In this study the use of variables for the abbreviation of disjunctive sets of phonological rules is examined. We ask whether there are rule sets which require more than one variable, and answer the question in the affirmative. The evidence consists of a new formulation of the English vowel-shift rule, using a pair of asymmetric features which distinguish four tongue heights for vowels. Within the context of the proposed rule, several theoretical issues concerning rule formalism and phonological change are discussed.

The problem of non-unique phonological representations is examined in its historical context. The discussion treats two types of phonological representations: the traditional phonemic representations and the underlying representations of generative phonology. Both systems allow for non-unique representations. In phonemics, a resolution of this problem was found within the distinctive-feature framework, whereas in generative phonology, unique representations become possible only if one recognizes a theory of phonological markedness.

Although it is generally admitted that the thematic verb stems in Indo-European are of more recent origin than the athematics, and have increased at their expense,1 no relative chronology has, to our knowledge, been ventured thus far. The conventional view of the distribution of athematics and thematics seems to be that both types existed even in quite early Proto-Indo-European. Surprise has therefore been expressed that so few simple (uncharacterized) thematics are attested in Hittite,2 and Kronasser even asserts that simple thematics inherited into Hittite were recast as athematics.8

In §1, a reconsideration of the orthographic evidence for Gothic yields a synchronic distinctive-feature analysis of its phonology. In §2, a general procedure is proposed for determining a Stammbaum, by positing that tree which assumes the fewest rules of sound change from the proto-language to the dialects. In considering the diachronic rules of early Germanic, North and West Germanic are shown to share more rules than do North Germanic and Gothic. Hence the optimal tree must split first into a North-West Germanic branch and an East Germanic one. Available non-linguistic evidence (§3) supports this view.

This study presents two main theses: (1) that distinguishing ‘primitive’ from ‘derivative’ forms and functions can be accomplished in a linguistic description by a base containing two types of rewrite rules, expansion and substitution; and (2) that adjuncts to a noun can, under certain conditions, dissociate from that noun and pass from ‘attributive’ to ‘adverbial’ function—in particular, that clausal modifiers to the nominative can so detach and become predicate appositives and absolutes. These two proposals suggest a number of corollaries—among them, that the major adverbial clause types (condition, concession, purpose etc.) are basically adjuncts to the subject noun.

Russian stress alternations are usually considered capricious and unrelated to grammatical meanings, but this paper argues that that is true only on the phonemic level. An analysis of such alternations in terms of underlying morphophonemic entities and ordered rules makes it clear (1) that Russian stress shifts are clearly correlated with the grammatical categories of number and case, and (2) that all stress shifts in the Russian substantive can be described by a single alpha-switching rule with varying environmental conditions. By listing the unmarked singular stem in the lexicon, generating the plural stem therefrom, and then generating the several case forms within each number stem, one obtains a simplification of the stress-shift rules and a generalization of their applicability. It is thus shown that a number of Russian words heretofore considered exceptional can all be covered as members of a single class.

Arabic verbs are traditionally considered to be inflected for number when they follow the subject, but not when they precede the subject. The resultant ‘nominal’ sentences of the former case can be related to the ‘verbal’ sentences of the latter case by a focus transformation which duplicates a noun before the verb and leaves the initial instance of the noun after the verb to be pronominalized. This pronominalization, in the case of subject nouns, is realized as the so-called number inflection of the verb. Such a treatment also simplifies the analysis of relative clauses, ostensibly subjectless sentences, and sentences containing copula pronouns.

This paper presents positive evidence for a genetic relation between Dravidian and Uralian. The major evidence consists of systematic phonemic correspondences between 153 Proto-Dravidian and Proto-Uralian lexical items. Impressionistic structural resemblances support the lexical evidence.

Ten terms for Buddhist religious functionaries in standard Thai are subjected to semantic analysis. The data are grouped into a semantic set which is exploited to reveal redundant, hence predictable, semantic features. In addition, semantic patterning within the set is discussed and is related to the relevant subset of classifiers.

Alternative hypotheses regarding the vowel system of Proto-Sahaptian and the transition to the two daughter languages, Nez Perce and Sahaptin, are examined. The proto-language may possibly have had a five-vowel system ∗/i e a o u/, identical to that of Nez Perce, and already showing the same pattern of vowel harmony alternations. But a reconstruction of the vocalic system before the development of vowel harmony, whether or not this predated Proto-Sahaptian, presents several possibilities. A six-vowel system ∗/i ə e a o u/ would allow a straightforward development. A suggested three-vowel system is rejected. The possibilities of development from four- or five-vowel systems are examined, but these, while conceivable, are less plausible. Examination of the comparative evidence suggests that Proto-Sahaptian possessed the six-vowel system and may have lacked vowel harmony. Some areal similarities to such a system are noted.