Formal Problems in Semitic Phonology and Morphology

John J. McCarthy

• Publisher: Taylor & Francis
• Year: 2018
• Print ISBN: 9781138604247
• eBook ISBN: 9780429888243
• Edition: 1
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

First published in 1985. Two basic issues figure in this study. The first concerns the representation of syllabic and accentual structure, and the effects of those structures on the formulation of phonological rules. In the second section of this title, a solution to the traditional problem of the root and pattern morphological system of Semitic is proposed and illustrated by an extensive treatment of Classical Arabic. This title will be of particular interest to students of linguistics.

Chapter 1: Prologue

This study is based to a great extent on the basic assumptions of generative phonology, and for that reasons assumes a certain familiarity with representative works like Chomsky and Halle (1968) and subsequent literature. This is not to say that it is a purely descriptive work within that theoretical framework; rather, it deviates in fairly fundamental ways from Chomsky and Halle’s modes of representation and rule formulation. In fact, the underlying thesis here supports a variety of far-reaching changes in the received generative theory with a number of empirical consequences.

Two basic issues figure in this study. The first, dealt with in chapters 2 and 3, concerns the representation of syllabic and accentual structure, and the effects of those structures on the formulation of phonological rules. An essentially hierarchic model is developed, along the lines first introduced in Liberman (1974). This model is shown to have very broad consequences for the segmental phonology and accentual system of Tiberian, and equally important results in the accentual systems of Classical Arabic and two Arabic dialects.

In chapter 4, a solution to the traditional problem of the root and pattern morphological system of Semitic is proposed and illustrated by an extensive treatment of Classical Arabic. The solution basically runs along the lines of theoretical proposals developed most clearly in Goldsmith (1976). Although the Semitic problem is itself of great inherent interest, the morphological model as conceived here is shown to lead to a variety of other consequences, in particular a strong constraint on the form of morphological rules and a deeper understanding of nonconcatenative morphology in general.

Although these two aspects of this study are to some extent independent, and in fact any of the three following chapters can be read separately with little loss of comprehension, there is one unifying idea behind all. The thesis is that several formal enrichments, along basically prosodic lines, of the theory of Chomsky and Halle (1968) are both descriptively necessary and theoretically desirable. The descriptive necessity emerges throughout the discussion, while the theoretical desirability of these enrichments lies in the possibility, explicitly followed at several junctures, of either constraining or eliminating the earlier apparatus.

One important point about the mode of presentation is in order here. The discussion throughout this work almost invariably eschews polemic in favor of more direct arguments in support of the proposals made. Thus I have avoided the construction of straw men and like rhetorical devices on the grounds that they properly belong to the process of scientific discovery and not to the exposition of finished results. Another aspect of this arises in the development of the model of metrical structure here. I have benefited a great deal from reading Halle and Vergnaud (1979) as well as other recent, often unpublished works on this subject, and this debt is ackowledged throughout the text. In many cases these other treatments conflict with mine on matters of varying significance. I have not felt it necessary to give direct recognition to all of these disagreements for two reasons. First, for most of them the data now known and understood with any degree of clarity do not determine whether the issue is substantive or merely notational. Second, in view of the very rapid changes in such a novel theory, I have thought it best to present a single, relatively consistent model which is fairly simple formally and which is supported by several thorough analyses.

This brings us to another point, the descriptive basis of this work. It goes without saying that any analysis that tends to disprove any proposals made here will have to be based on an empirical foundation equal to or greater in depth than the analyses here. I do not claim to offer an exhaustive treatment of the phonology or morphology of any language, but a fair degree of coverage, particularly in Hebrew phonology and Arabic morphology, is achieved. This aspect of the study has been aided by the existence of two previous generative treatments, Brame (1970) on Arabic phonology and Prince (1975) on Hebrew phonology. Prince’s work especially contributed much to the analysis of Hebrew in chapter 2 and some preliminary metrical insights in chapter 3. Where some phenomenon is known to me solely through a published description (as in the case of Tigre and Maltese in chapter 3), I have noted this explicitly.

A few practical matters. Because of the very large number of forms and rules cited in displays, I have adopted a policy of numbering displays anew beginning at each major subdivision of a chapter. Thus unique reference to any display will require three integers, like “chapter 3, section 4, (43)”. To abbreviate the footnotes I have left out glosses and have sometimes used Orientalists’ technical terms where the alternative is a very long explanation. I urge those with sufficient interest to consult a reference grammar of the appropriate language for a definition of the term and often an extensive discussion of the relevant phenomenon. With this exception, however, the notes are mostly quite accessible.

Finally, the mode of transcription. In both Hebrew and Arabic, 9 and ḥ are the voiced and voiceless pharyngeal glides respectively. A subscripted dot, as in ṭ, ḍ, ṣ, and δ indicates emphatic (pharyngealized) articulation. q is a voiceless unaspirated uvular stop, and ? indicates glottal stop. All other consonants have their familiar values. Because of the difficulties in devising a suitable transcription, I have not marked spirantized allophones of Hebrew stops except when relevant (in chapter 2), when they are indicated by an extra subscript line in b, t, d, k and by a superscript line in p and

. Hebrew ś is a consonant of unknown value, possibly palatalized s.

Long vowels in Arabic are indicated simply by gemination. Gemination in Hebrew long vowels is also the formal representation adopted here and followed in discussions of syllable structure and accentuation, but actual cited examples use a somewhat more elaborated mode of transcription. Long vowels written without a mater lectionis have a macron (ā, ē, ō), while those, except ā, written with mater have a circumflex (û, î). A breve over a vowel (ă, ĕ, ŏ) transcribes one of the

, an extra-short or reduced version of that vowel. I should point out here that my assumption that the basic distinction in Hebrew vowels is quantitative is supported by the Qimḥi school as well as a vast number of internal phonological considerations. I do not exclude the possibility of an earlier pronunciation like the modern Ashkenazic, but I would point out that this pronunciation involves a simple mapping from vowel quantity onto [tense], with an adjustment for the rounding of ā. Thus the quantitative distinction is basic, though some traditions superimpose a qualitative one onto it.

Portions of chapter 3 on Arabic stress appear in my article “On Stress and Syllabification,” Linguistic Inquiry 10,3. A very early version of the treatment of Arabic vocalism in chapter 4 was presented in 1976 at the North American Conference on Afro-Asiatic Linguistics.

Chapter 2: Syllable Structure and Segmental Phonology

1. Introduction

A theory of syllable structure is presented here in which segments are hierarchically arranged into higher-order constituents of a binary-branching tree. In general, one and only one such tree is associated with each syllable, rooted on the syllable node σ with individual segments as the terminal nodes. This first section of this chapter develops the very broad outlines of a theory of these syllable trees, illustrating the points with examples from Arabic and Hebrew.

The second section deals with the application of phonological rules to these enriched segmental representations. Phonological rules are allowed to operate on the trees directly, as well as on the segments, and some general principles governing their interaction are proposed. In this chapter our attention is mostly confined to segmental phonological rules, in the familiar sense, while chapter 3 deals in detail with a variety of accentual issues.

The final section of this chapter offers an extensive illustration of these principles of rule application and of the hierarchic syllable structures from the segmental phonology of Tiberian Hebrew. An even more thorough analysis of the accentual phenomena of this language can be found in section 3 of chapter 3.

The basic notion that syllables have internal hierarchic structure — that they can be parsed into units smaller than a syllable but larger than a segment — is scarcely new. The earliest explicit reference to this idea that I have located is Pike and Pike (1947), though undoubtedly one could find earlier treatments, perhaps even in antiquity. In fact, the so-called syllabic orthography of Akkadian in the second millenium BC is not strictly syllabic, but depends on a hierarchic treatment of this sort. Thus, the writing i-na-ad-di-in for inaddin ‘he gives’ implicitly reflects a division of closed syllables (CVC) into two partially-overlapping subunits each larger than an individual segment.

The notational foundation of this theory of syllable structure comes from two separate sources. First, there is the idea of an essentially autosegmental characterization of syllable membership developed in Goldsmith (1976) and most extensively in Kahn (1976). Rather than say that syllables are delimited by boundary symbols in the segmental string, this claims that for every syllable there is a node a on an autosegmental tier which is associated with just the segments in that syllable. Second, an extension of this notation by allowing σ to dominate a full binary-branching tree permits us to give an internal constituency to the syllable. This first appeared in Prince (1975), where it was intended to describe some processes of compensatory lengthening in Hebrew. A further extension of this by Paul Kiparsky (in class lectures and in Kiparsky (to appear)) involves labeling the nodes of this tree for a relationship of relative strength, along the lines of the theory of stress prominence in Liberman (1974) and Liberman and Prince (1977).

2. Syllable Structure

The justification of constituent structure in classical transformational syntactic theory has scarcely been uncontroversial in particular cases, but the methodology is generally agreed upon. First, application of rules of movement, deletion, agreement, and concord to syntactic strings is usually taken as prima facie evidence that they form constituents. Second, the statement of distributional regularities — like specifications of lexical subcategorization — is usually supposed to be confined to constituents. Third, the consistent appearance of similar strings in a variety of different rules of the first and second sorts leads to a theory of syntactic types.

Evidence of this sort exists in phonology as well, though there is a fundamental difference. While syntactic co...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Acknowledgments
• Chapter 1: Prologue
• Chapter 2: Syllable Structure and Segmental Phonology
• Chapter 3: Syllable Structure and Accentuation
• Chapter 4: Prosodic Structure, Morphology, and the Lexicon
• Bibliography