' ne vymovljajemo dvox *n' . . . a til'ky majemo podovzenu vymovu . . . . Since Shevelov agrees with Ziiyriski (summarized above, Sec. 1.3111) that " . . . meza V V 21 skladu proxodyt1 cerez podovzenyj pryholosnyj . . . ." his only argument in favour of consonant length is clearly based on a failure to trace the audible intensity "valley" 22 that separates geminates to its articulatory source. 19 Shevelov, Narys, p. 379. 20 Loc. c i t . 21 Loc. c i t . 22 In his chapter on the syllable, Shevelov does do this: '*. . . clenuvannja na sklady maje xarakter cysto fonetycnyj Dejna advances only one argument for his "korelacja podwojenia," the fact that geminates are found within single 23 morphemes. This argument is far from convincing. That geminates occur within single morphemes can be considered the minimal prerequisite for their interpretation as unit phonemes. But i t can by no means be considered sufficient grounds for such an interpretation.^ Within the framework of Morris Halle's theory of phonology no a priori considerations of the relative merits of alternative solutions are relevant. It was stated above that, from the point of view of economy of description, geminates should be treated as clusters unless strong argu-ments against such a solution could be found. The only argument strong enough to force the rejection of this solu-tion is that i t f a i l s to meet Condition 3 and/or Condition 4. It will be evident from the following paragraphs that not only does the solution adopted here not violate Conditions 3 and 4, but i t alone makes i t possible to state certain simple operations in C1U inflection and derivation in a simple fashion. 23 Cf. the quotation in Sec. 1.3112. And yet, he does not set up any *n, _js, or _*v, a l l of which occur within morphemes, cf. Hec. 1.3131. 24 Apparently Dejna has misunderstood Troubetzkoy's statement to this effect, Principes, p. 185. 31 1.3131. Shevelov sets up the following phonemes not recognized "by Dejna: */bs v: m: d: z: z: n: s: j : / . They are a l l exemplified in Shevelov*s Category A., where any attempt at morphological analysis will require their inter-pretation as clusters, and not as unit phonemes. Examples of these geminates are given in Chapter V as illustrations of Rule P 11a. Only */v: n: s:/ are attested outside Category A. */n:/ is attested both in Category E. and among the unclassifiable items, i.e. in single morphemes; whereas */v:/ and */s:/ are attested only in the latter group. There are no cogent reasons why these cannot be con-sidered clusters. As for */v:/, which outside Category A, is attested only in bowan 'image' (and derivatives), i t can be added that the maximally distinct allophones of {v} here, [bowv^n], make their interpretation as a functional unit rather unsatisfactory. */s:/ is attested outside Category A. only in ssaty 'suck' (and derivatives). That the [s:] here represents a cluster {ss} is evident from the equation {rv-u}: {vy-ryv-£j-u} = {zv-ii}: {vy-zyv-^j-u} = {Is-ii}: {vy-sys-£j-u} 'tear; tear out; ca l l ; call out (trans.); suck; suck out' (1.sg.pres.). 1.3132. Shevelov's Category D. exemplifies Dejna's Volhynian "*c*" (i.e. % : / ) , as well as y§:/. Here, as in Category A., an attempt at morphological analysis would require that §:/ be reinterpreted as clusters, 32 respectively /eg/ and/ss/. T§te distribution of the "allo-morphs" of the 2.sg. person marker (/myj-e-s/, /myj-e-s-ga/ 'you (sg.) wash' (trans, and intr.) ), the third person mark-er (/myj-u-£/, /myj-u-c-ca/ 'they wash' (trans, and intr.)), and the reflexive enclitic (cf. /myj-u/, /myj-u-sa/ 'I wash' (trans, and intr.)), is accounted for by the dis-tributional constraints of the language (the sequences [ss] and [$s] are not admitted within word boundaries). This means that, given the rules P 4a, 4b, 9a, 9f, and 11a, the words just mentioned can be adequately represented as follows: {myjes}, {myjessa}, {myju$}, {myju^sa}, {myju}, {myjusa}. 1.3133. Dejna and Shevelov agree on setting up */d: $: n: z: s: c: ?: |: |: %'/• These are a l l represented in Shevelov's Category C. It is impossible to establish a single stem for nom.sg.. and in.sg. in words like the following unless these geminates are considered clusters: /mid/, /mid-du/ 'copper'; /$in/, /$i*n-nu/ 'shadow'; /maz/, /maz-zu/ 'ointment'; / s i l / , / s i l - ^ u / 'salt' (cf. /krov/, /krov-ju/ 'blood' and /dvir/, /vir-ju/ 'door'). A similar situation is found in Category B., where i t is not possible to isolate from the underlying bases the derivative suffix that forms the numerous collective and abstract neuter nouns in this group unless the geminates are analysed as clusters. Compare /liud-y/ 'people', /bez-^ud-33 d-a/ 'underpopulation'; /son/ 'sleep', /bez-so^-^i-a/ 'insomnia*; / l i s / 'forest', /bez-l£§-s-a/ 'un-forested area'; /(Jil-o/ 'act', /bez-dil-1-a/ 'inaction'; and /slub/ 'wedding', /bez-slub-j-a/ 'common-law marriage'; /vfr-a/ 'faith', /bez-vir-j-a/ 'atheism'. In both categories an automatic alternation is involved. The only clusters of a consonantal segment followed by / j / that are admitted in the native vocabulary of CLU are those beginning with a labial consonant or / r / . The examples just listed partly illustrate how / j / (both the i n i t i a l of the desinence /ju/ and the derivative suffix) alternates with the duplicate of whatever single acute con-sonant or continuous liquid precedes i t . An accurate formu-lation of the distributional constraints involved can be found in rules P 2e, 5b, 5c, 6, 7, 1Gb, 11a, and i l l u s t r a -tions of these rules, in the corresponding sections of Chapter V. This alternation, incidentally, also accounts for the unclassifiable 11 ju — 11jes, as might be seen from the equation /byj/:/bju/:/bjes/ 'beat' (2.sg.impr.; l.sg. pres.; 2.sg.pres.) = /pyj/:/pju/:/pjes/ 'drink' = / l y j / : /lJ.u/:/^les/ 'pour' (cf. the examples to rule P 8g). 1.3.34. One additional point should be mentioned in this connection. It was mentioned above (Sec. 1.312, n.18) that geminates are subject to certain constraints that can be considered general for a l l clusters. In word final 34 position and in the context C...V no sequences of consonan-tal segment followed by / j / are attested. In these con-25 texts, geminates are not admitted. Thus, for instance, to /pid-borid-$-a/' •chin* (nom.sg.) corresponds the gen.pl. /pid-borf^/ with a zero desinence; and, while the in.sg. of /su$/ 'gist 1 is /sd|-|u/, the corresponding form of /cas$/ 'part' is /ca's-Jj-u/ (cf. rules P 6 and 10b). In the con-text R...Y, sequences of labial consonant plus / j / are - admitted, but sequences of / r / plus / j / are not attested. Neither are geminates; the in.s. of /smer$/ 'death' is /smer$-u/ (cf. rule P 10b). Thus, not surprisingly, there is a fair degree of similarity between the constraints governing geminates and those governing clusters of labial consonant or / r / followed by / j / . To sum up the preceding few pages, there is no need to set up a series of geminate morphonemes and a distinctive feature of consonant length in CLU. The vast majority of geminates straddle morpheme boundaries, where they arise as the result either of the chance meetings of identical mor-pheme final and morpheme i n i t i a l segments, or of the dis-tributional constraints of the language. Only in a handful 25 Dejna's assertion that geminates do not occur in word i n i t i a l position (cf. the quotation in Sec. 1.3112) does not appear to be factual. Cf. the forms of ssaty;-' suck' and lyty 'pour', discussed in Sec. 1.3131, 1.3133. 35 of instances are geminates found within morphemes. Some of these have been mentioned already (Sec. 1.3131)• A few more will be mentioned in the discussion of rule MS 12b in Chapter IT, Sec. 2.2. 1.32. The question of distinctive sharping of gener-ally only non-distinctively sharped consonants involves two groups of consonants, the palatals and the labials. They are most conveniently discussed separately. 1.321. In position before a vowel, palatal con-sonants are, generally speaking, sharped before / i / and plain elsewhere. An exception are the soft geminate pala-tals which may precede not only / i / , but also /a/ and /u/. Thus, for instance, parallel to /dvir/, /dvir-ju/ 'door' (nom.sg. and in.sg.), and /su$/, /SVL^-^VL/ 'gist 1, there is [nic], [nfgru] 'night*, which could be adequately phonemi-cized as /nic/, /nfc-ju/ since a l l [g:] represent / e j / (the three in.sg. forms are morphonemically {dvirju}, {su$ju}, {nicju}). In addition, however, single sharped palatals occur in a small number of words before /a/ and/u/. No descrip-tion known to this writer sets up special phonemes to take care of these palatals, and indeed there is no need to. Like the corresponding geminates, they are found only at morpheme boundaries. Unlike the geminates, however, they 36 are always preceded by a consonantal segment. Prom a com-parison of /ces$/, /ces-fc-u/ 'honour* (nom.sg. and in.sg.), and /bez-ce's-fc-^-a/ 'ignominy' on the one hand, and [zowc], [z6*w§u] 'gall' and [bez:6wga] 'acholia' on the other, i t is evident that the last two items result from the constraints that were discussed in Sec. 1.3134. Morphonemically they can be written {zovcjuj and {bez|3!z6"vcja} • 1.322. The distinctively sharped labial consonants, like the palatals discussed in the preceding section, are limited to a small number of words. Shevelov l i s t s the most important of these "nexarakterystycni restky." He recog-nizes the distinctive function of the sharped labials, but does not set up any sharped labial phonemes. Dejna does not mention distinctively sharped labials. It was mentioned above (Sec. 1.3134) that no sequences of consonantal segment followed by / j / i s attested in the context C...V. It is in this context that the distinc-tively sharped labials occur. It is consequently possible to represent every distinctively sharped labial as a morpho-nemic sequence of labial plus {j} and formulate the phono-logical rules accordingly (cf. rules P 5a and 7). This solution makes i t possible to formulate general morphological rules describing the {j} "epentheticum" added 26 Shevelov, Narys, p. 377. 37 to morphemes ending in a labial consonant before suffixes like {an} and {ac}. Cf. {cierev-o} •wood', {cterevj-^n-y-j} 'wooden* aiid {$m-£} 'darkness*, {^mj-an-yj} 'darkish' (the adjectives ultimately become [derevj^nyj] and [$many;j]); {kordv-a} 'cow', {kordvj-ac-yj} 'pertaining to a cow' and {mavp-a} 'monkey', {mavpj-ac-yj} 'pertaining to a monkey' (ultimately [kordvjacyj] and [mdwnacyj]). The distinctively sharped labials occur not only at morpheme boundaries, but also within morphemes. There seems to be no better way of representing such words as [sya^to] 'holiday', [cyax] 'nail' than {svj^t-o}, {cvja^x}. This solution, however, has certain consequences for the morpheme structure rules. These will be discussed in Chapter IV, Sees. 1.1 and 2.2. 1.323. It was mentioned above (Sec. 1.3133) that the only sequences of consonantal segment followed by / j / which are admitted in the native vocabulary of CLU are those beginning with a labial consonant or non-continuous liquid. In the preceding section i t was shown that i t is possible to analyse sharped labial consonants in the context C...V as morphonemic sequences of labial followed by {j}. In the context C....V also the sharped non-continuous liquid occurs. Since / r / is a phoneme of CLU, there would be no motivation in a traditional phonemic analysis (leading to the establishment of a phonological representation satisfying 38 both of Conditions 3 and 3a) to push the analysis of, for instance, [bezyitru] 'with no wind' and [bezyitru] 'dead calm' (dat.sg.) beyond /bezvitru/ and /bezvitru/. The parallelism of /vira/ 'faith', /bezvirja/ 'atheism' and /viter/ 'wind', /bezvitra/ 'dead calm', however, suggests that the latter contains the morpheme {j}. In a morpho-nemic representation, i t can be adequately written as {bezfl3flvf|r-j-a} provided a rule is formulated to the effect that in the context C...V {rj} is realized as [r]. Cf. rules P 5d and 7. The Vowels 2.11. The distribution of the principal allophones of the CLU vowels is described in rules P 11c to 14 and exemplified in the corresponding section of Chapter V. In the present section the vowels will be briefly characterized in articulatory-auditory terms and related to the distinctive feature definitions decided on for this description. The following section will add a few remarks on some possible problems connected with these definitions. On the quadrilateral in Figure III the circles indi-cate the approximate values of the six accented vowels as they are realized in unchecked position, e.g. in isolation (as names of the letters _ i , u, £, a, o). 27 The quadrilateral represents the eight primary cardinal Figure III. The six accented vowels of CLU in unchecked position. Cf. p. 38 f f . 40 From this figure i t ean he seen that the CLU vowels constitute two series opposed articulatorily as wide vs. narrow, i.e., acoustically, as compact vs. diffuse. {u}, {u}, {6}, {o}, while generally realized as back vowels, undergo varying degrees of fronting when adjacent to sharped non-vowels (stressed and unstressed [ii], [6]), par-ticularly when surrounded by sharped non-vowels (where {u}, {o} may be realized as [ii], [6]). The feature that consis-tently distinguishes them from the remaining vowels is labialization. In the present description, therefore, they are defined as fl a t , as opposed to the rest of the vowelst which are natural. {u} and{{o} are generally more lax than the corre-sponding accented vowels, and, at the same time, less extreme: {u} is less diffuse, {o} , less compact (e.g. when not contiguous to a sharped non-vowel, respectively [u v] and [o*]). The raising of {o} is particularly marked in pre-tonic position when the following syllabic of the same word is [u] or [ i ] . In this position {o} may be realized as [u v] in allegro speech. Though {y} and {e} vary considerably in terms of" cardinal vowels of the IPA. Cf. The Principles of the  International Phonetic Association, London, Dept. of Phonetics, University College, 1949 (Reprinted 1958). Note that the symbols used here are not those of the IPA. 41 tongue height, and {£.} , {a} vary along the front-hack axis, ii)> {y} a n d {£}, {a} are consistently opposed to respec-tively {£}, {i} and {£}, {e} as retracted to advanced. In the present description, therefore, {f}, {j} and {&}, {a} have been defined as grave, and. {f}, {i} and {e}, {e}, as acute. {y} and {e} are realized as more lax, and less extreme, than the corresponding accented vowels. {y} is generally lowered to a [y r ] . {e} is usually realized as [y v] when the following syllable of the same word is a diffuse vowel, otherwise, as [e A], A careful style, however, may distinguish {y} and {e} in a l l positions. {£.} and {a} are somewhat advanced (i.e. become less grave) when adjacent to a sharp non-vowel (stressed and unstressed [a]). Their fronting is especially marked when they are surrounded by sharp non-vowels (unstressed, and occasionally also stressed, [a]). 2.12. In view of the complete absence of any acoustic data on the CLU vowels, i t is obviously impossible to discuss their feature composition in any detail. Even the following very general remarks are fraught with a l l the weaknesses of pure speculation, but some comments on the vowel definitions decided on are clearly required. The sonority feature compact vs. diffuse seems the 42 28 least problematic. As defined in Fundamentals, this feature is related to the ratio of the volume of the front resonance chamber of the vocal tract to that of the back resonance chamber. This ratio is higher for wide vowels than for narrow vowels. In terms of the positions of the fi r s t two formants (Fl and F2), a high ratio between the two resonance chambers means centrally located F l and F2, a low ratio, widely separated Fl and F2. In practice, the posi-tion of F l may in some languages be sufficient for the OQ identification of the feature. It is likely that this is the case in CLU, though possibly a more complex identifi-30 cation procedure is required. In any case, the difficulty of defining a threshold between compact and diffuse vowels will be the same. Such a threshold will have to 28 P. 29 f f ; also the discussion in Fant, o_p. c i t . , p. 217. 29 This is the situation in Russian (where the feature is split into two, diffuse vs. non-diffuse- and compact vs. non-compact). Cf. Sound Pattern, p. 126: "[Diffuseness] . . . is acoustically signalled by a maximally low f i r s t formant.?; and p. 127: "Compactness is signalled by a maximally high f i r s t formant." The same view is held by Y.I. Grigor'ev ("differencial'nye priznaki russkix glashyx /u, y, i / " , Voprosy Jazykoznanija, XI:1, 1962, pp. 10-30), who, on the basis of the dependency of Fl on the fundamental, argues against defining absolute values of diffuseness and terms the diffuse vowels of Russian simply "krajnie", with refer-ence to their extreme, i.e. maximally low, F l . 30 Cf. Roman Jakobson, C. Gunnar M. Fant and Morris Halle, Preliminaries to Speech Analysis: M.I.T. Acoustics Labora-tory Technical Report Ho. 13, 2nd printing, Cambridge (Mass.), Massachusetts Institute of Technology, 1952, p. 27. 43 discriminate properly between the high Fl positions of [y v] and [u v] and the low Fl positions of [e A] and [o A]. The tonality feature which is considered primary for the CLU vowels, the feature flat vs. natural, is related to the shape of the front orifice of the mouth resonator: Labialization causes a downward shift of some of the corn-el ponents of the spectrum. In Russian, the feature flat vs. natural " . . . poses no problem. Its acoustical correlate •50 is the position of the second formant."^ It seems reason-able to assume that a similar situation exists in CLU. The most advanced articulations of [u] are possibly more advanced than [y], but can, even then, be expected to have a lower F2, due to the l i p rounding. The same relation probably holds between [6] and [a]. It is suggested, then, that flat vowels can be defined as having an F2 at, or below, a yet to be defined c r i t i c a l frequency. The second tonality feature relevant to the specifi-cation of the CLU vowels, the feature grave vs. acute, presents a less clear-cut picture than the two features dis-cussed above. In Fundamentals, gravity is defined as a con-centration of energy in the lower frequencies of the spec-k's trum. ^ The feature is thus the acoustic correlate of the back-front opposition in the vowels. The retraction of the 31 Cf. Fundamentals, p. 31; and Fant, p. 219. 32 Sound Pattern, p. 130. 33 Fundamentals, p. 31. tongue from the [i] position to that of [y], or from that of [e] to that of [a], is accompanied "by a downward shift of F2 and thus lowers the energy concentration of the spectrum. It should he noted, however, that this lowering of the energy concentration is somewhat counteracted by a concomi-tant upward shift in Fl ([y] is less diffuse than [ i ] , [a] more compact than [e]). In view of the lack of acoustic data no specific identification procedure can be 34 suggested. 2.13. The present description of the CLU vowels conflicts with the statement in Preliminaries that Ukrai-nian has " . . . two classes of grave vowels — flat as /u/,, plain as / i / — and only a single, optimal class of acute vowels — plain as / i / . " ^ This statement agrees well with the view, expressed in Preliminaries and elsewhere, that the feature grave vs. acute is more fundamentally important in languages in general than the other tonality features. It does not, however, agree with the phonetic facts of CLU. 34 Cf. Preliminaries, p. 29 f f . 35 P. 35. Roman Jakobson presents the same view in "Remarques sur l 1Evolution phonologique du russe, compared a celle des autre langues slaves," TCLP, II (1929), p. 71. 36 Preliminaries, p. 32 f f . ; also Fundamentals, p. 40 f f . 45 As mentioned above, the labialized vowels have both back t and front allophones. Gravity, then, cannot properly be considered an invariant attribute of the labialized vowels, but must be treated as a redundant feature. 2.2. Most treatments of CLU phonology describe the vowel system in terms of six vowel phonemes and a phoneme •37 of stress. In addition, i t is usually recognized that the distribution of the six vowel phonemes, independently of the stress, is somewhat uneven. The present section will out-line the distributional characteristics of the CLU vowels as they present themselves to the analyst. The discussion will in particular concern the phonemes / i / and /y/, which have been the object of special attention in the literature on CLU. In this connection, some attempts to reduce the vowel inventory to five phonemes will be reviewed. But f i r s t a statement of the distributional facts is in order. 2.21. Since vowel chains do not occur in the native vocabulary of CLU, i t is necessary to distinguish only two typical positions for vowels in the CLU word. In the 37 Or, as A. V. Isacenko, "Versuch einer Typlogie der slavischen Sprachen", Linguistica Slovaca, I (1939), p. 71, in terms of six stressed and six unstressed vowels. In the following, the feature accented vs. unaccented wi l l , for simplicity's sake, be left out of consideration. 46 following they will he termed i n i t i a l (following a pause) and medial (following a non-vowel) respectively. The latter position is numerically the more important^ and will he dealt with f i r s t . 2.211. Figure IT presents, in a rough phonetic transcription, a l l possible sequences of non-vowel followed by vowel, with the exception of those described in notes (b) and (c). The latter are best left aside for a moment, but will be taken up below. From this table several facts are evident. Firstly, CLU has six vowel phonemes, / i e y a u o/. Secondly, the consonantal phonemes f a l l into two categories, that of the "paired" and that of the "unpaired" consonantal phonemes. The unpaired consonantal phonemes — the labial, the palatal, and the velar consonants— are sharped before / i / and plain before the rest of the vowels. In the paired consonantal phonemes, on the other hand — the dental consonants and the liquids — sharping plays a distinctive role. Thirdly, the paired consonantal phonemes form two groups, which may be called "the strong" and "the weak". In the strong group, 38 I. Z. Petlycnyj ("Do pytannja pro systemu holosnyx fonem v sucasnij ukrajins'kij literaturnij movi", Pytannja  ukrajins1koho movoznavstva, I (1956), p. 16) counted 13000 vowels and found that 94 per cent were in medial position, 5.6 per cent, in i n i t i a l position; .4 per cent were the second segment in vowel chains ( a l l in foreign words). 47 [i] [e] [y] [a] [u] [o] + + + + + + + [j] (a) [b p v f m] _ + + + + + + [p p y f m] (b) + - - - - • - -[d t n] + + + + + + + [4 % (c) + - - + + + + [z s 3 c] + + + + + + [ ? ? J « ] ( ° ) + - - + + + + r v v v v "i L 3 C Z S J _ + + + + + + [3 § i §] (b) + - - - - - -[g k h x] [1] - + tl] (c) [r] + + ' + + + + + + - - + + + + - + + + + + + + - - + + + -WORD INITIAL + + + + + + + Figure IV. The occurrence (+) and non-occurrence (-) of vowels following non-vowels and pause, and of non-vowels preceding vowels and pause. NOTE: To the generalizations of the table the following details should be added: (a) The sequence [jy] occurs only across morpheme boundaries; cf. Sec. 2.2213. (b) Sharped labials and palatals occur before [a u] under the circumstances discussed in Sec. 1.32. (c) Sharped dentals and [1] occur before [y e] across the suffix boundary ( l 4 l 5 ; cf. Chapter III-B. which comprises the non-strident dentals and the continuous liquids, sharping is distinctive before / i a u o/, but not before /e y/. In the weak group, which comprises the s t r i -dent dentals and the non-continuous liquids, sharping is distinctive before /a u o/, but not before / i e y/. In a phonological representation the distributional regularities just summarized should obviously be exploited in conformity with Condition 5. They result from distribu-tional constraints which can be stated in a few simple rules. Since the exceptions mentioned above (notes (b) and (c) on Figure IT) are easily delimitable, they do not restrict the validity of these rules (i.e. they are not genuine exceptions), but merely limit their applicability (i.e. they make the rules applicable to specific contexts only). The rules that describe the distribution of non-distinctive sharping before vowels are numbered P 5a, 5c, 8b, 8c, 8d, 8f, 8g. Examples and details of the operation of these rules are found in the corresponding sections of Chapter V. Any of the six vowel phonemes can occur after [ j ] . However, as stated in note (a) on.Figure XV, the sequence [jy] occurs only across morpheme boundaries. This means that the feature grave vs. acute in non-flat diffuse vowels following {j} need not be specified in the representation of morphemes in the dictionary, but can be specified by a 49 morpheme structure rule. Cf. rule MS 12c. 2.212. In i n i t i a l position, as stated in Figure IV, any of the six vowels can occur. 2.22. The last statement, as well as several of those preceding i t , needs qualification. The fact that the orthoepic norms of CLU tolerate certain deviations from the standard just surveyed makes i t necessary to consider also other varieties of CLU. This will be done in the following pages. At the same time, the findings of some other studies of CLU phonology will be discussed. 2.2211. In some varieties of Ukrainian the group of strong paired consonantal phonemes includes the strident dentals (while the non-continuous liquids alone constitute \ 39 the weak group). The differentiation of sharped and plain strident dentals before / i / is by some observers considered atypical of GLU,^ but is considered acceptable according; 39 This is characteristic of dialects of the Dniester Region, of Podolia, and in the south of the Middle Dnieper Region. Cf. F. T. Zylko, Hovory ukrajins*koji movy, Kyjiv, Radjans'ka Skola, 1958, pp. 73 f f . , 107 f f . , 150 f f . 40 Cf. Oleksa Synjavs'kyj, Normy ukrajins'koji literatur-noji movy, 2nd ed., L'viv, Ukrajins'ka Vydavnyctvo, 1941, p. 11; and George Y. Shevelov (pseud. Jury Serex), "Phonema Errans," Lingua, II (1949), p. 401: . . the consonants s, z will always be palatalized before 1 in the literary Tanguage . . . ." 50 to the current Soviet norms.^ For the present description i t was decided to disregard this phenomenon, but i t could easily be accomodated. The present set of phonological rules presupposes a dictionary and a set of transformational rules that would yield sequences like &{|ik}&masc.& past& •chopped1, &{slk}&sg.&nom.& 'juice' t and &{sll}&pl.&gen.& 'villages'', and turn them into {sik}, {sxk}, {sil}. Rule P 8c then turns these into [sik], [gik], [ s i l ] . To yield the pronunciation [sik], [sik], [ s i l ] , typical of the dialects in question, only one change in the phonological rules is necessary, viz the suspension of rule P 8c. In addition, however, i t is necessary that a l l strident dental consonants be specified as sharped before {!}, {i} in the dictionary. 2.2212. In some varieties of Ukrainian the division of the paired consonantal phonemes into strong and weak has been eliminated: a l l paired consonantal phonemes are weak, i.e. are non-distinctively sharped before / i / . This 41 "Usi c i vafiarity [palatalizacija i nepalatalizacija pryholosnyx — I. P.], z ohljadu ha jix posyrenist', treba vyznaty za rivnorjadni, paralel'no isnujuci V mezax l i t e r a -turnoji movy•" 1. A. Bulaxovs'kyj,'ed., Kurs sucasnoji ukraj ins'ko j i literaturno j i movy, I. Kyjiv, 1951, p. 182; quoted by I. Z. Petlycnyj, op_. cit . , p. 18. But see further below. -42 This situation predominates in the south-eastern group, but is also typical of some south-western dialects, e.g. in 51 phenomenon was formerly considered inconsistent with the orthoepic norms of CLU, but is currently not merely accepted, but preferred, in the URSR.^ The spread of this phenomenon seems to be favoured, not only by the attitude of educators like Zylko; and the absence (since 1928) of an orthographic means of differentiating minimal pairs like [dij] 'milk'.' : [cli j ] 'actl' (both spelt d i j ) , [nis] 'nose' : [nis] 'carried' (both nis), or [liz] 'of willows' : [liz] 'crawled' (both l i z ) ; but by the drift of the language i t s e l f . The varieties of CLU that,do not dis-tinguish sharped and plain paired consonantal phonemes before / i / are covered by the alternative rules given in Chapter V, Sec. 3. 2.2215* The opposition of [ji] and [jy] across morpheme boundaries is not maintained by a l l speakers of the Prut Region, and in the Carpathian group. Cf. Zylko, Hovory, pp. 75 f., 105, 154, 151, 160 f., 167. 45 Synjavs'kyj, p. 10 f. 44 Cf. the following remarks by Zylko on the formerly orthoepic distinction between sharped and plain paired con-sonantal phonemes before / i / : "Dexto z poslidnykiv namaha-jet'sja cju rysu obstojuvaty nibyto jak prytamannu sucasnij ukrajins'kij movi. Prote dlja c'oho nemaje pidstav . . . . My wazajemo cju osoblyvist' vymovy peredn'ojazycnyx pryho-losnyx . . . dialektnoju, i j i j i treba unykaty v sucasnij literaturnij movi." ("D.ialektni vidminnosti pryholosnyx fonem ukrajins'koji movy", Ukrajins'ka mova v skoli, XI:4 (1961), p. 12).. CLU. It is described as standard by Synjavs'kyj, but according to Ziiyilski i t is not typical of the speech of the "educated": KaSde akcentowane i , bez waglgdu na swe pochodze-nie, brzmi po ^  w wymowie wyksztalconej zwykle jak ^£ (,1i) . . . . W gwarach wschodnich i zachodnich przewaza w takiej pozycji ^ i (jL szerokie), ktdre w gwarach galicyjskich obniza si§ do £ . . . .46 Unfortunately, this writer has found no mention of this matter in the recent Soviet literature on Ukrainian phonetics available to him. For the speakers that do not distinguish [jl] and [jy], the feature grave vs. acute in non-flat diffuse vowels is predictable after {j} when no word boundary intervenes. For these speakers, the rule MS 12e must be deleted, and a phonological rule with analogous effect must be inserted at a later place (cf. the optional rule given in Chapter V, Sec. 3). 2.222. It was stated above that a l l the six vowel phonemes can occur in i n i t i a l positi on. This statement requires a few comments regarding the status of i n i t i a l [i] and [y]. 45 Synjavskyj, P. 173; and reaffirmed by Shevelov in "Phonema Errans", p. 402. 46 Zilyriski, p. 12. Ziiyriski makes the following observation: Na poczatku wyrazu wymawia sig etym. _i w gwarach wschodnich i w wymowie wyksztalconej wschodniej i zachodniej w pozycji akcentowanej i nieakcentowanej z nleznacznemi wyjatkami jak y® . . . . Gwary za-chodnie wykazuja pod tym wzgleplem wahania. ¥ jed-hych z nich slyszy sie przewaznie y e, w innych zad w pozycji akcentowanej • • •• i w zglosce nieak-centowanej najcze^ciej a. . . . .47 The "unimportant exceptions" he l i s t s are ispyt, im'jat Ivan, a l l native Ukrainian words. Synjavs'kyj limits himself to stating that the stressed orthographic i; in words like insyj, inodi, inde, istyk, iskra, ihraska, irod, ±6 is pronounced as y_, and adds that ". . . ce odna z mozlyvyx vymov, bo vymovljajut' • „48 sja vony 3 z a.." Shevelov mentions the problem of i n i t i a l [ i ] , [y] both in his textbook (Narys, p. 365) and in the article "Phonema Errans" (p. 404, n. 7). In both places he dis-cusses the "intermediate place of /y/ between / i / and /e/. As an illustration of the close affinity of /y/ and / i / , h mentions in Narys that U'toj"cas, jak u•vymovi sliv insyj, inodi, ind^k, istyk, i r i j zberihajet'sja vymdva (xoc i pyset'-sja i - ) , V insyx slovax zvycajno vymovljajet 1sja j.-, napr., ity, im' ,ja, ispyt. 47 Ziiyriski, p. 20. 48 Synjavs'kyj, p. 173. 54 In "Phonema Errans" he evaluates this situation in these words: . . . the indiscriminate use of y_ — i does not exist either, except . . . at the beginning of a word: although linguistic consciousness distin-guishes the pronunciation of y_ — i at the begin-ning of words (ynsyj — ihsyj, orthographical: insyj 1 other *),but the two variants of the pronun-ciation are perceived as equivalent. Prom the above quotations i t is evident that the distribution of i n i t i a l [i] and [y], at least in some varieties of CLU, is determined not by the phonetic context, but by the linguistic habits of the speakers. This means that i n i t i a l [i] and [y] represent two phonemes. The paucity of words with i n i t i a l [i] and [y], and the absence of minimal pairs — together with the regional variations mentioned by Ziiyriski — suffice to justify the orthographic convention of writing only i n i t i a l i . But these facts do not constitute a sufficient basis for considering i n i t i a l [i] and [y] positional, or stylistic, or free variants.^ In this writer's view, Shevelov's assertion that ". . . the two variants of the pronunciation are perceived 49 In the article quoted above, Shevelov seems to say that i n i t i a l [i] and [y], after a l l , are variants of a kind. In the absence of genuine paronyms, a pair like [i*spyt] 'test' and [jfskra] 'spark' would tend to indicate the opposite. In any case, whereas the presence of minimal pairs implies a phonemic opposition, the absence of minimal pairs does not imply the absence of a phonemic opposition. as equivalent." is not accurate. Pronunciations with [i] and [y] are probably equivalent only when the context makes any deviation from the listener's norm relatively insigni-ficant. This is undoubtedly usually the case, but i t is not always. Since the elimination of the opposition between {i} and {y} in i n i t i a l position is accepted by the current Soviet norms, an optional phonological rule has been formu-lated (cf. Chapter V, Sec. 3). It should be noted that, since a l l varieties of Ukrainian distinguish morpheme i n i t i a l {i} and {y} in derivational and inflectional suf-fixes, the morpheme structure rules are not affected. In some of the varieties of Ukrainian that do dis-tinguish i n i t i a l { 1 } and {y} there is some variation from one speaker to another as to the exact distribution of these two morphonemes. Thus, one speaker may habitually say [yskra] 'spark', [ystyk] 'plough-scraper', [£spyt] 'test', 50 Asked for the meaning of the word [iva], one informant repeated the sound sequence several times, but could not assign i t a meaning. Supplied with the additional informa-tion that this might be a sort of plant, possibly a tree, he exclaimed, "Aha, bacyte, iva, a u nas skazaly yva." and proceeded to define iva as a tree similar to the loza (both are kinds of willow). Cf. Troubetzkoy's 2nd rule (P"rin-cipes, p. 49 f»): "Si deux sons apparaissent exaetement dans l a meme position phonique et ne peuvent pas etre sub-stitue's l'un a 1'autre sans modifier la signification des mots ou sans que le mot devienne miconnaissable, alors ces deux sons sont des realisations de deux phonemes diffe'-rents." (Underlining supplied. H.A.). 56 another, [yskra], [istyk], [ispyt]. This sort of variation can only be accounted for as due to differences in morpheme shapes on the dictionary level and is consequently, outside the scope of a phonological description. 2.5. The literature on Ukrainian phonology contains, as was mentioned above, a number of references to a Ukra inian five-vowel system. Since some of these have received general recognition, they deserve a few words of comment. 2.51. Roman Jakobson, in his study of the history of the Russian phonological system, sets up a five-vowel system for the south-eastern dialect group: Dans les parlers orientaux de l'ukrainien meri-dional , les consonnes ont pris le son mou devant tout 1. p rovenant de diphthongue. 2 est naturelle-ment envisage" comme variante extragrammatieale cdmbinatoire de i conditionnee par position apr&s \ eonsonne dure; i l est characteristique que, meme objectivement, J L soit plus proche de i dans les parlers orientaux que dans les parlers occidentaux de l'ukrainien. ....... . ->i Jakobson's description f i t s not a l l the dialects of the south-eastern group, but only those in which non-distinctive softening of a l l consonantal phonemes before / i / coincides with the elimination of the / i / : / y / opposition in i n i t i a l 51 "Remarques sur 1'evolution phonologique duerusse compare'e a celle des autres langues slaves," TCLP II, Prague, 1929. position as well as after / j / . In these dialects, / i / and /y/ can indeed he considered to be in complementary distribution, but only i f labial, palatal, and velar consonants be considered paired. Thus, collapsing / i / and /y/ into one phoneme, * / i / , entails the establishment of a dozen more consonant phonemes having a distinctivecfunction only in one position, viz before * / i / . This is clearly not a very economical solution. It is interesting to note that the phonetic properties of [i] and [y], which Jakobson mentions in passing, do not per se recommend this 52 solution. 2.32. Karol Dejna, in the article mentioned above, follows Jakobson in setting up a six-vowel system, . e a o , for the south-western group, and'a five-vowel I y u • system, ^ e a o u , for the south-eastern dialects (p. 149). At the same time, he considers sharping non-distinctive in labials (p. 149)» velars, and palatals, as well as in the non-continuous liquids (p. 150). This creates a problem: the differentiation of words depends on whether the positional variants [m p b y r] stand before */i/> o r the "positional variant" [y] follows */m p b v r/. It is obviously not satisfactory to represent both [mij] 52 Cf. Preliminaries, p. 5 f. and [myj] as */ja±±/[piw] and [pyw] as */piv/, [bid] and [byj] as and so forth, so Dejna proposes to mark the occurrence of [y] by means of a diacritical mark on a preceding symbol representing a labial consonant or non-continuous liquid (p. 150)."^ This, of course, is tant-amount to recognizing either / i / and /y/ or distinctive sharping in labials and the non-continuous liquids while obscuring which distinction is being recognized as phonemic and which is held to be allophonic, It should be noted that Dejna's description implies that / i / and /y/ are not distinguished after palatal or velar consonants in the south-eastern dialects (i.e. that sequences like [ki], [ky], unlike sequences like [pi], [py], are phonemically identical). This implication finds no corroboration in the dialect texts available to the 54 present writer, and i t is incorrect with reference to CLU. On the other band, Dejna's */r/, corresponding to CLU {rj and ( r j , is typical of most dialects in the south-eastern group. 53 Note in preceding examples that Dejna's * / i / includes the phoneme / j / ; cf. Sec. 1.2. 54 2ylko, Narysy contains 28 texts (comprising some 12000 words) from the Middle Dnieper Region (pp. 271-277). the Sloboda Region (pp 277-288), and the Steppe Region (pp. 288-299). 2.55. Edward Stankiewicz1s characterization of Ukrainian in his;:.paper on Slavic typology agrees largely with Dejna's description:^ Standard Ukrainian has five vowels, the Western Ukrainian dialects, six (p. 311). The feature sharp vs. plain is distinctive only for dental consonants and continuous liquids. He adds that " . . . in Standard Ukrainian . . . / r ' / is a free variant [sic]." (p. 309) There would be l i t t l e point in discussing this description furtherif i t were not for the fact that Stankiewicz has tried to apply i t , without major changes, 56 in his study of Ukrainian morphology. Instead of the automatic distribution of plain and sharped consonantal segments before / i / and /y/ which was outlined above (Sec. 2.211), Stankiewicz finds a maze of morphophonemic alternations which have no ". . . clear morphological function, but are distributed, as i f hap-hazardly, in specific grammatical categories . . . ." and 55 "Towards a phonemic typology of the Slavic languages," American Contributions tothe Fourth International Congress  of Slavicists, 's-Gravenhage, Mouton. 1958, pp. 301-19. 56 "The consonant alternations in the Slavic declen-sions," Word.XVI (I960), pp. 183-205. From the examples */javor»i/ and •/ja.v'ori/ (p. 190, n. 11) i t is apparent that Stankiewicz has modified his views regarding the status of /r?/. (The two words should be written /javoyi/ (with the proper /i/) and /javory/.) 57 are " . . . quite complicated." The reasons are obvious. Given the CLU forms of the adjective 'lean' listed in Column A below, the eastern forms in Column B can easily be inferred by means of a rule that sharps a l l consonantal segnents before / i / . If / i / and /y/, however, are col-lapsed into a * / i / , as in Column C, i t becomes necessary to state for each single suffix beginning with * / i / whether or not i t requires a preceding czmsonantal segment to be sharped.^ A B C nom.sg.masc. /xud-yj/ /xud-yj/ */xud-ij/ dat.sg.fem. /xud-£j/ /xud'rfj/ */xu$-ij/ in.sg.masc. /xud-ym/ /xud-^ m/ */xud-i*m/ nom.pl. /xud-i/ /xu^-i/ */xa§-i/ gen.pl. • /xud-yx/ /xud-yx/ */xud-ix/ 57 0|>. c i t . , p. 189 f. and n. 11 58 Stankiewicz's remark, o£. c i t . , p. 197, that ". . . i t is interesting to notice the phenomenon of free variation in the final consonants of adjectival stems; the tendency is however, to generalize either the hard or the soft con-sonant throughout the declension of adjectives . . . ." may be noted in passing. It is supported with a reference to Zylko, CTarysy, pp. 165, 174. Characteristic of the dialects in question is not the free variation in the final consonants of adjectival stems, but the fact that "Porivnjano najbil's vyrazno zberezene . . . davnje rozriznjuvannja tverdyx i m'jakyx osnov prykmetnykiv. . . .' (Zylko, Narysy, p. 162). The examples on pp. 165 and 174 illustrate how the generalization of hard stem final con-sonants in the adjectival paradigm, typical of CLU (cf. Column A above),. cannot be maintained in some of the dialects Stankiewicz is discussing (cf. Column B above). / Prom the above considerations i t should be clear that a morphonemic solution analogous to. the phonemic solu-tion advocated by Dejna and Stankiewicz would f a i l to satisfy Condition 5 hy violating Conditions 3 and 4. Such a solution could be modified so as to satisfy Condition 3 in one of two ways: by the addition of a dozen consonant morphonemes, or by the recognition of {£} and {y} (and {i} and {y} ) as distinct functional units. The latter solution is the more economical and, unlike the former, makes i t possible to satisfy Condition 4. It is therefore to be preferred. The Consonants. The use of the features consonantal vs. non-consonantal and vocalic vs. non-vocalic to define the class 50 of consonants in CLU requires no comment. The feature compact vs. diffuse, which is the acoustic correlate of the back-front opposition, and the feature grave vs. acute, which is the correlate of the opposition between peripheral and medial point of articulation, are also used in conform-60 ity with the definitions in Fundamentals.- They divide the consonants into four groups, the labials, the dentals, 59 Cf. Fundamentals, p. 29, and Fant, p. 215 f f . 60 Pp. 29, 31} and Fant, pp. 217-219 the palatals, and the velars, which will he discussed below in that order. Of the remaining features that function distinc-tively in the consonants, those which are relevant only to one or two of the four groups - the features strident vs. mellow, nasal vs. non-nasal, and sharped vs. plain - will be discussed in the respective sections below (3.11 and 3.21). In the same connection a few remarks will be made on the use of the feature continuous vs. interrupted. In the present section, only the feature tense vs. lax, which is distinctive for a l l consonants except {m}, {n}, {n}, will be commented on. 3.OIO. Ukrainian is generally described as having a distinctive feature of voicing. At the same time, however, i t is well known that most varieties of Ukrainian, in-cluding CLU, differ from the majority of other Slavic languages by the distribution of this feature. While in other Slavic languages only voiceless obstruents are tolerated in word final .position, CLU here preserves the distinction between voiced and voiceless consonants. While in other Slavic languages obstruent clusters follow the rules "voiceless before voiceless" and "voiced before voiced", only the latter rule is generally valid in CLU (with the notable exception that voiceless obstruents are admitted before {v}; cf. rules P 1 and 2a). The solution adopted here has, to this writer's knowledge, been considered before only by Shevelov. In "Phonema Errans" (p. 418), Shevelov expresses his view that in CLU voicing is distinctive, and not the feature tense vs. lax, but admits the possibility of a develop-ment in the direction of the latter feature. In this con' nection he, refers to Troubetzkoy's discussion of these features. 3.011. In the passage in question, Troubetzkoy considers the cases in which a distinctive function is fu l f i l l e d by several secondary manner features in com-bination. This is clearly the case in CLU, where the voiced obstruents always are lax, and the voiceless ob-struents always tense. In such cases i t may be possible to determine which of the features is phonologically relevant,and which redundant, i f the opposition is neut-ralizable. For in the position of neutralization, as a rule, the unmarked member of the correlation represents the arehiphoneme . ^  As mentioned above, the opposition between the feature complexes voiced-lax vs. voiceless-tense is 61 Principes, p. 167; Shevelov refers to Grundziige. p. 141. 62 Cf. also Principeey p. 84. neutralized in position before a voiced-lax obstruent other than {v}. In this position only voiced-lax obstruents are admitted. Since the feature whose unmarked term appears in the position of neutralization is the feature tense vs. lax, this feature must be deemed distinctive. The analytic technique just outlined, and the principles on which i t is based, seem so valuable that they should not be abandoned except - to quote Troubetzkoy -. . . l a ou le systeme phonologique en question con-tient des indications directes sur une autre repar-tition 'non naturelle 1 du caractere marque et du caractere non marque* des termes de 1' opposition. . . .63 In declining to accept the conclusion reached above, Sheve-lov makes no mention of any evidence that the feature voiced vs. voiceless should function distinctively in CLU despite its "irregular" distribution. This writer knows of 64 no such "direct indications." 5.012. On the contrary, i t has repeatedly been observed that in many varieties of Ukrainian, in the position before a voiceless-tense obstruent, the opposition of voiceless-tense and voiced-lax obstruents is reduced to 63 Principes, p. 167. 64 It is'clear that the dialects in the extreme West in which the distribution of voicing follows the general Slavic pattern must be left out of consideration here. In these dialects, voicing is evidently distinctive. Cf. Zilyriski, p. 120. an opposition of voiceless-tense and voiceless-lax obstruents. This is an unmistakeable sign that the feature of voicing is merely accessory. Another striking indication thatthe feature tense vs. lax must be considered distinctive is provided by CLU {h}, the lax counterpart of {x}. {h} is traditionally defined as voiced, but outside the eastern dialects where voicing usually accompanies laxness i t is hardly realized as voiced with any degree of consistency, except in a careful, "maximum" redundancy" style. In a relaxed style, there is probably only one position in which {h} is consistently voiced by a l l speakers, viz the position before another lax (and voiced) obstruent. This is the position in which the opposition {h}:{x} is neutralized. In a l l other positions, {h} may be voiced or voiceless, but i t is always distinct from.{x} by its relative lack of friction, i.e. by its smaller amount of energy. On this basis i t is clearly 65 0. Broch's article, "Zum Kleinrussischen in Ungarn," Archiv fur slavische Philologie, XVII, p. 321 f f . , unfortunately could "not be consulted. Cf. Olaf Broch, Slavische Phonetik, Heidelberg, Carl Winter, 1911, p. 69 f.; and F. T. Zylko ? Rovory? pp. 40, 44, 103, 108, 152, and 167 f f . ; according to Ziiyriski (p. 119), partial devoicing of lax obstruents in this position is steadily becoming more wide-spread (before 1932). 66 Cf. Broch (Slavische Phonetik, p. 80) on voiced and voiceless [h]: ". . . "eine'feste Abgrenzung der einen Form von der anderen ist nicht durchgef iihrt. . . . " This impossible to consider voicing an invariant characteristic of {h}. It is evident from the preceding remarks that the conclusion reached above on the basis of distributional considerations finds support in the phonetic data of the language. It may be argued that, since the orthoepic norms of CLU do not sanction the devoicing of lax obstruents, but stipulate a complete distributional congruence of the features of voicing and tension, CLU could be described equally well i f voicing were considered distinctive. This is obviously true. However, since the .solution adopted here accomodates several varieties of Ukrainian without any loss of accuracy of description with regard to CLU, i t is to be preferred. 3«02• The distinctive feature tense vs. lax is defined as ". . . higher (vs. lower) total amount of energy in conjunction with a greater (vs. smaller) spread of energy in the spectrum and in time." According to Pant, the greater length of tense consonants may be more statement is elaborated and confirmed by Zilynski, p. 102. Though" this fact has been known for more than half a century, its significance has formerly been ignored due to the traditional adherence to exact articulatory definitions: when /x/ is defined as velar, and /h/ as laryngeal, their structural relationship is obscured. 67 Fundamental^, p. 30. 67 important, from an acoustic point of view, than their great-go er noisiness. Auditorily, however, the latter property seems to he the more prominent. Needless to say, the re-dundant feature of voicing may he an important cue for the tenseness-laxness opposition. 3.03* In his article "Phonema Errans", as mentioned above (Sec. 3.010), Shevelov admits the possibility of a development in the direction of the feature tense vs. lax. This admission seems to involve two assumptions, neither of which is acceptable. 3.051. The f i r s t assumption, which Shevelov shares with a l l other investigators known to this writer, is that Ukrainian hitherto has had a distinctive feature of voicing. This assumption is not well founded. With regard to present day Ukrainian the role of the feature of voicing has been outlined in the preceding pages. It was seen there that voicing cannot be considered distinc-tive. With regard to earlier forms of the language the following can be said. It is generally supposed that Common Slavic had a distinctive feature of voicing. It is 68 Pant, p. 224 f. 68 further supposed that this distinctive feature of voicing persisted in a l l the Slavic languages, undergoing no changes except.the distributional ones produced by the f a l l of the gq jers. y In Ukrainian no distributional changes are attest-. . . . . . s • ed except the elimination of the opposition between "voiced" and "voiceless" obstruents in position before a "voiced" obstruent. It must not be forgotten, however, that the only evidence for a distinctive feature of voicing in Common Slavic is comparative: voicing is presumed to have been distinctive in Gommon Slavic because i t is assumed to be distinctive in a l l modern Slavic languages. Acknowledging that C1U has a distinctive feature of tension, rather than voicing, means the introduction of new comparative evidence, which must prompt a revision of the history of the Slavic obstruents. Shevelov's second assumption is that Ukrainian at present is going through a development which may lead to the establishment of the distinctive feature of tension. Shevelov's statement is not very explicit, but presumably i t is the spreading phenomenon of voiceless lax obstruents (cf. Sec. 3*012, n. 65) he has in mind. This assumption 69 A. Meillet, Le slave commun, Paris, Champion, 1924, p.22. seems to be based on an incorrect interpretation of the data. Aside from the Ukrainian dialects in which voicing is distinctive (cf. Sec. 3.011, n. 64), Ukrainian dialects treat the feature of voicing in two typical ways. In one part of; the dialects (predominantly in the East, and in-cluding CLU), lax obstruents (with the possible exception of /la./) are fair l y consistently voiced. In the other part of the dialects (predominantly in the West, but also sporadically in the East), lax obstruents are consistently voiced in voiced contexts, but tend to be voieeless :Jbefore voiceless obstruents. In both groups there is only one "position of neutralization", viz before a lax-voiced obstruent. There appear-- to be no reasons for assuming any structural difference between these two groups with regard to the 'obstruents. They differ only in the distribution of the redundant feature of voicing. One might say that the dialects that admit voiceless lax obstruents merely carry the consequences of the structural facts one step further and eliminate the redundancy of voicing wherever it is convenient. While i t is-clear that a development in this direction is taking place, i t is equally clear that this development is not bringing about any reorganization of the phonological system.^0 Regarding the history of the distinctive feature of tension, the following hypothesis can be tentatively for-mulated. 3.052. Before the f a l l of the jers two series of obstruents were opposed. The members of one series were voiced and lax, those of the other were voiceless and 71 tense. Their relation answered Troubetzkoy1s description: . . . l a ou plusieurs princij>es de differenciation sont combine's entre eux et . . . ou la nature de la neutralisation de 1'opposition ne donne aucune indication sur le caractlre marque" ou non marque^  des teriaes de 1'opposition, l'exacte determination de l a nature d'une correlation de mode de franchisse-ment du second degre est a proprement parler impossible.72 For in obstruent clusters a l l members were voiced/lax or voiceless/tense, while no obstruents were admitted in final 73 position. 70 Such a rebrgahizatibh is possibly taking place in those south-western dialects where voiceless/lax obstruents tend" to become" tense This development, which is replacing the distinctive feature of tension with "that of Voicing, cannot be what Shevelov had in mind. It is geographically strictly limited,and i s , as yet, irrelevant for the greater part of the Ukrainian dialects including CLU. Cf. Ziiyriski, p." 121 f. In this connection i t is to be regretted that a l l the work On Ukrainian dialectology has been based oh theassumption that voicing, and not tension, was distinc-tive in the obstruents. One cannot read Ziiyriski'without feeling that many "valuable generalizations have been omit-ted because no structural importance was attached to the feature of tension. 71 Cf. Meillet, p. 13. 71 Before the f a l l of the jers and the appearance of word final obstruents as well as heterogeneous obstruent clusters, the opposition voiced/lax vs. voiceless/tense had been interpreted differently in different Slavic dialects. In some dialects this opposition had been interpreted as one of voicing. These dialects subjected the new obstruent Clusters to the -rule "voiced before voiced and voiceless before voiceless" and tolerated only voiceless obstruents in final position. In other Slavic dialects, the opposi-tion voiced/lax vs. voiceless/tense had been interpreted as one of tension. These dialects developed the phono-logical rule "lax before lax". Some complemented this rule with the rule "tense before t e n s e " C h a r a c t e r i s t i c of a l l these dialects was the persistence of the tense vs. lax opposition in word final position: the "voicelessness" of a following pause has no influence on distinctively tense or lax obstruents. It is impossible to attempt an exact delimitation of these two groups of Slavic dialects. That must be the subject of a separate study. But i t seems clear that, for 7 2 Principes, p. 167 f f . 73 Meillet, p. 16 74 Cf. Ziiynski, p. 121. instance, Polish and the Russian dialects on which Standard Russian is based,, developed distinctive voicing. Whereas the South Russian dialects and Ukrainian (with the excep-tion of the dialects in the extreme West) belong to the second group. In the following discussions of the CLU obstruents, an attempt will be made to interpret some of the charac-teristic traits of Ukrainian in the light of this hypothesis. It will be seen thatit goes a long way towards giving a rational explanation to some of the "phonemata errantia" of Ukrainian. The Labial Consonants 5.1. The non-distinctive sharping of labials before [i] and the distinctive sharping of labials before [a], [u]> [°]» reflecting a morphonemic sequence of labial f o l -lowed by {j}, have already been discussed above (respective-ly Sees. 2.211 and 1.522). In the present section only two problems will be discussed: the feature composition of the labial consonants, and the status of two of them, {f} and {v}. 5.11. The feature composition of the labial conso-nants which is suggested in Chapter II-A is not maximally 73 economical. The five segments are differentiated by thirteen feature statements, while the practical minimum is twelve statements. A more economical statement could be achieved by adopting a solution analogous to the one proposed by Morris Halle for Russian: {b} {P} {m} {v} {f} Strident - - -. - + + Nasal - + o o Tense - + o - + The fact that CLU {v} is consistently represented by a non-fricative [w] after a vowel when not followed by a non-vocalic segment, however, does not recommend this solution. The solution adopted here seems more realistic, and i t was decided to forgo the more elegant statement presented above. 3*121. The vocabulary of CLU includes a considerable, and increasing, number of items containing the sound [ f ] . Only a few of these words are of native origin, some of the latter being onomatopoeic or having a more or less marked regional flavour (e.g. f 1 jakaty 'hiss', or flin'katy 'sob'). The vast majority of these words are loan words, drawn partly from Western European languages, partly from the stock of international words of Greco-Latin origin. Many of the Ukrainian dialects do not have a phoneme 74 / f / and render CLU words like fakt 'fact*, profesija 'trade', profspilka .'trade union' [xteakt], [proxfe^sija] , [proxspilka], or [xwakt], [proxwesija]. 5 In his description of "General Ukrainian" ("jgzyk 76 og.-ukr."} Karol Dejna asserts that "Nie jest fonemem f". In view of what was said above, Dejna's position is undoubt-edly defensible. In a description of CLU, however, i t seems more realistic to grant that the language has a morphoneme {fj and then note its limited occurrence. That is the solution adopted here. 3*122. While there is no doubt that {v} is part of the phonological system of CLU, i t seems natural to question its position within the system. The definition decided on in the present paper does not, mutatis mutandis, differ from those given /v/ in other descriptions of CLU. Never-theless, {v} deserves a few words of comment. The fact that friction is not an essential attribute of any of the allophones of {v}, and in particular the existence of the allophone [w], suggest the possibility of interpreting {v} as the grave counterpart of the acute .{j}. The great phonetic latitude of {v} itself ([w] after vowel 75 Of. Zilynski, p. 57. 76 Dejna, p. 149-75 when no non-vocalic segment follows; [y] before [ i ] , [v] and [te] in free variation elsewhere) seems to indicate a less strictly circumscribed place in the system for this morphoneme. (Cf. Troubetzkoy, Trincipes, p. 74 ff.) If {v} is interpreted as a glide, *{w}, the close affinity of this glide with which is suggested by the facts described above, seems to be confirmed by the dis-tributional data. In position after a consonant, *{w} and *{j} are in complementary distribution: after labial con-sonants, *{w} is not admitted, and after non-labial con-sonants, *{j} is not admitted. At the same time, however, such an interpretation creates a number of difficulties. If *{w} and con-stitute the class of glides, the distributional character-istics of this class are identical with those of the class of consonants, for {v} has none of the distributional characteristics that set {j} apart from the rest of the seg-ments. As a consequence, this,interpretation makes impossi-ble a l l the morpheme structure rules specifying the features consonantal vs. non-consonantal and vocalic vs non^vocalic in segments contiguous to glides. Since, also, a l l state-ments which in this description and the morphological rules i t presupposes refer to grave diffuse consonants would be complicated appreciably i f this interpretation were accepted 76 ({v} patterns with the labials* cf. rules P 5a, 8a and 9c;'* and the grammatical alternations {v}~{vj}~{vl}), i t should not be. While the partial complementation of *{w} and described above would be expressed in a morpheme structure rule assigning the feature grave vs. acute to any glide in position after a consonant, the partial complementation of {v} and {j} is only partially reflected in the morpheme structure rules of the present description. The non-occurrence of {j} after consonants other than labials results in the formulation of rule MS 12b. But the non-occurrence of {v} after {b}, {p}, {f}, {m} is not reflected in a morpheme structure rule. For a few further remarks, see Chapter XV, Sees. 2 .2 and 3 . 2 . 5 . 1 3 . While {f} thus is included among the CLU morphonemes, i t cannot be ignored that its status is rather different from that of, for instance, the corresponding Russian morphonemes. It is clear that, unlike its Russian correspondents, CLU {f} is the consequence of heavy borrow-ing during the last century or so. The earlier inability of Ukrainian to accept loanwords containing [f] without modification, as compared with the apparent ease with which Russian has assimilated a steady trickle of such words during the last several centuries, demands a 77 structural explanation. Roman Jakobson discusses the borrowing o'f foreign phonemes in his "Principes de phonologie historique", where he makes the following general statement: Les phonemes etrangers que la langue s'approprie le plus aisement sont ceux qui s*incorporent dans les correlations deja existantes.' 8 This statement should apply to the borrowing of / f / equally well from the point of view of Russian and from the point of view of Ukrainian, for in both languages {v} was an "unpaired" participant in a correlation embracing a l l the obstruents. It is plain, however, that Jakobson's state-ment does not explain the different treatment of [f] in the two languages. If the hypothesis outlined above (Sec. 3.032) is accepted, the different treatment of [f] in Russian and Ukrainian correlates with the different status of {v} in the two languages. In early Russian, where, after the f a l l of the jers, {v}, {y} were marked unpaired participants in the correlation of voicing, foreign [f], [f] were identical with the realizations of {v}, {y} in positions of neutralization, and their phonologization provided {v}, {y} with unmarked counterparts. In Early Ukrainian, {v}, {y} were unpaired participants in the correlation of tension. 78 Troubetzkoy, Principes, Appendice I, p. 323. Since they were unmarked, they could be realized as lax in a l l positions, as is their modern correspondent, CLU {v}. In his discussion of dephonologization in the article quoted above, Jakobson mentions that II est caracteristique que, dans la suppression des correlations, ce^soit d'ordinaire ^ustement le terme correlatif marque qui est supprime". . . . '9 The different treatment of foreign [f] by Russian and Ukrainian suggests that this statement is true "with opposite signs" with reference to the extension of a cor-relation. An existing correlation is most easily extended to embrace a new phoneme when the latter is unmarked. Cf. the discussion of foreign [3] below (Sec. 3.23). With reference to the borrowing of words containing [f] Jakobson makesthe following statement: Dans les cas ou existe une tendance a russifier com-pletement le'mot emprunte ayant un f, ce f a ete rem-place par xv, x, °£L P_«^ The replacement of foreign [f] with {p} or {kv}, more often with {x} or {xv}, which is typical of various historical and geographical varieties of Ukrainian, demands a structural investigation. It would appear that, whereas the rendering of foreign [f] as [p] or [x] is equally 79 Op. c i t . , p. 321 80 0p_. c i t . , p. 323. 79 possible whether the [f] is heard as voiceless or tense, the heterogeneous clusters [kv] and [xv] indicate the [f] is perceived as tense (the deficiency of the lax [v] is eked out with a tense grave obstruent). A study of the times and places at"which such.replacements have been or are, made may provide important clues to an understanding of the geographical and temporal distribution of the dis-tinctive feature tense vs. lax, and may, at the same time, throw light on the chronology of other phenomena in the history of the Slavic languages. A further remark on the status of {v} from a his-torical point of view is made in Sec. 3 . 4 3 . The Dental Consonants I 5 . 2 . The dental consonants have already been com-mented on several times in the preceding sections. Geminate dentals were dealt with in Sec. 1 . 5 1 3 . The section on vowels discussed the distribution of sharped and plain dentals before {i}, {i} (sees. 2 . 2 1 1 , 2 . 2 2 1 ) . Their d i s t r i -bution before {y}, {y}, {e}, {e} was touched upon in Sec. 2 . 2 1 1 and will be discussed in Chapter III-B, Sec. 4 . The distribution of the feature sharp vs. plain in sequences of consonantal segments is most conveniently discussed in con-nection with the phonological rules in Chapter V. In the present section, only two points will be commented on, 80 the feature composition of the dental consonants, and the status of the dental affricates. '5»21. The specification of the dental consonants in terms of distinctive features presents no problems. The feature strident vs. mellow naturally divides the dentals into two groups. In the non-strident dentals, the feature nasal vs. non-nasal distinguishes {n}, {n} from the mellow stops, whereas, in the strident dentals, the feature con-tinuous vs. interrupted separates the dental fricatives from the corresponding affricates. The three features just mentioned are used in the present description in conformity with the definitions in Fundamentals (p. 30 f.) and require no further comment. The feature sharp vs. plain, which is distinctive for a l l dentals and, among the consonants, is distinctive only for the dentals, also requires no further comment. (Cf. Fundamentals, p. 31) 3.22. The dental consonants present no particular problems of analysis. None of the earlier descriptions of CLU known to this writer differs from the present des-cription with respect to either the number or, mutatis mu-tandis, the definitions of these consonants. The presence, in CLU, of a f u l l series of non-continuous strident dental consonants makes a lengthy discussion of the analytic procedures required to determine their monophonematicity 81 seem idle. Suffice i t , in passing, to point out that a reasonably efficient formulation of the phonological rules requires that the affricates be considered strident (which precludes their interpretation as sequences of mellow stop followed by continuant) as well as monophonematic. As examples of rules that would be complicated considerably i f the affricates were considered sequences, rules P 4a, 4b, 8b, 9c, 9d, 9f may be mentioned. The solution adopted here agrees well with the phonetic facts: affricates are pro-duced with a one-stage release (rather than the release plus friction characteristic of clusters of stop followed by homorganic fricative), and their length does not per-ceptibly exceed that of single segments. At the same time, this solution facilitates the simplest possible statements of the grammatical alternations in the morphological rules (cf. the parallelism' in, for instance, {k}~{5}, {$}~{c}, {c}~{c}; {!}~(3}, {3H5}). . 3 . 2 3 . Of the four dental affricates, {9} has by far the greatest functional yield. It occurs in a considerable number of native Ukrainian roots, as well as in some pro-ductive suffixes (e.g. {#9}, {y9}). {c} { 3 ) , and {^ }, on the other hand, might be considered marginal. Their dic-tionary frequency is relatively low, and an appreciable portion of the roots in which they occur is of non-Ukrainian 82 origin. In this connection i t must not be forgotten, however, that their existence has a good deal of systemic support. The status of {3} and {^ } has been questioned by Shevelov, f i r s t only cursorily in "Phonema Errans" (p. 417 f . ) , later in a separate study, "A latent phoneme in making: 0-1 The affricate 3 in Slavic." His views require a few words of comment. In "Phonema Errans" Shevelov limits himself to a few observations on the opposition {§}:{z} in Ukrainian. He notes, f i r s t l y , that in some dialects of Ukrainian, the i n i t i a l sequence [zv] has been consistently dissimilated to [3V], whereas in other dialects (as in CLU), an opposition between i n i t i a l [zv] and [3V] is maintained (cf. CLU {zvfr} 'ravine' : {3vin} 'bell'). Secondly, he asserts that ". . . i n certain words 3 - z are used indiscriminately (zygary - zygary 'the clock'), and in other words they are retained merely by tradition ^erkalo 'the mirror', 3yga 'the top'). . . ."(p. 418). He draws the conclusion that {3} and {^} are ". . . in a state of oscillation." If this conclusion means that the distribution of 81 Miscelahea homenaje a Andre Martinet: Estructuralismo e hi scor ia , ed. Diego Catalan^ Biblioteca filo'logica de la Universidad"'de la Laguna, I, Madrid, Editorial G-redos, 1957, pp. 251-276. 83 {z} and {3} in related morphemes varies somewhat from dia-lect to dialect, i t is undoubtedly correct. But i t f a i l s to be of any significance with respect to the status of the opposition {z}:^} in any one variety of Ukrainian.®^ In !\A latent phoneme in making" Shevelov partly-reaffirms the hypothesis that Polish influences are respon-sible for the appearance of {3} in Ukrainian in the six-teenth century (p. 266), partly proposes the new hypo-thesis that Ukrainian contacts with a Roumanian speaking "shepherd stratum" in the Carpathians led to the intro-duction of {3} somewhat earlier (p. 269 f f . ) . He estab-lishes that {3} was used in the native vocabulary of Uk-rainian as early as 1596 (p. 263) and l i s t s some sixty root words containing {3} (p. 260 f f ) . "The l i s t is certainly not exhaustive." (p. 263, n. 30). In addition he mentions that {3} f i l l e d a gap in the phonological system, where {c} lacked a "voiced" partner (pp. 266, 274). Prom these premises Shevelov draws the non sequitur that {3} is ". . ..almost . . . a phoneme, but not a real phoneme." (p. 275).Q-3 82 Shevelov's view (reflected also in the quotation in Sec. 1.311, n. 16) that certain phonemic distinctions are part of the phonological system while others are retained "merely by tradition" is unacceptable to this writer. It' seems d i f f i -cult to imagine how any phonemic distinctions could be re-tained except by tradition. 84 Despite its somewhat illogical argumentation and i t s complete disregard for the synchronic facts on the basis of which alone i t is possible to establish whether a given sound type i s , or isnot, a phoneme, Shevelov1s "A latent phoneme in making" contains much valuable material. In connection with the discussion of the possible Polish influences on Ukrainian at the time of the emergence of {3}, Shevelov states that 1. 3 spread very quickly to the easternmost regions of The Ukraine where no major Pol. influence was exerted /dzveno - 1665, Poltava, dzvon - 1670, Rese-tylivka/, and 2. encompassed there and everywhere words and categories of words never affected by the change [of z to 3] in Pol. /e.g. dzykhar 1695, Poltava [/]. . .-."84 Shevelov's. doubts that the appearance of {3} in the Eastern Ukraine was due to Polish influences seem entirely reasonable. It could similarly be doubted that the Rouma-nian speaking population in the Carpathians can have exerted any "major direct" influence in that part of the Ukraine. Whatever were the external factors that helped br^ing about the emergence of {3}, i t seems obvious that strong internal 83 Shevlov's historical criteria for determining whether a given phoneme is "a real phoneme" are these, i t would seem: its existence must be attributable either to ". . . rise by flood of foreign words borrowed from a given foreign language." (p. 275) 84 0p_» c i t * , p. 266. 85 forces were at work at the same time. Otherwise the spread of {3} in the Eastern Ukraine would not have been possible. With regard to a potential internal force, Shevelov repeatedly refers to the empty slots,, in the series of s t r i -dent dentals, where only the continuous segments formed a complete set. He also mentions the Ukrainian " . . . ten-dency to have fully symmetrical oppositions in voicing . . . ." (p. 266). If the hypothesis outlined in Sec. 3.032 is accepted, a more compelling internal force can be seen, {c}, which was unpaired and marked, was an element of instability in the system. The introduction of [3] or [3] in positions other than where the tense vs. lax opposition was neutraliz.ed sufficed to eliminate the imbalance. This was achieved in various ways. The change of [z] to [3] in i n i t i a l p osition before [v], for instance, made [3] a positional variant of both {z} and {c}, and would tend to dissociate [3] from either. On the other hand the bor-rowing of foreign words containing [3], and the introduction of [3] in foreign words not containing this sound, would lead, to the same result. With regard to the " . . . tendency to have.fully 85 Thus, through a distributional constraint on {zj. a new"morphoneme would be established at the expense of the functional yield of the feature continuant vs. interrupted (cf. R. Jakobson, "Principes de phonologie historique," P. 327 f.) 86 symmetrical oppositions in voicing. . . ." i t is interesting to compare the different ways in which the foreign sounds {f} and {3} were treated in Ukrainian. Neither of these sounds arose as a consequence of "a normal and general sound law." Both of them f i l l e d empty slots in the phonological system. But, while {f} could he established only after "a flood of foreign words," {3} was preserved in early loans, was introduced in loanwords where i t did not belong, appeared in a number of native words, and has been retained in the phonological system of Ukrainian for several centuries despite its relatively low dictionary frequency. This dif-ference cannot be accounted for by the tendency towards symmetry. It may, however, be accounted for by the principle proposed in Sec. 3.13 to explain the different treatment of foreign [f] in Russian and Ukrainian. The lack of regularity in the change of [z] to [3] in the native vocabulary of Ukrainian suggests a development that was begun and halted. A structural investigation into the conditioning factors of this development, with a proper-ly balanced emphasis on both internal and external motive forces, is of the highest relevance for an understanding of the history of Ukrainian. The possible existence of similar developments in the history of other Slavic languages makes such an investigation relevant to the history of the Slavic languages in general. The Palatal Consonants 5*5« The palatal consonants have already been touched upon in preceding sections. The non-distinctive sharping of palatals before {1} was mentioned in bec. 2.211. The two different reflexes of morphonemic sequences of palatal f o l -lowed by {j}, the sharped geminates and the distinctively sharped palatals, were discussed in Sec. 1.321. The auto-matic alternations involving palatal and dental consonants are most conveniently discussed in connection with the phonological rules P 3a, 4a, and 4b in Chapter V, Sec. 1.2. 3.51» The palatal consonants present only one pos-sible problem of analysis, the question of the status of {3}. According to Shevelov, {3}, like {§}, is in "a state f\f> of oscillation." With regard to CLU, however, the status of {3} is clear. It occurs in only a very small number of roots, some of them of native, some of them of foreign origin. In addition, i t is in regular grammatical alter-nation with {3} in one of the productive verb classes ( f u l l stems ending in {y.}). It is thus - to borrow a phrase 86 "Phonema Errans," p. 417 87 Shevelov's view that the status of {3} is unstable is possibly" based on the' fact' that" the" alternation |d}""{3} has been eliminated"in the conjugation of verbs in {y} in some Ukrainian dialects (predominantly' in the Sloboda and Steppe Regions). In these dialects,^ forms like /xo<|u kru$u.i vogu nosu/ correspond to CLU / X 0 3 U krucu vozu nosu/ (cf. 2ylko, 88 from Stankiewicz - ". . . stable in the system, but rare in op contexts." The question of the origin of Ukrainian { 3 } has been 89 the subject of a great deal of discussion. Essentially, the problem is whether the Ukrainian alternation {cl}~{3} is the product of a relatively recent analogy with {$}~{c}, or has been preserved in the language since [dj] f i r s t changed to [ 3 ] in Early East Slavic. In the former case, the early change of [ 3 ] to [z] can be assumed to have been general, in the latter i t must have left unaffected those [ 3 ] that a l -ternated with [d] in Ukraninian alone. This is not the place to add to the debate on this dilemma. But i t should be noted that the hypothesis out-lined in Sec. 3.032 may be of some relevance to the solution. If this hypothesis is accepted, the history of [ 3 ] can be i\iarysy, p. 163) It is not clear to this writer to what ex-tent this alternation is preserved elsewhere in the mor-phology of these dialects.' As was remarked in the case of { 3 } (cf. Sec. 3.23), the uneven distribution of a given linguistic phenomenon among different varieties of a lan-guage does not mean that this phenomenon is in "a state of oscillation" in any one variety of that language. 88 "Towards a phonemic typology of the Slavic languages," p. 3 0 7 . . . 89 Of. Ii. A. Bulaxovs'kyj, Pytannja poxodzennia ukrajin-s'koji movy, Kyjiv, Vyd. Ak. Nauk URSR, 1956, -passim', particularly Chapter XI, pp. 169-194. 89 described in maximally simple terms, viz in terms of two general and well motivated developments. The f i r s t develop-ment changed both [ 3 ] (from [dj]) and [ 3 ] (from [g]) to the homorganic fricatives, thus precluding the establishment of a series of affricates marked for voice (cf. the quotation from R. Jakobson, "Principes de phonologie historique," Sec. 3.13, n. 79). The second development introduced [ 3 ] , [ 3 ] , and [ 3 ] in those East Slavic dialects that had evolved a distinctive feature of tension, and provided the unpaired, marked affricates with the unmarked counterparts necessary for the stability of the system. The Velar Consonants. The non-distinctive sharping of the velar con-sonants before {1} was mentioned in Sec. 2.211. Aside from the automatic distribution of the feature tense vs. lax in obstruent clusters (rules P 1 and 2a), the velars are in-volved in only one automatic alternation, viz their replace-ment by the corresponding palatal consonants before {j}. This alternation, which might have been mentioned in Sec. 1.. 321, but was left unconsidered for the sake of simplicity, is described in rule P 2c. Examples can be found in the corresponding section of Chapter V. 3.41. In his "Ponologiczny system jgzyka ukrain-skiego" Karol Dejna states that " . . . nie jest fonemem ,: 90 £." (p. 150). According to Shevelov, "the sound £ . . . has stood t i l l this day on the borderline between the system of 90 Ukrainian consonants and consonants outside this system."-' {g}, then, requires a few words of comment. {g} is limited to a relatively small number of native roots and assimilated loans. Formerly, {g} furthermore rendered foreign [g] in the Western Ukraine, where Polish influence was strong. In the east, however, where Russian influence prevailed, foreign [g] in bookish loans was usually rendered by {h} (Ukrainian h equals Russian g). In an attempt to standardize the written language, the letter £ was abolished in 1946. As a consequence, {h} is replacing {g}, not only in words that formerly varied regionally, but also in words that formerly were pronounced r t 91 with |_gj everywhere. The future of {g} in the URSR is uncertain. It is interesting to note Zylko 1s view that De j aky j zanepad f onerny £ az ni jak ne svidcyt' , sco neju treba nextuvaty v literaturnij movi. Fonema £ 90 "Phonema Errans," p. 417 •91 E. T. Zylko, "dialektni vidminnosti pryholosnyx fohem," p. 11: "Za ostannij cas . . . ' [JJ] pocynaje zahepadaty v literaturnij movi ." * Ge zuinovleho k i l 1 koma prycynamy. U skolax casto sposterihajet'sja neprypustyma zaleznist 1 orfoepiji ucniv' vid' orfohrafiji. literaturna vymova zde-bil'sbho tez zazhaje vplyvu orfohrafiji. \' . . Vhaslidok zanepadu fonemy £ vynykajut' orfoepicni paralelizmy: hn>it - gn>it, hjratj£ - graty. . . . " hadaje pevnoho kolorytu i svojeridnoji riznomanit-nosti zvukovomu j fonemnomu skladu ukrajins'koji literaturnoji movy; vona zbahacuje j i j i suto zvukovi j fonemni mozlyvosti.92 In any case, the possibility cannot be ignored that the same structural forces that have maintained {g} in Ukrainian for the last several centuries despite its marginality will con-tinue to support its existence. 3.42. The introduction of {g} in early Ukrainian is best accounted for as part of a development whidh provided a l l the unpaired marked obstruents with unmarked counter-parts. In this connection i t is worth noting the example dzykhar quoted above (Sec. 3.23, n. 84) with the digraph kh for the new [g]. The earlier development, which changed [g] via [j*], to [h], has to this writer's knowledge not received a structural explanation. The "popular" explanation offered by Andre Martinet, that [g] is more difficult to articulate 9 3 than any other voiced stop, is not very enlightening. ' It fa i l s to account for the preservation of [g] in other Slavic dialects. In Slavic Languages, Roman Jakobson calls this change a "local innovation in part of the final Proto-92 Loc. c i t . 93 "Role de l a correlation dans la phonologie diaehroni-que," TCLP, VIII, 1939, p. 281, and n. 3. 92 S l a v i c period," notes its complementary distribution with, the coalescence of [e] and [c] i n the East Slavic dialect group, but offers no clarification of the connection between these developments.^ He mentions that the change of [g] to [g] (and on to [h]) i s common to the entire central Slavic area (Czech, Slovak, High Sorbian, and part of Slovenian in addition to Ukrainian, Belorussian, and South 95 Great-Russian). If the hypothesis sketched in Sec. 3.032 is accepted, the different treatment of [g] in the different Slavic lan-guages receives a natural explanation, i t would seem. In those Slavic languages which at the time of the f a l l of the jers had distinctive voice, /g/ came to be in constant auto-matic a lternation with /k/ in positions of neutralization. Phonologically, /g/ was "/k/ plus voice," and phonetically 94 Slavic Languages; A Condensed Survey, New York, King's Crown Press ("Columbia University Press), 1955, p. 14. He thus' implicitly reaffirms his view that "Tout fait dialec-tal russe ayant precede la chute des jers faibles possede sa propre isoglosse, non motivee au point de vue linguis-tique." ("Remarques sur l 1evolution phonologique du russe . . . ," p. 46.) '95 "Op. "cit i , " P>" 15. "NY S. Troubetzkoy's article "Die Ehtwicklung der Gutturale in den slavischen Spracheh," Sbornik v cest' na prof. L. Miletic,~S0fija, 1933, pp. 267-279, unfortunately could not'be consulted . According to Shevelov ("A latent phoneme in making," p. 252) Troubetzkoy considered the spirantization of [g] general " . . . where this was not precluded by the merger of c with £ (North Russian; Lower Sorbian), or by the tendency to eliminate x (Mac, Southern Polish,'parts of SCr. and Bulg.)." 93 [g] was "[k] plus voice." While there was no motivation for a change affecting /g/ in these languages, the unpaired /x/ has in some of them been curtailed or eliminated. Converse-ly, in those of the Slavic languages which had developed the distinctive feature tense vs. lax by the time of the f a l l of the jers, /g/ was the only lax i#elar consonant. As such, i t was not of necessity distinctively interrupted. In the positions of neutralization, /k/ was realized as [g], and /x/ as [ g ] , but there was no need for /g/ to be identified with one of them rather than the other. There was thus nothing in the phonological structure of these languages to prevent a change of [g] to [g] (and on to [h]). The change of [g] to [g] s t i l l need 'not be explained as due to the difficulty inherent in articulating voiced interrupted [g]. The "weakening" of voiced-lax stops to the homorganic continuants is a common phenomenon in lan-guages with distinctive tension. That only {g} changed, while {D} and. {$} remained intact is natural enough: {D} and {3} were both distinctively interrupted. It is clear that the present day geographical dis-tribution of the feature tense vs. lax does not correlate with the ar£a in which [g] changed to [g] (and on to [h]). It is necessary to assume that the tenseness feature in large parts of the Slavic language area has been replaced by distinctive voicing. The fact that in some Slavic 94 languages the development of [g] to [h] ran its f u l l course before voicing superseded the tenseness feature (as, for instance, in Gzech), while in others, voicing was introduced at the [g] stage (as in the South Great-Russian dialect group), suggests the possibility of establishing the rela-tive chronology of these developments. 5» 45» The early division of the East Slavic language area into a northern dialect group, characterized by the preservation of interrupted [g] and the coalescence of [c] and [c], and a southern dialect group, characterized by the change of [g] to [g] and the consistent differentiation of [c] and [c], was mentioned above. This writer knows of no attempts to explain why the isoglosses of these two im-portant features largely coincide. As was seen in the preceding few paragraphs, the assumption that the southern group of dialects had a dis-tinctive feature of tension can account for the change of [g] to [ g ] . It seems natural to ask, whether the same assumption can account for the preservation of the [c], [c] distinction. At the time in question, the two dialect groups had the following inventory of unpaired phonemes: North South marked v c c x unmarked c c x v The unpaired phonemes can be assumed to have been elements of instability in both systems, and were, then, ". . . ex-pose's a disparaxtre, a se creer un partenaire correlatif. . qg . ."y It is evident that the latter solution was adopted in both dialect groups for the marked phonemes, while the former was applied to the unmarked phonemes. The ten-dency to reduce the number of unpaired phonemes would 97 / / naturally affect the affricates f i r s t . Where /c/ and /c/ were unmarked, a reduction in their number could only be achieved through a sound change, and in the northern group they coalesced. This development did not spread to the southern group, for there the problem was different. From the fact that the South Great-Russian dialect group preserved the distinotionobetween /c/ and /c/, i t is clear that the tendency to fuse /c/ and /c/ had run its course by the 'time these dialects developed distinctive voicing. Where /c/ and /c/ remained marked, the possibility existed of stabilizing them by the introduction of unmarked 96 Andre Martinet, ap_. c i t . , p. 276. 97 Fundamentals, p. 42. counterparts. This possibility, of course, also existed with regard to /k/, which in Ukrainian succeded /x/ as the unpaired velar consonant, just as i t existed in North Great-Russian with regard to /v y/. This possibility was realized in both dialect groups. The Liquids In Sees. 1.3133, 1.3134, and 1.323 the different treatments of {!} and {r} followed by {j} were discussed. It was mentioned in Sees. 2.211 and 2.22 that in the varieties of CLU that distinguish weak and strong paired consonantal segments, {r} is weak whereas {£} is strong. In addition, i t was stated in Figure III that {r} is the only paired consonantal segment that does not occur in word final position (cf. rule P 8f). 4.1. The liquids present no analytical problems. It was noted in Sec. 2.32 and 2.33 that some earlier descrip-tions of Ukrainian have considered [r] and [r] positional or free variants.. This is entirely realistic with regard to a great many regional dialects of Ukrainian, but i t is characteristic of CLU that the sharped and plain interrupted liquids are distinguished before {a}, {u}, { 0 } and are in 98.Cf. Zylko, Hovory, pp. 25, 40, 44, 53, 75, 115, 134, 152, 161. 97 complementary distribution (and not free variation) else-where. Furthermore, the distribution of {r} and {%} before {a}, {u}, {o} is in CLU etymologically correct. In this respect, the literary language differs from a number of regional dialects which do maintain the opposition of {r} and {r}, but in which the historical distribution of the two morphonemes has been disturbed. The Distinctive Features The distinctive features as they relate to individual morphonemes and morphoneme classes have been discussed above. Here only two points will be mentioned, the place of the tonality features in the hierarchy of distinctive features, and the degree of economy achieved in the branching diagram (Figure I). 5.1. It is evident from the branching diagram, and from the distinctive feature matrix (Figure II), that the features flat vs. natural and sharp vs. plain are in com-plementary distribution in the CLU morphonemes. The feature flat vs. natural is distinctive only for the vowels, while the feature of sharping is distinctive only for consonants and liquids. Flatting and sharping are, as the names of these features suggest, opposite phenomena. Acoustically, f l a t -ting consists in a downward shift of some of the components 98 of the spectrum, whereas sharping means an upward shift of some of the components of the spectrum. Artieulatorily, the former feature is effected through a narrowing of one of the orifices of the mouth resonator, in CLU, the front o r i -fice, while the latter results from a widening of the hack orifice, the "pharyngeal pass."^(Fundamentals, p. 31 f.) The possibility of considering the two features manifestations of a single invariant feature, which might be termed "modified tonality," cannot be ignored. In Preliminaries (p. 33 f.) i t is suggested that " i f there is only one tonality feature in the vowels of a given language, then i t may be lumped with the primary (or only) tonality feature of the consonants. . . . " This suggestion, however, is not relevant to CLU, for here two tonality features are distinctive for the vowels, and the feature sharp vs. plain cannot be considered the primary tonality feature of the consonants. For the present description, therefore, i t was decided to forgo this reduction in the number of distinctive features. It is worth noting that i f the two features flat vs. natural and sharp vs. plain were collapsed into a feature of "modified tonality," this feature would rank differently in different parts of the hierarchy of featuresi In the vowels, i t would hold a higher place than the feature grave vs. acute, but the latter feature would precede i t in the consonants and liquids. This is the type of situation envisaged by Morris Halle in Sound Pattern, p. 35 f. The distributional variants of the feature "modified tonality" would be predictable from the feature consonantal vs. non-consonantal: "modified tonality" means flatting in non-consonantal segments, and sharping in consonantal segments. 5.2. The forty-four CLU morphonemes are fully specified by means of 277 feature statements (branches in Figure I, pluses and minuses in Figure II) , that i s , by an average of 6.3 statements per segment. This compares favourably with the theoretical lower limit of 5.46 (= logg 44) statements. That a greater measure of economy can be achieved in the specification of the CLU morphonemes QQ than in the description of the Russian segments-^  is not surprising. The relatively low average number of state-ments per segment reflects the relatively high degree of symmetry apparent in the branching diagram (Figure I). 99 Cf. Sound Pattern, p. 44 f. B. THE BOUNDARIES The five phonological boundaries of CLU, which are listed in Chapter II-A, have been postulated in conformity with Condition 1. They are introduced in the description to account for certain limitations on the sequential con-straints. In other words, their function is to delimit the domains within which the phonological rules hold. The rules which eliminate morpheme boundaries or convert them into phonological boundaries are part of the transformational level and cannot be included in the phonology of the language. Since, furthermore, none of the transformational rules of CLU have been given a definitive formulation, the remarks in the following sections will be limited to a characterization of the con-texts in which the phonological boundaries are found and a brief mention of some of their effects. Not surprisingly, the close relationship of CLU and Russian is reflected in the existence of similar phono-logical boundaries in the two languages. As will be seen below, the contexts in which the CLU boundaries are postulated are largely identical with those of the cor-responding Russian boundaries. 1 0 0 Naturally, the effects 100 Cf. Sound Pattern, pp. 48-50. 101 of the boundaries differ from one language to the other. 1. The phrase boundary (fllQ) is introduced in the following places: 1. At the beginning and end of sentences, and 2. before and after the longest immediate constituent containing not more than two and not less than one accented vowel. The phonological phrase is the domain in which the rules describing the distribution of the feature tense vs. lax hold (cf. rules P 1 and 2a). The rules describing the distribution of the feature compact vs. diffuse in sequences of acute obstruents, and of the feature sharped vs. plain in sequences of acute diffuse consonants, are obligatory only within the phonological word. They may, however, apply to the phonological phrase in allegro speech (cf. rules P 3a, 4a, 4b; and P 9a). Finally, the rules of prominence hold within the phonological phrase (cf. rules P 15 and 16). 2. Word boundaries (121) are introduced in the f o l -lowing places: 1. before and after unaccented proclitics, enclitics, conjunctions, and adverbs; 2. after the morpheme class symbol &imperative&; 3. at a l l phrase boundaries; and 4. before and after the longest immediate constituent con-taining a single accented vowel. It is interesting to note the parallelism between Russian and CLU with regard to the word boundary postulated 102 after the imperative desinence. In CLU, sequences of sharped and plain acute diffuse consonants are not admitted within word boundaries (cf. rule P 9a). Since rule P 9a clearly has no effedt on items like [ldzte] •crawl', [vyste] 'hang*, [tra^te] 'spend', (2.pl.impr.), these words must be analysed as {laz]j2|]te}, {vysl2[lle}, {tra^l2Dle} . 1 0 1 As mentioned in the preceding section, rule P 9a may be ex-tended to cover a phonological phrase in a casual style. Under such circumstances, the imperatives mentioned above may be pronounced [lazte], [vyste], [tr£t:e]. 3,. The prefix boundary (j ] 3 l ) is postulated after prefixes and unaccented prepositions. The prefix boundary is important for the description of the distribution of the feature sharped vs. plain in certain sequences of consonantal segments (cf. rules P 9a, 9c, and 9d). It also accounts for the occurrences of plain strident acute diffuse consonants before {1} (cf. rule P8c. Furthermore, the prefix boundary prevents the sharping of certain acute diffuse consonants before {j} (rule P 5b), as well as their gemination in the same con-text (rule P 6). 101 Cf. Roman Jakobson, "Russian Conjugation," Word, IV (1948), p. 159. 4^  As mentioned in Chapter III-A, Sec. 2.211, the general rule that acute diffuse consonants and continuous liquids are plain before {y} and {e} has certain "except-ions.". Apart from the forms of the verb {i j) 'pour' which are taken care of by rule P 8g, these "exceptions" are ac-counted for by the suffix boundary (fl4fl). The suffix boundary is introduced before adjectival desinences beginning with {y} and {e}. Thus, the distinc-tion between, for example, {blahor6dn[]4lle} 'noble' and {horddn|4le} 'horticultural' (nom.sg.neut.) is maintained in [blahorddne] and [horodne] (cf. rule P 8b). It can be noted that, since a l l adjective stems ending in a sharped morphoneme are accented, there is no need to introduce the suffix boundary before desinences beginning with {y} and {e*}. 5. The compound boundary (|J5fl) is introduced in compeundx. abbreviations of the type {partl5flzfl3lhory} •party meeting'. In abbreviations of this type sequences of tense and lax obstruents which are otherwise not admitted within word boundaries can occur (cf. rule P 2a). f CHAPTER TV COMMENTS ON THE MORPHEME STRUCTURE RULES The present chapter offers some comments pn the morpheme structure rules of CLU. In Sees. 1. and 2., whose numbering parallels that of Chapter II-B, the material on which the individual morpheme structure rules are based is surveyed and discussed. Sec. 3« contains some incidental observations on some problems connected with the formulation of these rules. 1.1. Rules MS 1 to 3 are based on the following attested morpheme i n i t i a l cluster types: RC {rtu-fc} 'mercury' CCJ (sv.jat} 'holiday' CJ {p.iatl 'five' CCR {zhraj} 'crowd' CR {xlib} 'bread' CCC {|MI} 'stink' CC {svit} 'world' Two longer word i n i t i a l clusters are attested. One, the CCCR of {pslr-uh} 'trout', contains a zero alternating with a vowel on the derivational level ({p#slr}, cf. {pes^r-yj} 'variegated'). The other, that of the river name Stry'Jaz (a tributary of' the Dniester rising in Poland), does not occur in an indisputably Ukrainian morpheme and can safely be discounted. It was mentioned in the discussion of the multiple 105 realizations of {j} (Chapter III-A, Sec. 1.322) that the distinctively sharped labial consonants can he analysed as morphonemic sequences of labial followed by {j}. They occur only in the context C...V, where sequences of labial f o l -lowed by [j] are not attested. While the adoption of this analysis has quite attractive consequences for the formu-lation of the morphological rules, i t alone makes i t neces-sary to take morpheme i n i t i a l sequences of the type CCJ into account. While this may seem slightly less than ideal, i t seems to be adequately compensated for by the existence of rule MS 12b. Partly as a consequence of the latter, the five-segment morpheme {syjjrfc} does not require more feature statements in the dictionary representation than does the four-segment morpheme \svat} 'match-maker'. Compare V 1 o a V o a 1} Consonantal + + 0 + + + + Vocalic - - - 0 — — — + — Compact - 0 0 + — - - + — Flat 0 0 0 — 0 0 0 — 0 Grave — 0 0 + — — + + — Strident + 0 0 0 — + 0 0 — Nasal 0 — 0 0 — 0 — 0 — Continuous + + 0 0 0 + + 0 0 Tense + — 0 0 + + — 0 + Sharped 0 0 0 0 — — 0 0 — Accented 0 0 0 0 0 0 0 0 0 After the morpheme structure rules and the morphological rules have been applied, the words {svjat} 'holiday' (gen. pi.) and {svat} 'match-maker' (nom.sg.) are by the phono-106 logical rules turned into [gydt] a n d [sv£t] (rules P 5a, 7, and 9c apply to the former, rule P 5a to the latter). 1_.2. Rules MS 5 to 7 are based on the following corpus of morpheme final cluster types: eery} 'worm' GJ 'crown of head' RC JR • sluggard' CC RR ( M i l ) 'e stuary• JCC CR {xylr} 1 sly' RCC CGR {hol^r} * sharp' CCC JC {a,1v} ' quince' RCCC mlIt} 'bridge' obijIt} 'farm yard' borsc} 'beet soup' tdvst} 'fat' cdrstv} 'stale' In a few words of patently non-Ukrainian origin, a final sequence RCR is attested, as, for instance, in the word f i l ' t r ' f i l t e r ' . Since words of this type deviate in structure from the native vocabulary ofCLU, there is l i t t l e point in taking them into account in the formulation of morpheme structure rules. A few remarks on the consequences of the inclusion or exclusion of foreign words for the morpheme structure rules are presented below (Sec; 3.1). 1.3. Rules MS 9 to 11 are based on the following attested medial cluster types: {bajrj.kj i r a v i n e ' J C {biljvll} 'buffalo' {fourjan} 'weeds' RR {barllh} 'lair' {tarjian} •cockroach' OR {izumrud} 'emerald' RCR {verhlud} »camel' CC {jacmin} 'barley' OJ {rumjln} 'ruddy' OCR {j£s$rub} 'hawk' Medial cluster types other than those listed above are attested in a few clearly non-Ukrainian items. Thus, for instance, RCC in halstux 'necktie', burstyn 'amber'; OOCC in havptvaxta 'main guard', mundstyk 'bit'; COCR in menstruacija 'menstruation', abstrahuvaty 'to abstract'. They can safely be disregarded as atypical. 2.1. Rule MS 12a describes the non-occurrence of clusters consisting of velar followed by velar or velar followed by palatal consonant. The rule is exactly analogous to Morris Halle's rule MS 11c for Russian.1 The similarity between the two rules extends even to the existence in CLU of a single counter-example which may raise doubts as to whether the rule is justified. CLU {ldks} 'noodles', unlike Russian {ve*k-S+a} 'squirrel', cannot be considered a morpheme sequence. It i s , however, of non-Ukrainian origin (Kazan Tatar, Uygurian lakca).^ It is unknown to this writer how 1 Morris Halle, The Sound Pattern of Russian^; A Lin- guistic and Acoustical Investigation, 's-Gravenhage, Mouton & Co., 1959, p. 60. 2 Cf. M. Vasmer, Russisches etymologisches Worterbuch, Heidelberg, Chr. Winter, 1953. It is^interesting to note that the Russian correspondent is lapsa. 108 long the morpheme has "been in Ukrainian (Russian lapsa is attested in Domostroj, according to Vasmer) and -uncertain to what extent i t might be considered "naturalized." For a few further remarks on {I^ 5ks} , see Sec. 3.1. 2.2. Since among the consonants only the labials can precede {j}, there is no need to specify a consonant followed by {j} as diffuse and grave in the dictionary. The relevance of rule MS 12b for morpheme i n i t i a l sequences of the type CCJ was mentioned in Sec. 1.1. In addition, this rule was referred to in connection with the discussion of the place of {v} in the phonological system (Chapter III-A, Sec. 3.122). If {v} were to be reinterpreted as a glide, *{w}, opposed to *{j} as grave to acute, the feature grave vs. acute would be predictable in the glides in position after a consonant. It would be possible to formulate two morpheme structure rules, one specifying as acute following a grave diffuse consonant, the other specifying *{w} as grave following a compact or acute diffuse consonant. Each of these rules, then, would predict one feature statement in a context defined by means of six or (in the case of a compact consonant) five feature statements, as illustrated in the following partial feature matrix, 109 Consonantal Vocalic Compact Grave - o + o where the f i r s t rule would replace the last sero with a minus. While the decision to treat {v} as a labial consonant, as mentioned in Chapter III-A, Sec. 3.122, facilitates a considerably more efficient formulation of the rules MS 1 to 10, i t clearly leads to a loss of predictability in sequences of consonant followed by {v}. On the other hand, however, i t increases the predictability in sequences of consonant followed by {j}. While the rule exemplified above predicts one feature statement in a context defined by means of six, rule MS 12b predicts two in a context defined by only four. mentioned. In the discussion of the multiple realizations of {j}, i t was shown that the.vast majority of sharped geminates should be analysed as morphonemic sequences of a consonantal segment followed by {j} (Chapter III-A, Sees. 1.3133 and 1.321). It was also mentioned (Sees. 1.3122 and 1.312, n. 23) that sharped geminates may occur within single morphemes. Rule MS 12b presupposes that no aharped geminate consonants have been analysed as monomorphemia sequences of consonant followed by {j}. One additional consequence of rule MS 12b must be 110 In the corpus examined for the present description, this is relevant to only one item, suddja 'judge'. This word has tentatively been analysed as containing a morpheme {SUCL#J}. The "mobile vowel" is not attested, but postulated in analogy with words like stattja 'article'. The latter word contains a morpheme {sta£#j\ Its gen.pl. is {stalj-iv} (parallel to {suclj-fv}) or {sta^ej}. If this analysis should turn out.to be undesirable, there s t i l l exist the possibilities of considering the stem of suddja a single morpheme with a final morphonemic. gemination., and of con-sidering i t a morpheme sequence.. 2.5. The circumstances that, le ad to the formulation of rule MS 12c were stated in Chapter III-A, Sec. 2.211 and fur ther discussed in Sec. 2.2213 of the same chapter. No additional comments seem necessary here. 2.4. The only sequences of liquids attested are those consisting of an interrupted liquid followed by a continuous liquid. In view of this., there is no need to specify the feature continuous vs. interrupted in sequences of liquids. The feature can be assigned by a rule, viz rule MS 12d. 5.1. Since there is hardly a linguistic structure which is not heterogeneous in some respects, i t is impera-I l l tive that a distinction be made between the typical and the atypical when descriptive statements are formulated. In the preceding sections i t was repeatedly noted that part of the material attested was disregarded in the formulation of the morpheme structure rules. The exclusion of patently foreign elements from the material on which the morpheme structure rules are based is a necessary consequence of Condition 5, for their inclusion would mean a marked decrease in the economy with which the native morphemes can be represented. (For example, i f the final cluster type RCR, discussed in Sec. 1.2, is included in the description, the efficiency of rule MS 5c is halved; i f the one morpheme {loiks} is to be taken into account, the considerable economy ex-pressed in rule MS 12a must be relinquished.) The fact that foreign morphemes are left out of consideration in the formulation of certain morpheme struc-ture rules naturally does not mean that they cannot be represented. It merely means that no particular economy can be achieved in their representation. Where the feature vocalic vs. non-vocalic can be left unspecified before a morpheme final sequence CR in native morphemes (cf. rule MS 5c), this feature must be specified in the morpheme One more remark should be made in this connection. The function of the morpheme structure rules is to assign 112 values to unspecified non-phonemic features. They are there-fore formulated in such a way that they never apply.to a l -ready specified features. Those non-native morphemes which have been disregarded in the formulation of the morpheme structure rules constitute the only exceptions to this gener-al rule (e.g., rule MS 5c will contradict the already speci-fied feature vocalic vs. non-vocalic in the antepenultimate segment of {filjr}). Since, however, the morpheme structure rules cannot change the values of already specified features, nothing happens. 5.2. In the formulation of the morpheme structure rules i t is naturally necessary to satisfy Condition 5. This can be done in several ways: through the ordering or the wording of the rules, or through the elimination of un-economical rules. It was suggested in Sec. 2.2 that the efficiency of a rule can be expressed simply as the ratio of the number of features the rule assigns, to the number of feature statements required to define the enntext to which the rule applies. The number of features assigned can be called the 3 This is stated unequivocally in Sound Pattern, p. 56, but appears to be contradicted in the discussion"of Russian {ajv}, p. 57 f. The "statement on p. 58 that rule MS la . . turns the second segment . . . into a vowel." must be a lapsus calami. 113 "yield" of the rule; the number of feature statements required to define the context, its "cost; and the ratio of these figures, the "relative cost." The relative cost of the rules discussed in Sec. 2 is as follows: Ms 12a: 6/1; MS 12b:'4/2; MS 12c: 6/1; MS 12d: 4/2. On the average, then, the relative cost of these rules is 3.67.^ The implications of these figures are obvious. In order that a given rule can satisfy Condition 5, there must be a reasonable relationship between its relative cost and the number of morphemes to which i t applies. It seems worth while CD nsidering a specific example. In CLU, clusters of labial consonants are not attested in i n i t i a l position. While an i n i t i a l labial consonant can be followed by an acute diffuse or an acute compact conson-ant (e.g. {vdlv} 'widow1, {^5l} 'bee'), a following grave consonant is invariably compact (e.g. {ohdj}'bend'). Furthermore, in morpheme medial and final position, the only clusters of labial consonants attested are those that begin with {m} or {v} (e.g. {klumb}'threshing floor', {scovb}'rock'). 4 The specifications " i n i t i a l " and "non-initial", which are necessary in the definitions of certain contexts have been assigned the value 1. 114 These distributional facts invite the formulation of a rule like the following: in position after a grave diffuse consonant other than non-initial {v} or {m}, a 5 grave consonant is compact. Such a rule, however, while clearly giving an accurate account of s>me of the d i s t r i -butional characteristics of labial consonants, would defeat its own purpose, that of introducing a measure of economy in the description. At a relative cost of 15/1, this rule would reduce the number of feature statements by one in only a handful of morphemes. This writer would like to propose an amendment to Morris Halle's statement that ". . . a feature must remain unspecified in the phonological representation whenever the feature is nonphonemic by virtue of its occurrence in a g particular context." In view of Condition 5 , i t would seem that the function of the morpheme structure rules is not to describe the sequential constraints, but to predict sequences that are admitted. Thus, a feature must be left unspecified in the representation of morphemes only where i t is more economically specified by a rule. In order that 5 Cf. the similar rule for Russian, Sound Pattern, p. 60: "MS l i b . In position after a grave noneompact ("labial) consonant other than non-initial {*m}. a [grave H»-A»] consonant is compact." 6 Sound Pattern, p. 30. a morpheme structure rule can produce a saving in the representation of morphemes, its relative cost must he lower than the number of morphemes to which i t applies. The relative cost of the rules discussed in Sec. 1, MS 1 to 10, is somewhat lower than that of rules MS 12a to 12d: MS l a : 2/2 MS 5a: 6/2 MS 7: 6/14 MS lb: 6/2 MS 5b: 3/2 MS 9a: 3/2 MS l c : 3/2 MS 5c: 10/2 MS 9b: 3/4 MS 2: 6/2 MS 5d: 6/2 MS 9c: 3/4 MS 3: 6/8 MS 6: 6/2 MS 10: 6/12 The overall average relative cost of these rules is 1.21. The greater efficiency of these rules is naturally conditioned by the fact that they involve only two distinc-tive features. In addition, however, the fact that the rules are partially ordered contributes considerably to their economy. Thus, after the relatively costly rules MS l a to 2 have been formulated, a more general rule, MS 3, can be formulated which reduces the average relative cost of the whole set (cf. the similar relationship between MS 5a to 6 and MS 7; and between MS 9a to 9c and MS 10). 7 f\ 7 Morris Halle provides a good example of the importance of ordering the rules properly in Sound Pattern, p. 58 f. Rules MS l a to 3> which are maximally economical, have an overall relative cost of 26/19 (= 1.37); rules MS 9a to 116 Considerations of the relation between relative cost and actual number of morphemes concerned are relevant to rules;.MS la to 10 as well as to rules MS 12a to 12d. In particular, i t might be asked whether rules MS 9a to 10, which are applicable only to morphemes with three or more vowels, increase the number of unspecified features in sufficiently many morphemes to justify their relative cost. In view of the low overall relative cost of these four rules, (15/22 = .68), this question can undoubtedly be answered positively. CHAPTER V COMMENTS ON THE PHONOLOGICAL RULES The present chapter offers some comments on the phonological rules. Its primary purpose is to illustrate the operation of these rules. In this connection, some remarks will he made on the order in which the rules must he applied. In addition, a number of rules which are alter-native to rules listed in Chapter II-C, or which are not mandatory, will be discussed. The f i r s t two sections have been numbered to parallel the sections of Chapter II-C. The alternative and optional rules are given in Sec. 3 below. 1 . 1 . Rules P 1 and 2 a describe the automatic d i s t r i -bution of the feature tense vs. lax in obstruent clusters. Examples of rule P 1 . {dxnuv} [txnuw] 'took (masc.) a breath' (cf. {vy|]3l]dxnuv} [v^dxnuw] 'exhaled' (masc.)); { v y ^ Q j s o v 5 2 f l z | 3 E x a t y } [vyjsowsxaty] 'came (masc.) out of the house', but fern. {vyj]3fl j sla[l2llz!3flxaty} [vyj slazxaty] . It should be noted that the exception to the general statement in rule P 1 cannot be formulated rigorously. There appears to be a great deal of variation with regard 118 to the preposition and prefix {z}}' Examples of rule P 2a. {prdsba} [prdzba] 'request' (cf. {prdsa£} [prdsa$] 'they ask'); {xripfl2[]ze} [x:rib£e] •hut he snored* (cf. {xropla} [xropla] 'she snored'). But across a compound boundary there is no neutralization: {part[J5flSyscyplina} [partdyscyplina] 'party discipline'. The persistence of the distinction between tense and lax obstruents outside the positions defined in rules P 1 and 2a is abundantly exemplified in the following sections. In addition, compare: {lis} [lis] 'forest' and {liz} [liz] 'crawled' (masc); {riska} [riska] 'crumb' and {rizka} [rizka] 'stick'; {tvir} [tyir] 'creation' and {dvir} [dyir] 'door'. 1.2. Rules P 3a, 4a, and 4b describe the distribu-tion of the feature compact vs. diffuse in clusters of acute obstruents. Rules P 2b, 2c, and 3b are prerequisite for the proper operation of the above rules and require a few words of comment. Rule P 2b assigns the non-phonemic feature inter-1 Cf.., for instance, M. A. Zovtobrjux, Sucasna ukrajin-s'ka literaturna mova, Kyjiv, Radjans'ka Skola, -1961, p. ToTZ 119 rupted to the dental stops. This must he done before rule P 3a is applied since this rule turns some dental stops into palatal affricates, which are distinctively interrupted. Rule P 2 c , which is parallel to the rules discussed in Sec. 1 .3, must be placed before rule P 3a since the l a t -ter may apply to the palatal consonants resulting from the application of rule P 2 c . In the material analysed for the present description, there are no examples of dental ob-struents preceding velars affected by rule P 2 c . It i s , however, not unlikely that a larger corpus would contain such examples. Rule P 3b assigns the non-phonemic feature of stridency to the palatal consonants in order to ensure that a l l acute diffuse obstruents resulting from the application of rule P 4a are strident. Examples of rule P 3 a . Dental stops before palatal consonants: {br^tcyk} [brac:yk| ' l i t t l e brother', {tcu|} [c:u|] 'they weave'; {nadfl3!zaly} [na^z^ly] 'began reap-ing' (pi.); {vid[j3l]cepyv} [ y i 3 c e p y w ] 'detached' (masc). Strident dentals before palatal consonants: {piscanyj} [rciscanyj] 'sandy'; {z f l3isyla} 'sewed' (fem.) (cf. rule P I ) ; {bezfl3flsumu} [bezsumu] 'noiselessly'; {z]]3Querela} [z^erela] 'from the source'; {xlopec]]2{Ize} [ x l o p e 5 z e ] 120 '"but the boy' (cf. rules P 2a, 8e, and 10a). Examples of rule P 4a. {poQ3[lkazsa} [pokazsa] 'show yourself; {u[]3iplasci} [uplasci] 'in the bottle' (cf. {plaska} [pl&ska] (nom.sg.) and rules P 8c and 9a; {docci| [doc:i] 'daughter' (dat.sg.) (cf. {docka} [docka] (nom.sg.) and rules P 8c, 9a, and 11a). Examples of rule P 4b. {u[J3[]budci} [ubu^gi] 'in the booth' (cf. {budka} [budka] (nom.sg.) and rules P 8c and 9a); {pecatci} [pecacti] 'stamp' (dat.sg.) (cf. rules P 8a, 9a, and 11a). As was mentioned in Chapter TV (Sea. 1), these rules are obligatory within the phonological word, but may, in a casual style, apply also across word boundaries within a 2 phrase provided no pause intervenes. 1.3. Rules P 5a to 5d and 8a to 9e describe the automatic distribution of the feature sharped vs. plain in consona ntal segments. The division of these rules into two groups is necessitated by the rules that generate the multiple realizations of {j}. Eor example, rule P 9c presupposes that - 2 Cf. Oleksa Synjavs'kyj, Wormy ukrajins'koji literatur-noji movy, 2nd ed., L'viv, Ukrajins'ke Vydavnyctvo, 1941, p. 174. 121 the feature sharped vs. plain has been assigned to a l l labial consonants. Rule P 8a, however, must be limited to specifying these consonants as plain before consonants and liquids, and in word final position. It could not be applied to labials in other positions without interfering with the reflexes of sequences of labial followed by {j}. Thus, the distribution of the feature sharped vs. plain in labial consonants followed by a vowel or {j} must be specified by a rule preceding rule P 7, viz rule P 5a. Examples of rules 5a to 5d which involve {j} are given below in connection with rules P 6 and 7. Examples of rule P 5a. Labial consonants before a vowel: {bfk} [bik] 'side* (nom.sg.) and gen.sg. {b6ku} [boku]; {byk} [byk] 'bull'; {buk} [buk] 'beech1. Examples of rule P 5c. Palatal consonants before a 9a), but {ses^ero} [se'stero] 'a group of six' (cf. rules P 8b and 9a); {kryci$} [krygl^l ' cry 1 (2.pi.impr.), but {kryc^} [krycy$] '(one) cries' (cf. rule P 8f'). 'clang' (cf. rule P 9c), but {zl^flvjdz} [ z vJaz] 'connection* (cf. {vjaz} [vjaz] 'clasp'); {mertvjdk} [mertydk] Examples of rul s P 5a and 7. {3vjdzk} [^yazk] 122 •corpse', but {cervjak} [cervjak] 'worm'. Examples of rules P 5b, 6, and 7. {pid[)3.[Jboridja} [|>idborid:a] 'chin1 (cf. rules P 5 a , 5 d , and 1 1 a ) , but gen. pi. {pid|3[Ibor£d j} , to which rule P 6 does not apply, becomes [;oidborid]; similarly {proQ3|]valja} [proval:a] 'precipice', but gen.pl. {pro|3lvalj}, [prova].]. Rules P 5b and 6 have no effect across the prefix boundary. Thus, for instance, {z ; l 3 l j i d a ^ } 'they will eat' becomes [zjida$] rather than the non-standard [z:ida$]. Examples of rules P 5 c , 6 , and 7. {roz | 3 i d o r i h j a } [rozdorig:a] 'fork of a road' (cf. rules P 2 c , 5 d , and 1 1 a ) ; gen.pl. {roz[]3Ddorihj} , to which rule P 6 does not apply, becomes [rozdoriz] (cf. rule P 8 e ) . Examples of rules P 5d and 7 . {poj}3|vi|rja} [poyitra] 'air' (cf. rules P 5a and 9 b ) , but {bez|3|jvir ja} [bezyirja] 'atheism' (cf. {vira} [yira] 'faith') and {su|l3[jzirja} [suzirja] 'constellation' (cf. {zora} [zora] •star', and rule P 8 c ) . Rule P 8 a complements rule P 5a by specifying labial consonants as plain in a l l positions not mentioned in the latter rule. Examples of rule P 8 b . Mellow acute diffuse conson-ants: {kdnyk} [kdnyk] ' l i t t l e horse', {kdnej} [konej]'horses' 123 (gen.pl.) (cf. {kin} [kin] 'horse'); across the suffix boundary there is no neutralization: cf. {$ertl]4[lyj} [te'rtyj], {Se'rtSAfle} [te'rte] 'ground' and {£re'$ Hflyj} [tre$yj], {?re||]4le} [tre$e] 'third' (nom.sg.masc. and neut.) , Strident acute diffuse consonants: {hugy} [husy] 'geese* (nom.pl.), {hugej} [husej] (gen.pl.) (cf." {hugam} [hugam] (dat.pl.)). Across the suffix boundary, interrupted strident dentals are affected, cf. {kuc[j4|]yj} [kticyj] 'short' (nom.sg.masc.) (fem.,' {kuca} [kuca]) and {biloiyc]l4|]yj} [biloly'cyj] 'white-faced' (nom. sg.masc.) (fem.,{biloiyga} [bilolyga]); but the opposition sharped vs. plain is maintained in the continuous strident dentals: {lysl4lyj} [lysyj] 'bald' (fern, {iysa} [lysa]), but {hus|4Dyj} [hugyj] 'pertaining to a goose' (fern, {huga} [huga]). Examples of rule P 8c. {sik} [sik] 'chopped* (masc.), {sik} ,;.[-gik']i 'juice' (cf. gen.sg. {sdku} [soku]), {gil} [gil] 'villages' (gen.pl.) (cf. nom.pl. {gela} [sela])-; but across the prefix boundary, rule P 8c has no effect, cf.. {zi|3ihnanfi4|yj} [zihnanyj] 'rounded up' and {zfl3l 3 Cf. Synjavs'kyj, |>. 60. 124 i g n o r 6 v a n f l 4 D y j } [zignorovanyj] •ignored'. Examples of rule P 8d. {kin} [kin] 'horse' (cf. gen.sg. {kona} [kona]), hut {kyn} [kyn] 'throw• (sg.impr.). Rule P 8d complements rule P 2c by specifying a l l velar consonants not affected by the latter rule. Examples of rule P 8e. Rule P 8e specifies the so far unspecified palatal consonants as plain before conson-ants (e.g. {^mil} [^mil] 'drone'), before liquids (e.g. {clen} [clen] 'member'), and in word final position (e.g. {pic} [pic] 'oven'). The rule changes the sharped palatals in word final position which result from the application of rules P 5c and 7. Thus, for instance, {rozfljjfldorihg} 'fork of a road' (gen.pl.), which after P 2c is {roz|3Edorizj}, after 5a and 5d, {rozjJ3Bdori|3}, after 7, {rozfl3fldori|}, is by 8e turned into {roz|]3fldori§} (ultimately [rozdorfz]. Similarly, {pid|3Qzamkj} 'vicinity of a castle' (gen.pl), which after P 7 is {pi3H3Ezam£} , after P 8e is {pi3J]3|zamc} (ultimately [pi3zamc]). Examples of rule P 8f. {hora} [hora] 'mountain' {hori} [hori] (loc.sg.), {hory} [hory] (gen.sg.), {hir} [hir] (gen.pl.), but {zora} [zora] 'star', {zori} [zori] (loc.sg.), {zore*;ju} [zoregu] (in.sg.), {zir} [zir] (gen. pi.), {zoryca} [zoryca] 'star'. 125 Examples of rule P 8g. {zemla} [zemla] 'earth', {zemleju} [zemleju] ( i n . s g . ) , {zemlyst|]4|yD} [zemlystyj] -'earthy 1; across a s u f f i x boundary: c f . {uH3lImerl|]4[ly;j} [umerlyj] 'deceased' and { d r l j ] 4 l l y o } [ o r l y j ] 'pertaining to an eagle'; but {iju} [ l l u ] 'pour' (l.sg.) ( c f . .{iyj} [ l y j ] (impr.) and r u l e s P 5 b , 6, and 7 ) . {ije} t i l e ] (3»sg.), and {zafl3Dije} [ z a l l e ] 'inundates'. Examples of r u l e P 9a. {kxS"{;} [kig$] 'bone', and gen.pl. {kos£ej} [ k o s t e j ] ; {|xtka} [£itka] 'aunt', {$xtci} [£xg:i] (dat.sg.) (of. r u l e s P 4 b , 8c, and 11a); {^ryvdzga} [tryvdzsa] 'be frightened' (impr.); {pxsna} [gxsna] 'song' ( c f . gen.pl. {pise^i} [oxse:a]), and i n . s g . {pi^neju} [pxsneju] ( c f . r u l e P 8 b ) ; {sdnce} [sdnce] 'sun' ( c f . gen.pl. {sdneg} [sdneg]) and gen.sg. {sdnga} [sdnca]. In a deliberate s t y l e , rule P 9a does not hold across a p r e f i x boundary: {s$xny} [s^xny] 'walls', but {zJ^FlJi^gu} [s$xn:u] 'with shade' ( c f . r u l e s P 1, 5 b , 6, 7 , and 11a); or across a word boundary: {banty} [banty] ' t a s s e l s ' , but {sta^|4|?e} [stante] 'stand' (2.pi.impr.). But a casual s t y l e may ex-tend the domain of the rule to the phonological phrase, n e u t r a l i z i n g the oppositions: [§s|xny], [s$xn:u]; [banty], [stante;.]. Note that t h i s rule must follow r u l e s P 8 b and 8c ( c f . the forms of {txt#k} and {son#c) above). 126 Examples of rule P 9b. {kozla} [kozla] 'billy-goat' (gen.sg.) (cf. nom.sg. {kozel} [kozel]), like {vuzla} [vuzla] 'knot' (gen.sg.) (cf. nom.sg. {vuzol} [vuzol]); {vi:£ru} [yitru] 'wind' (gen.sg.), {vi£r£v} [yitriw] (gen. pi.) (cf. nom.sg. {vi$er} [yiter]). Examples of rule P 9c. Sharped and plain dental consonants are opposed before plain labials: {pysmo} [pysmo] 'letter' (cf. gen.pl. {pysem} [pysem]), and {pasmo} [pasmo] 'skein' (cf. gen.pl. {pasom} [pasom]); but the op-position is neutralized before sharped labials: cf. loc.sg. {pysmi} [pysmi] and {pasmi} [pasmi]. Cf. further {zvir} [zyir] 'animals' (collective) (gen.sg. {zviru} [zyiru]) and {zvir} [zyir] 'ravine' (gen.sg. {zvoru} [zvoru]). Across a prefix boundary, rule P 9c does not apply: {spily} [spily] 'ripened* (pi.), but {zfl3l]pil} [spil] 'from the fields'. For some speakers, rule P 9c might be extended to include the continuant grave compact consonants in the same position: cf. {xmil} [xmil] 'hops', gen.sg. {xmelu} [xmelu]; the pro-nunciation [xmil], however, may be as common. Examples of rule P 9d. {kozli} [kozli] 'billy-goat' (loc.sg.) (cf. nom.sg. {kozel} [kozel]) and {vuzli} [vuzli] 'knot' (loc.sg.) (cf. nom.sg. {vuzol} [vuzol]). Rule P 9d has no effect across a prefix boundary in a careful style: 127 c f . {zQ3lldzi} [ z l o g i ] 'from the loge' and {sldzy} [sldzy] 'tears' (pi.) . Examples of rule P 9 e . {hdlci} [hdl^-] 'needle' (dat.sg.) ( c f . nom.sg. {hdlka} [hdlka] and gen.pl. {holdk} [holdk]; {vdlzi} [ v d l z i ] 'the Volga' (dat.sg.) ( c f . nom.sg. {vdlha} [vdlha]); contrast {halka} [halka] 'jackdaw', l o c . sg. {halci} [ h a l c i ] and {halka} [halka] (proper name), loc.sg. {halci} [ h a l c i ] . 1.4. Examples of rule P 9 f . {smije^sa} '(one) laughs' and {smiju^sa} 'they laugh', which a f t e r r u les 4b, 5a, and 9c are {smijegsa} and {smijugsa}, become {smijegga} and {smijugga}, u l t i m a t e l y [smijeg:a], [ smijug:a]. The rule may be ^ e f f e c t i v e across a word boundary where there i s no pause: {tra£|2[]sa} [trag:aj 'be spent' (sg.impr.), {nef l2 J Imucjjsa} [nemug:a] 'do not t i r e y o u r s e l f ( c f . rul e P 4a). Rule P 9g assigns the non-phonemic features sharped and acute to {j } . {j} must be spe c i f i e d as sharped f o r the proper operation of the rules that assign allophonic features to the vowels (P 1 1 c to 14), and i t must become acute before the glide allpphones of {v} are generated by rule P 1 0 c . 128 Examples of rule P 10a. {jakis-{;ri.j|Aflyo} [ j a k i s n y j ] ' q u a l i t a t i v e ' ( c f . {jakis$} [jakis$] ' q u a l i t y ' ) : { P6l3ijizdn[l4llyD} [pdjiznyj] 'of a t r a i n ' ( c f . {poQ3fljizd} [pdjizd] ' t r a i n ' ) . I t can be noted that the r u l e does not cover the a l t e r n a t i o n i n {^yzclen} [tyzden] 'week', gen.sg. {lyzna} [tyzna], the only example of the elimination of a dental stop a f t e r a p a l a t a l continuant. The a l t e r n a t i o n here i s not automatic, c f . [tyzenno] 'weekly' (adv.), i . e . {lyzen-n-o). I t might be mentioned also that r u l e P 10a can re f e r simply to "interrupted acute consonants" since there are no sequences of strident dental followed by an acute compact consonant ( c f . rule P 3a). Examples of rule^P 10b. This rule applies to the products of rule P 6 which fol l o w a'consonant or l i q u i d : {pidH3lgruntja} 'background', which a f t e r P 6, 7, and 9a i s {pidfl3flgruntta} , becomes [jpidgrun^a]; {smertju} 'death' ( i n . s g . ) , which a f t e r P 6, 7, and 8f i s {smer^u}, becomes {smer^u}. There are no examples of geminate l i q u i d s a f f e c t -ed by t h i s rule i n the present l i m i t e d corpus. Rule P 10b generates the d i s t i n c t i v e l y sharped p a l a t a l s discussed i n Chapter III-A, Sec. 1.321: {pidj^flzamkja}, which a f t e r r u l e s P 2c, 5a, 5c, 6, and 7 i s {pid^lzamgga} becomes [pi-jzamga]. Rule P 6 could e a s i l y be formulated i n such a way that i t 129 would not generate geminates in position after a consonantal segment. Such, a reformulation, however, would complicate rule P 6 without leading to the elimination of rule P 10b, which is needed for sequences like {xustci} 'kerchief (loc. sg.) (cf. nom.sg. {xustka}), where there are no geminates t i l l after rule P 9a (i.e. {xustci}, after P 4b: {xuscci}, after P 8c: {xuscgi}, after P 9a: {xusgci}, ultimately [xusci]). Postvocalic {v}, when not followed by a vowel, is realized as a glide, [w] (cf. Chapter III-A, Sec. 3.122). The essential difference between [v] and [w] is that the latter is articulated without the labial obstruction that characterizes the former. Acoustically, then, [w] is non-consonantal.^ The change in the feature composition of postvocalic {v} not followed by a vowel is effected by rule P 10c. It can be noted that [w] is opposed to {j}, which has already been specified as acute by rule P 9g. Examples: {krdvy} [krovy] 'blood' (gen.sg.), but nomc.sg. {krdv} [krow], in.sg. {kroVju} [krdwju], diminutive {krivla*} [kriwld]. Rule P 10c must precede rule P 11a to ensure the 4 Cf. Roman Jakobson and Morris Halle, Fundamentals of  Language (Vol. I of Janua Linguarum, ed. C. H. van Schoone-veld), 's-Gravenhage, Mouton & Co., 1956, p. 26; and Gunnar Fant, Acoustic Theory of Speech Production, 's-Gravenhage, Mouton & Co., I960, p."" 170. 130 correct realization of, for instance, the {w} of {bowan} [bowva^i] 'image'. As was mentioned in the discussion of geminates in Chapter III-A, Sees. 1.311 and 1 . 3 1 2 , the CLU geminates are, from an articulatory point of view, sequences of identical segments separated by an intensity valley which reflects the interval between two chest pulses. If only the articulatory mechanism.of the oral cavity is considered, the geminates appear to be lengthened. It is this oversimplified analysis which is reflected in the traditional transcription of gem-inates (where geminate [d] is written [d:]). This transcrip-tion can be retained provided i t is understood to render the articulatory facts in a simplified manner. Examples of rule P 1 1 a . {ob ] ]3Dbyty} [obt^ty] 'to wallpaper'; { z n a j 6 m - m o l 2 ij 14-lyx} [znajom:ojyx] 'let us get them acquainted'; {pid[J3lddx.ja} [nid:^|:a] 'loft'; { z l 3 | zddu} [z:acLu] 'from behind'; {zfl3jz£ty} [z:£ty] 'to mow down'; {masyn-n[}4fly;)} [masyniyj] 'machine' (adj.); {vyl3D ssaty} [vystaty]••to suck out 1; {midju} [m£d:u] 'copper' (in.sg.); {su-fcDii} [su\t:u] 'gist' (in.sg.); {-fcinju} [££n:u] 'shade' (in.sg.); {micju} [m£c:u] 'strength' (in.sg.); {ma^ zju} [mdz:u] 'ointment' (in.sg.); {nis-sa} [n£s:a] 'was carried* (masc.); {mezy^flrikja} [mezyriQ:a] 'region between rivers'; {pid[J3Dnihja} [nidn'f|:a] 'pedestal'; {pidjdaxja} 131 [|)id:d|:a] 'loft'; {silju} [§il:u] 'salt' (in.sg.). It should be noted that rule P 11a does not affect sequences of {j} or postvocalic {vv}. Thus, {vij-j-d} 'cart shaft' (cf. {vij-g-e} [yijce] 'plough shaft') is realized with distinct allophones of {3} for the falling diphthong [ij] and the rising diphthong [ja], [vijjd]; similarly, postvocalic {w} becomes [wv] as a result of rule P 10c, e.g. {za{|3lvl3flvdha} [zawvaha] 'note1. 1.5. Rules l i b to 14 assign allophonic features to the vowels. Rule P l i b specifies {u} and {o} as fundament-ally back vowels and is prerequisite for rule P 12, which describes the fronting of a l l grave vowels in position after a sharped segment. Rules P 12 to 14 can be suspended in a maximum redundancy style, as in the examples contained in the preceding sections, but in a "natural" style they are opera-tive. It should be noted that the considerable vacillation in the realization of vowels in checked positions depending on variations in'style,, tempo, and individual speech habits makes i t undesirable to describe the data more precisely than has been done in rules P 12 to 14. Examples of rule P 12. {sad} [sad] 'garden', but {sddu} [sldu] 'I shall sit down'; {lox} [lox] 'river salmon' but {ldx} [lox] 'dungeon'; {luk} [luk] 'arch', but {luk} [luk] 'hatchway' (gen.pl.); {lysQ4ly.i} [lysy.i] 'bald', but 132 {husQ4l!yj} [nusyj] 'pertaining to a goose'; {tvaryna} [tvaryna] 'creature', but {ufl3lkra tjyna} [ukra,-jyna] 'the Ukraine'. Examples of rule P 1 3 . {vad} [vad] 'defect' (gen. p i . ) , but {v£d} [va^d] 'hinder' (sg.impr.); {t<5n} [ton] 'tone', but {don} [doiii] ' l i t t l e daughters* (gen.pl.); {syn} [syn] 'son', but {syn} [ s y % ] 'blueness'; {nud} [nud] 'bore-dom' , but {nud} [nuid] 'nausea'. The combined e f f e c t s of rules P 12 and 13 i s a r e l a t i v e l y strong f r o n t i n g , p a r t i c u l a r l y marked i n un-accented vowels, i n rapid speech also i n accented vowels. Por instance, {ga$} [gad] ' s i t down' (-sg,.impr.); {losi} [lb'si] 'sow' (dat.sg.) ( c f . {^oxa} [ l 6 x a ] (nom.sg.)). No r u l e s have been formulated to describe the d i f f e r e n t degrees of diffuseness i n accented and unaccented {y} and {u.} since t h i s v a r i a t i o n possibly i s not general ( c f . Chapter III-A, Sec. 2 . 1 1 ) . The opposite phenomenon, however, i s described i n rule P 14: unaccented {e} and {o} tend to become l e s s compact than t h e i r accented counter-parts. Examples of rule P 14. {©}: {met} [met] 'goal' (gen. p i . ) , but dat.sg. {me-fl} [myv-£i], gen.sg. {mety} [my-ty], acc.sg. {metu} [my vtu]; nom.sg. {metd} [met£] or [me*td], in.sg. {metoju} [metoju] or[me Atoju]. {o}: {kozd.} [kozd] 155 'goat' .(nom.sg.), but {kozti} [koAzu] (acc.sg.), and {kozi} [ko Azi] (dat.sg.). 2. As mentioned in Chapter 1-C, a study of the dis-tribution of relative prominence in unaccented vowels could not be undertaken in the course of the present investiga-tion. For this reason, the rules describing prominence have been limited to the bare essentials. 5.0. It was stated in the Introduction (Chapter I-A) that the differences between the several varieties of CLU essentially are differences in the distribution of phonemes, which are reflected in the existence of supplementary phono-logical rules. These rules are listed, and their place in the phonological system discussed, in the following . 5.11. It was mentioned in Chapter III-A, Sec. 2.2222 that the present orthoepic norms of Soviet Ukrainian accept the elimination of the opposition / i / : / y / in word i n i t i a l position. The following optional rule can be formulated: in position after a word or prefix boundary, {y} becomes acute. The place of this rule cannot be determined from the material examined for the present description. But since a prefix final labial consonant presumably will be sharped before an [i] resulting from this rule, the rule can tenta-tively be placed before rule P 5a and numbered PO 4c. 134 5.12. In some varieties of CLU, the / i / : / y / opposi-tion is eliminated after / j / (cf. Chapter III-A, Sec. 2.2215). -For these varieties of CLU, the following optional rule can he formulated: In position after {j}, {y} becomes acute. It is important that this rule be applied before rule P 12, and i t can therefore be numbered PO 11c . 5.15» As mentioned in the discussion of sequences of consonantal segment followed by vowel (Chapter III-A, Sec. 2 . 2 2 1 2 ) , the division of the paired consonantal morphonemes into strong and weak is not relevant for a l l varieties of CLU: in some varieties of CLU, the opposition of sharped and plain paired consonantal segments before / i / has been elimi-nated. In order that the present description can accomodate these varieties of the language, the following optional rule can be formulated: Mellow acute diffuse consonants and continuous liquids are sharped before {1}. To ensure the proper operation of the rules describing the automatic distribution of the feature sharped vs. plain in sequences of consonantal segments (rules P 9a, 9b, 9d, and 9e), this rule must be numbered PO 8h. 5.2. The fact that several varieties of a given language can be described as deviations from a given norm by 135 the inclusion of optional rules in the phonology of the language is one of the most interesting implications of the theory of phonology chosen as the framework for the present description. Morris Halle has recently suggested that linguistic change can be described as the introduction of new rules into the phonological system. While in the innovating generation a change would be reflected in an optional rule, a later generation would integrate the change in the obligatory rules to achieve a maximally economical 5 phonological system. This suggestion seems to be supported by the situation in CLU. Thus, for instance, i t seems certain that rule PO 8h, which must be an optional rule in a description that accomo-dates several varieties of the language, is an optional rule for many.speakers of CLU, who apply this rule in a colloquial style, but suspend i t in formal speech. It seems equally certain that for many speakers of CLU this rule has been completely integrated into the phonology. These speakers have in their speech pattern an alternative rule which in-corporates both of rules P 8c and PO 8h and may be numbered PA 8c: 5 Morris Halle, "^ Phonology in a Generative Grammar," Word, XVIII (1962), pp. 54-72. 136 Acute diffuse consonants and continuous liquids are1 sharped before {1}. On the other hand, i t would seem that more than the phonological rules can be involved. Thus, i t is clear that in the varieties of CLU where rule PO l i e has become rule P 11c, rule MS 12c has become redundant and, most probably, deleted. Again, in the varieties of CLU where rule PO 4c is obligatory, i t can be assumed that the opposition of mor-pheme i n i t i a l / i / : / y / has been eliminated in the dictionary representation of lexical morphemes and is preserved only in grammatical morphemes. BIBLIOGRAPHY Andrusyshen, C. H. and J. N. Krett. Ukrainian-English Dictionary. Toronto, University of Toronto Press, 1957 [copyright: University of Saskachewan, 1955]. Avanesov, P. I. Ocerki russkoj dialektologii. Cast' I. Moskva, Gosudarstvennoe Ucebno-Pedagoglceskoe Izda-tel'stvo Ministerstva Prosvescenija RSFSR, 1949. Bloch, Bernard and George L. Trager. Outline of Linguistic  Analysis. Special Publications of the Linguistic Society of America. Ed. Bernard Bloch et a l l . Baltimore, Md., Waverly Press, 1942. Broch, 01af. Slavische Phonetik, Sammlung slavischer Lehr- -und Handbucher. Herausg. A. Leskien und E. Ber-neker. Heidelberg, Carl Winter, 1911. Bulaxovs'kyj, L. A. Pytannja poxodzennja ukrajins'koji . movy. Kyjiv, Yydavnyctvo Akademiji Nauk URSR, 1956. Cherry, E. Colin, Morris Halle, and Roman Jakobson. "Toward the Logical Description of Languages in their Phonemic Aspect." Language, XXIX (1953), 34-46. Chomsky, Noam. Syntactic. Structures. Yol. IV of Janua  Linguarum. Ed. Cornells H. van Sehooneveld. 1s-Gravenhage, Mouton & Co., 1957. 138 Dejna, Karol. "Fonologiczny system jezyka ukrairiskiego." Biuletyn Polskiego Towarzystwa Jezykoznawstwa, X (1950), 149-153. Durnovo, Nikolaj. Ocerk i s t o r i i russkogo jazyka. Vol. XXII of Slavistic Printings and Reprintings. Ed. C. H. van Schooneveld. 's-Gravenhage, Mouton & Co., 1959. Pant, Gunnar. Acoustic Theory of Speech Production. Vol. II of Description and Analysis of Contemporary Standard  Russian. Eds. Roman Jakobson and C. H. van Sehoomrveld. •s-Gravenhage, Mouton & Co., I960. Grigor'ev, V. I. Review of M. Halle, The Sound Pattern of  Russian. Voprosy Jazykoznanija, X:l (1961), 129-133. ________ "Differencial'nye priznaki russkix glasnyx /u, y, i / . " Voprosy Jazykoznanija, XI:1 (1962), 10-30. Halle, Morris. The Sound Pattern of Russian: A Linguistic and Acoustical Investigation. Vol. I of Description  Analysis of Contemporary Standard Russian. Eds. Roman Jakobson and C. H. van Schooneveld. 's-Gravenhage, Mouton & Co., 1959. "Phonology in a Generative Grammar." Word, XVIII (1962), 54-72. Holoskevyc, H. Pravopysnyj slovnyk. 8-e vyd. New York, Knyhospilka, 1955. 139 Horalek, Karel. tfvod do studia slovanskych jazyku. Praha, Nakladatelstvi* Ceskoslovenske Akademie Ved, 1955. International Phonetic Association. The Principles of the  International Phonetic Association. London, Dept. of Phonetics, University College, 1949 (Repr. 1958). Isacenko, A. V. "Versuch einer Typologie der slavischen Sprachen." Linguistica Slovaca, I (1939), 64-76. Jakobson, Roman. "Remarques sur l 1evolution phonologique du russe, comparee a celle des autres langues slaves." TCLP, II (1929). , "Russian Conjugation." Word, IV, (1948), 155-167. , C. Gunnar M. Pant, and Morris Halle. Preliminaries to Speech Analysis: M.I^ T.- Acoustics Laboratory Tech- . nical Report No. 13. 2nd Printing. Cambridge (Mass.), Massachusetts Institute of Technology, 1952. , Slavic Languages: A Condensed Survey. 2nd ed. In Columbia Slavic Studies. Ed. Ernest J. Simmons. Hew York, King's Crown Press (Columbia University), 1955. , and Morris Halle. Fundamentals of Language. Vol. I of Janua Linguarum. Ed. Cornells H. van Schooneveld. 's-Gravenhage, Mouton & Co., 1956. , "Principes de phonologie historique." In 140 N. S. Troubetzkoy, Principes de phonologie, trans. J. Cantineau, Paris, 0. Klincksieck, 1957, pp. 315-336. Jassem, Wiktor. "The Phonology of Polish Stress." Word, XV (1959), 252-269. Kofinek, J. M. Od indoeuroskeho prajazyka k praslovancine. Bratislava, Slovenska Akaddmia Vied a Umeni, 1948. Kovaliv, P. K. "The Problem of Change of 0, E in new closed Syllables in Ukrainian Dialects." Orbis, V (1956), 244-249. Kuraszkiewicz, Wiadyslaw. Zarys dialektologii wschodnio-s^owianskie."] z wyborem tekstdw gwarowych. Warszawa, Panstwowe Wydawnictwo Naukowe, 1954. lees, Robert B. Review of N. Chomsky, Syntactic Structures, Language, XXXIII (1957), 375-408. "Cto takoe transformacija?" Voprosy Jazyko-• znanija, X:3 (1961), 69-77. ' , "0 pereformulirovanii transformacionnyx grammatik." " Voprosy Jazykoznahija, X:6 (1961), 41-50. Marouzeau, J. Lexique de l a terminologie linguistique. 3-e ed. Paris, Geuthner, 1951. Martinet, Andre. "Role de l a correlation dans l a phonologie diachronique." 1CLP, VIII (1939), 273-288. 141 , Economie des changements phonetiques. Berne, A. Francke, 1955. Matthews, W. K. "The Ukrainian System of Declension." Language, XXV (1949), 117-129. Meillet, A. Le slave commun. Paris, Honore Champion, 1924. Ohijenko, Ivan (Mytropolyt Ilarion). Ukrajins'ka l i t e r a -turna mova. Saskatoon, The Gospel Press, 1951. _ , Ukrajins'kyj literaturnyj naholos. Winnipeg, Nasa Kul'tura, 1952. Parxomenko, 0. M. Ukraj ins'ka mova. Cast. I. 3-e vyd. Kyjiv, Radjans'ka Skola, 1961. Petlycnyj, I. Z. "Do pytannja pro systemu holosnyx fonem v sucasnij ukrajins'kij literaturnij movi." Pytannja ukraj ins'koho movo znavstva, I (1956), 15-22. Prystupa, P. I. "Tverdi i m'jaki pryholosni pered i v ukra-jins'kij movi." Dialektolohicnyj Bjuleten', no. 5 Richhardt, Rosemarie. Polnische Lehnworter im Ukrainischen. Band 15 der Verdffentlichungen der Abteilung fur tuts an der Freien Universitat Berlin. Herausg. Max Vasmer. Wiesbaden, Otto Harassowitz, 1957. (1955), 3-8. slavische Sprachen und Literaturen des Osteuropa-lnsti-142 Shevelov, George Y. (pseud. Jury S e r e x ) . "Phonema Errans." Lingua, II (1949), 599-418. (pseud. Jurij S e r e x ) . Warys sucasnoji ukrajins'koji literaturnoji movy. Baukove Tovarystwo im. S e v c e n k a , Biblioteka Ukrajinoznavstva. Vol. III. Munich, Molode 2yttja, 1951. (pseud. Yury S e r e x ) . "The Problem of Ukrainian-Polish Linguistic Relations from the Tenth to the Four-teenth Century." Word, VIII ((1952), 529-549. , "A Latent Phoneme in Making: the Affricate ^ in Slavic." Miscelanea homenaje a Andre Martinet: Estruc-turalismo e historia, ed. Diego Catalan. Biblioteca filoldgica de la Universidad de la Laguna, vol. I, Madrid, Editorial Gredos,T957, pp. 251-276. Simovyc, Vasyl'. Hramatyka ukrajins'koji movy. 2-e vyd. Kyjiv, Ukrajins'ka Uakladnja, [1919]. Stankiewicz, Edward. "Towards a Phonemic Typology of the Slavic Languages." American Contributions to the Fourth  International Congress of Slavicists, 's-Gravenhage, Mouton & Co., 1958,; pp. 301-519. , "The-Consonantal Alternations in the Slavic De-clensions." Word, XVI (1960$, 183-205. Stieber, Zdzislaw. "Dwa problemy fonologii slowianskiej." Lingua Posnaniensis, I (1949), 85-86. 143 Teresko, L. S. "Fzajemozv 1 jazok o r f o e p i j i t a o r f o h r a f i j i v sucasnij u k r a j i n s ' k i j l i t e r a t u r n i j movi." Ukrajins'ka  mova v s k o l i , XI (May-June 1961), 31-36. Troubetzkoy, N. S. Principes de phonologie. Trans. J . Cantineau. P a r i s , C. Kl i n c k s i e c k , 1957. Tschernych, P. J . Historische G-rammatik der russischen Sprache. Red. H. H. B i e l f e l d t . Slawist.ische B i b l i o t h e k Nr. 6. Herausg. H. H. B i e l f e l d t ,u. a. Halle (Saale), Max Niemeyer, 1957. Zalisnycko, V. N. "Z dosvidy zastosuvannja fonetycnoho r o z -boru." Ukrajins'ka mova v s k o l i , XI (July-Aug. 1961), 65-68. Z i i y r i s k i , Jan. Opis fonetyczny jezyka ukrainskiego. Polska Akademja Umiejetnosci, Prace Komisji Jezykowej, Nr. 19. Krakow," Gebethner i Wolff, 1932. 2 o v t o b r j u x , M. A. Sucasna ukrakins'ka l i t e r a t u r n a mova. K y j i v , Radjans'ka Skola, 1961. Zwoliriski, Przemyslaw. "Stosunek fonemu v_ do i w h i s t o r i i jezykow slowiariskich." Z p o l s k i c h studiow slawistycz-nych, I, Warszawa, Paristwowe Wydawnictwo Naukowe, 1958, pp. 52-60. 2 y l k o , Pedot Troxymovyc. Narysy z d i a l e k t o l o h i j i u k r a j i n -s ' k o j i movy. K y j i v , Radjans'ka Skola, 1955. 144 , Hovory ukrajins'koji movy. Kyjiv, Radjans'ka Skola, 1958. , "Dialektni vidminnosti holosnyx fonem ukrajin-s'koji movy." Ukrajins'ka mova v skoli, X (July- Aug. 1961), 18-25. , "Dialektni vidminnosti pryholosnyx fonem ukrajin-s'koji movy." Ukrajins'ka mova v skoli. XI (July-Aug. 1961), 9-15. 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items