Based on the distribution of vowel qualities in medial and final syllables, I argue that there is phonological gemination in English. The analysis is cast in terms of optimality theory and has important implications in several domains: first, ambisyllabicity is not the right way to capture aspiration and flapping; second, languages where stress is dependent on vowel quality are perhaps best treated in terms of the kind of covert gemination proposed here.

This article explores the idea that obviation systems like those found in the Algonquian languages are a less parochial solution to syntactic organization than is generally thought. Some simple constraints on obviation provide interesting analyses of key facts in Algonquian. Once certain language-particular differences are recognized, a number of syntactic problems in two unrelated languages, Tzotzil (Mayan) and Chamorro (Western Austronesian), yield easily to solutions based on obviation, despite the absence of obviative-based morphology in either language. Hierarchy alignment constraints play a central role in the analysis. The account is articulated within optimality theory, which provides an appropriate framework for representing the fact that these languages must select from their resources for expressing transitive propositions the optimal mode of expression for each such proposition.

Comaltepec Chinantec tone sandhi normally consists of rightward high tone spreading from syllables with a low-high tone pattern, and almost all sandhi outputs are non-neutralizing. This sound pattern may be understood by considering articulatory, aerodynamic, acoustic, and auditory principles, in necessary combination with phonetically rooted historical forces, and the principles of contrast maintenance, economy of effort, and what is termed pattern coherence.

This article examines two processes that affect vowels in Igbo: harmony and assimilation. Through these two processes, the relationship between autosegmental features and articulatory gestures is explored. Vowel harmony is argued to be featurally represented, but acoustic evidence shows that vowel assimilation is gradient and best represented in terms of articulatory gestures. Neither representation is adequate in itself to describe the full range of phonological and phonetic data; rather, I advocate a mapping procedure that takes advantage of the resemblances between autosegmental and gestural representations without collapsing the two. A complete account of Igbo vowel harmony and assimilation is provided, demonstrating the need for two kinds of representation, and illustrating the suggested feature-to-gesture mapping.

Among the common strategies for eliminating vocalic hiatus is vowel elision. In some cases, it is the first vowel (V1) that elides, while in others it is the second (V2). Analyses of elision have, virtually without exception, simply stipulated which vowel is elided, for example, by encoding this information directly in a language-specific rule. This implies that the targeted position is not predictable, but simply a matter of which of two equally available options is selected by the language. A cross-linguistic study suggests, however, that this is not strictly the case, but that in some environments the choice of target is universally determined. This article accounts for the observed restrictions on elision target within a constraint-based theory which claims that languages preferentially preserve phonological elements in certain prominent positions.

Roman Jakobson in memoriam, on the occasion of his hundredth birthday, October 12, 1996.

Some researchers have claimed that WH-movement in ASL is rightward, contrary to the apparent universality of leftward WH-movement. In contrast to this claim, we argue that WH-movement in ASL is to a leftward specifier of CP. We account for the occurrence of rightward WH-elements by independently motivated syntactic and discourse factors which lead to the appearance of WH-elements in sentence- or discourse-final positions—not by rightward WH-movement. Our analysis provides an account for a variety of ASL direct and indirect WH-questions and is in accord with cross-linguistic generalizations.

This article presents a linguistic analysis of a specific feature of a literary genre: the artificial punning found in the Egyptian Arabic narrative ballad (as described in Cachia 1989). A comparison of how these puns differ from regular processes in the phonology and morphology of the language reveals that this encoding by the poet-performer is very much a mirror image of regular processes. The audience's decoding of them, therefore, follows a pathway similar to regular processes. The dichotomy between the puns' linguistically based formal composition and their contextually based semantic interpretation is analyzed within a reinterpretation of a Jakobsonian structuralist framework involving a hierarchization of linguistic levels based on two factors: the degree of combinatoric freedom and the degree of semantic immediacy. This analysis reveals that the artificial punning in these ballads is actually the obverse of what one would expect to find following the definition of poetic discourse given by Roman Jakobson. This study thus shows that such artificial punning subverts normal expectations about poetic discourse and this has great implications for understanding the production and interpretation of literary word play in any tradition.

Pro-drop languages have restrictions on the reference of pro not found with the overt pronominals of non-pro-drop languages. In particular, while the overt pronouns of non-pro-drop languages may take clausal antecedents, C/IPs, pro may not take these elements as linguistic antecedents. This restriction on the referential properties of pro follows from a mismatch in Phi-features; pro, which is or is licensed by Phi-features, cannot corefer with a clause, which is Phi-featureless. We discuss implications of our analysis for linguistic theory.

It is well known that personal pronouns in Japanese such as kare ‘he’ and kanozyo ‘she’, unlike their English counterparts, cannot be construed as bound variables in logical form. The purpose of this article is to argue that this cross-linguistic difference is due to the difference in syntactic categories. English personal pronouns are determiners (Postal 1969), exemplifying what will be referred to as D-PRONOUNS, and can be construed as bound variables, whereas Japanese personal pronouns are nouns, exemplifying what will be referred to as N-PRONOUNS, and cannot be so construed. I argue that this follows from a general condition on binding that applies only to functional items, and not to lexical ones. I provide empirical and conceptual support for this hypothesis on the basis of the behavior of such elements as articles, determiners, and demonstrative pronouns as well as that of personal pronouns.

The ‘time‘-away construction, exemplified by We slept the whole afternoon away, proves to have a complex set of syntactic and semantic properties. In particular, the NP the whole afternoon behaves syntactically like a direct object, even though it is clearly not licensed by the verb sleep. This construction is shown to be distinct from two others that it superficially resembles, the resultative and the way-construction. It is also compared with a number of other semi-idiomatic VP constructions. Two approaches for licensing the NP object are compared: a lexical rule approach, in which sleep away is treated as a complex verb that licenses the object, and a constructional approach, in which V NP away is listed as a meaning-bearing construction that licenses both the verb and the object.

Georgian person-number affixes reflecting subject and object agreement coexist on verb stems. As a result of limited morphotactic space, in several contexts certain markers surface while others that are equally semantically motivated do not. Inflectional blocking, relying on general linguistic principles of specificity and analogy, accounts for surface verb forms in a novel, explanatory fashion. The implication for Georgian is that instead of employing inflection produced by syntactically relevant affixation rules, its verb morphology results from morphotactic constraints, blocking of potential affix combinations, and verb stem insertion into instantiated inflectional affix frames.

This article presents the results of two studies which show that gender bias and stereotyping are widespread in the example sentences of syntax textbooks. Results from both studies indicate that little has changed over the past twenty-five years: virtually all of the authors favor male-gendered NPs as subjects and agents and regularly stereotype both genders. Throughout the paper we make reference to the LSA guidelines for nonsexist usage, pointing out the need for such guidelines, and highlighting the gaps in the current version.

There is now, surprisingly, a sizeable literature on pronouns in Native American languages and what their historical significance may be (references cited in N&P 1996 and Campbell 1994). Nichols & Peterson 1996 (henceforth Nichols and Peterson) is significant both for what it reveals about these pronouns and for the questions it raises about how they might be explained. Nichols and Peterson investigate the distribution of the pronoun pattern with n ‘first person’ and m ‘second person’ and assert that it is limited essentially to certain western American Indian languages and to Austronesian languages of northern New Guinea, based on ‘a moderately large sample of the world's languages’ (337), largely copied from Nichols 1992. They argue that this distribution is no accident, but rather is ‘a single historical development … though we cannot determine the exact nature of the shared history (common descent? areal affinity? etc.), it can be given a chronology and a geography and tied in to a bigger picture of circumpacific migration’ (337). The interpretation of this single historical development, however, is obscured in the discussion. My purpose here is to offer a different interpretation of their findings and thus hopefully to contribute to greater understanding of the role of pronouns in comparative linguistics in general. N&P contributes a number of significant insights concerning criteria for genetic relatedness. The reinterpretation offered here does not concern these, but rather only their more specific claims about the geographical distribution of pronoun patterns and how they are to be understood.

The forms represented orthographically as 〈wanna〉, 〈hafta〉, 〈gonna〉, 〈gotta〉, 〈oughta〉, 〈usta〉, and 〈sposta〉 have standardly been analyzed as involving a syntactic rule or cliticization operation called to-contraction. Occasionally it has been suggested that the forms in question have been iexicalized', i.e., WANNA and HAFTA are synchronically distinct lexemes from WANT and HAVE. I argue here that neither approach is correct. The syntactic accounts are wrong to assume that the relation between wanna and want to must be syntactic, and the lexicalization accounts are wrong to assume that there is no synchronic relation: the link is one of derivational morphology. A morpholexical rule suffixes /tu/ ~ /tə/ to the base lexemes to form derived lexemes such as WANNA. These to-derivatives are headed morphological structures, as described by Stump 1994. They inflect on their heads, not their edges; they are synonymous with their bases but have different subcategorization and more colloquial style associations. Various morphological and phonological idiosyncrasies indicate that the derived lexemes are morphologically compound, but their sharing of the lexical idiosyncrasies of the base lexemes show that they contain those bases as heads. All the syntactic phenomena that have been claimed to be relevant to the debate over to-contraction fall into place under the assumptions advocated here, and some new insights emerge, particularly with regard to the ‘liberal dialects’ where the pronunciation written 〈wanna〉 has a wider distribution than in most American dialects.