Annual Review of Linguistics - Volume 2, 2016

Morris Halle has been one of the most influential figures in modern linguistics. This is partly due to his scientific contributions in many areas: insights into the sound patterns of English and Russian, ideas about the nature of metered verse, ways of thinking about phonological features and rules, and models for argumentation about phonological description and phonological theory. But he has had an equally profound influence through his role as a teacher and mentor, and this personal influence has not been limited to students who follow closely in his intellectual and methodological footsteps. It has been just as strong—or stronger—among researchers who disagree with his specific ideas and even his general approach, or who work in entirely different subfields. This appreciation is a synthesis of reflections from colleagues and former students whom he has formed, informed, and inspired.

The nineteenth-century conception that linguistic structure was to be explained by recourse to the histories of languages was largely abandoned with the rise of synchronic theories in the twentieth century, but has recently returned to prominence. Whereas traditional generative theories of language have tended to attribute crosslinguistic regularities to constraints imposed on the class of possible grammars by the human Language Faculty, some scholars have argued that this is often a mistake: that there are no (or at least very few) real substantive universals of language, and that the regularities in question arise from common paths of diachronic change having their basis in factors outside of the defining properties of the set of cognitively accessible grammars. This review surveys evidence for that position, primarily in phonology but also in morphology and syntax. I argue that in phonology, there are at present no convincingly demonstrated substantive universals governing the set of possible regularities, and that the generalizations we find should be attributed to a combination of contingent historical developments and biases in the learning algorithm that relates available data to the grammars learners acquire. In morphology and syntax, I argue that some apparent generalizations are indeed the product of diachronic change rather than synchronic constraint.

Phonological representations capture information about individual word forms and about the general characteristics of word forms in a language. To support the processing of novel word forms as well as familiar word forms in novel contexts, an abstract level of representation is needed in which many phonetic details and contextual features are disregarded. At the same time, evidence has accumulated that such details are retained in memory and used in processing individual words and indexical features of language. Taken together, these results mean that a hybrid model of phonological representation is needed. The abstract level supports generalizations based on lexical type statistics and fast adaptation to communicative requirements through the reuse of existing categories. A richly detailed level of representation is implicated in word-specific phonetic patterns, the detailed dynamics of regular sound changes, and active associations of phonetic patterns with gender, age, and dialect.

This article surveys the history of contrast in phonology from Bell's Visible Speech (1867) until Chomsky & Halle's Sound Pattern of English (1968). Phonological contrast can be viewed at the segmental and subsegmental (feature) levels. As contrast at the segmental level involves the phoneme, whose later history has been extensively documented, I concentrate on the origins of the concept in the work of Sweet. Subsequently, I focus on subsegmental-level contrast. After a look at its treatment in phonological analyses that operated without an explicit theory of features, I turn to Trubetzkoy, in whose work we find the seeds of later approaches. The article explores the foundations of the main methods of computing contrastive features: minimal differences and hierarchical feature ordering. It concludes with a discussion of contrast in early generative phonology and reviews some of the reasons for its decline at the end of the 1960s.

Research on spoken word perception and production has identified two hallmarks of spoken word processing: multiple activation of representations of the sound patterns of words in memory and subsequent competition among these patterns. Evidence for this activation-competition process has come, in part, from experimental studies examining the effects of phonological neighborhoods, which are collections of similar-sounding words that are activated in memory during both perception and production. In this article, we review more than 20 years of research on phonological neighborhood effects in spoken word processing that has demonstrated that the speed and accuracy of spoken word perception and production are, in large part, a function of the density and frequency of neighborhoods of spoken words. We conclude our review with a discussion of new avenues of research—based on recent advances in network science—that hold the promise of deepening our understanding of the mental operations involved in our uniquely human capacity for communicating with the spoken word.

Although consonantal variation has traditionally been studied using auditory coding, techniques now exist for measuring any kind of consonants acoustically and/or articulatorily. These methods have already been employed extensively for studying variation in many languages. Techniques and past studies using them are reviewed for rhotics, laterals, fricatives, stops, weakening and strengthening processes, and voicing. These methods are becoming well established in sociolinguistic inquiry. One of the greatest remaining challenges is to design studies that combine these methods with current sociological approaches to human interactions.

It is well known that syntax matters to phonology, but does phonology matter to syntax? This question is controversial, partly because judging whether a phenomenon is syntactic or phonological is not always straightforward, and partly because the relevant data (e.g., well-formedness judgments) are often noncategorical, yet grammatically structured in ways that elude performance explanations. This review discusses examples of phonologically constrained syntactic variation and shows how they can be understood in terms of the variation + filtering theory of syntax–phonology interaction. This theory has the following key properties: (a) Phonology refers to syntax, but syntax does not refer to phonology; (b) syntax predicts variation by admitting alternative linearizations as well as alternative choices among constituents; and (c) phonology acts as a filter by evaluating the phonological well-formedness of the variants. I conclude that phonological effects on syntactic variation are limited, but real.

The distinction between lexical and functional elements plays a major role in current research in syntax and neighboring aspects of the study of language. In this article, we review the motivations of a progressive shift of emphasis from lexical to functional elements in syntactic research: the identification of the functional lexicon as the locus of the triggering of syntactic actions and of syntactic variation, and the description and analysis of the complexity of functional structures in cartographic studies. The latter point leads us to illustrate current cartographic research and to present the maps created in the study of clauses and phrases. The maps of CP, IP, and other phrasal categories all involve a richly articulated functional sequence. We then address issues of the numerosity and typology of the functional lexicon, the constraints on the featural specifications of possible functional heads, and the relations between cartographic research and minimalism.

Some languages showing morphological ergativity in case and/or agreement also show ergative patterns in core syntactic domains—syntactic ergativity. The most-studied type of syntactic ergativity is a ban on the Ā movement of ergative subjects; an additional type concerns the distribution of absolutives in nonfinite clauses. This article first presents the standard view of syntactic ergativity, which is closely connected to the treatment of ergative as an inherent case. Evidence from Shipibo suggests that a ban on ergative Ā extraction does not require inherent ergative. This points to a view of syntactic ergativity centered around morphological case discrimination. One consequence is that pure head-marking languages cannot feature a true ban on ergative extraction, because ergative morphological case is not in use. This conclusion highlights the challenging tasks of diagnosing extraction restrictions in pure head-marking languages, as in the Mayan and Salish families, and of distinguishing extraction restrictions from instances where extraction merely interacts with agreement. A variety of crosslinguistic evidence suggests that agreement/extraction interactions are fully possible in morphologically ergative languages, and not only for ergative arguments. Special morphology in the context of transitive subject extraction is therefore not necessarily evidence of syntactic ergativity.

The explanation of island phenomena has been a central feature of formal grammatical theory practically since its inception. However, a growing number of linguists have provided explanations for these phenomena that are not based on purely syntactic constraints. Some have proposed alternative explanations that appeal to information structure or to semantic information, and others find the basis for island constraints to lie in processing. This review documents the full range of nonsyntactic explanations of island constraints.

Though the term “existential sentence” goes back at least as far as Jespersen (1924, p. 155) and is used in descriptions of many languages to refer to a designated construction, it is difficult to identify exactly what these constructions have in common crosslinguistically. Following McNally (2011, p. 1829), the term is used here to refer to sentence types that are “noncanonical,” whether due to some aspect of their syntax or the presence of a distinguished lexical item (e.g., Spanish hay) and that are “typically used to express a proposition about the existence or the presence of someone or something.” I discuss a representative sample of the different structural resources used to build existential sentences: distinguished existential predicates, on the one hand, and copular, possessive, and expletive or impersonal constructions, on the other. I then address the corresponding variation in the compositional semantics of existentials, as well as pragmatic or discourse functional variation. I contrast the variationist perspective with universalist approaches to existentials, such as that by Freeze (2001).

Languages may vary greatly in the way they express negation. Most languages exploit specifically designated negative markers, such as English not. Many languages may also use negative indefinites (such as English nobody or nothing) to express negation. The behavior of these negative indefinites is subject to crosslinguistic variation: In some languages, negative markers and negative indefinites cannot express a single semantic negation (nobody didn't come means that everybody came and not that nobody came), but in other languages they can. Languages with these properties, such as Italian, are called Negative Concord languages. In this review, I discuss the difference between negative indefinites in languages that exhibit Negative Concord and languages that do not. I also compare the behaviors of negative indefinites in languages that exhibit Negative Concord and so-called Negative Polarity Items. This article provides an accurate overview of recent developments in the study of negation and negative dependencies.

Free indirect discourse has traditionally been described as a form of reported speech or thought. It seems to be a mixture of both direct discourse (in allowing exclamatives, interrogatives, etc.) and indirect discourse (in following sequences of tenses and pronouns). It has been the object of more interest from literary theorists than from linguists, though Banfield (1982) offered what is still the best syntactic description of the phenomenon, and contemporary semantic accounts have brought new insights into it. Schlenker (2004) made decisive progress in proposing an account of two contexts and indexical shifting. Maier (2015) proposes an alternative quotational analyses, which Eckardt (2015) rejects, going back to Schlenker's model, suitably amended to answer Maier's criticism. We present these theories, criticize them, and propose an extension of the Schlenker–Eckardt model.

Recent years have seen a surge of experimental approaches to the study of natural language meaning, both to obtain solid data on subtle phenomena that are hard to assess through introspection and to understand how abstract characterizations of linguistic knowledge relate to real-time cognitive processes in language comprehension. This article reviews research in one of the most recent areas to see extensive experimental investigations, namely presupposition and presupposition projection. Presuppositions are at the very nexus of linguistically encoded content and contextual information, as they relate directly to the discourse context but also interact in intricate ways with their intra-sentential linguistic environment. They are thus extremely suitable for investigations of the interplay of linguistic and more domain-general processes in language comprehension, as well as for experimental investigations of subtle theoretical phenomena.

This article presents a review of works on expressive linguistic objects in the Principles and Parameters framework (Chomsky 1995a,b; 2001). It discusses case studies of expressives in Russian, German, Halkomelem, Maale, Walman (Valman), Kolyma Yukaghir, Itelmen, Tongan, and Spanish, as well as in Breton and Yiddish. The article demonstrates that two semantic types (size and attitude) of expressives correspond to a great variety of syntactic structures across languages, as well as within a single language (e.g., Russian). It shows that there is no direct one-to-one correlation between the form and function of expressives, which has important implications for the syntactic–semantic mapping of categorization (Wiltschko 2008, 2014). This review is of interest to theoretical linguists, language-area specialists, and language typologists. It is also relevant to the fields of education and endangered Aboriginal language documentation, maintenance, and revitalization, as one-third of languages investigated here are endangered and on the verge of extinction (e.g., Halkomelem, Itelmen, Kolyma Yukaghir, Walman).

Sentiment analysis is a growing field at the intersection of linguistics and computer science that attempts to automatically determine the sentiment contained in text. Sentiment can be characterized as positive or negative evaluation expressed through language. Common applications of sentiment analysis include the automatic determination of whether a review posted online (of a movie, a book, or a consumer product) is positive or negative toward the item being reviewed. Sentiment analysis is now a common tool in the repertoire of social media analysis carried out by companies, marketers, and political analysts. Research on sentiment analysis extracts information from positive and negative words in text, from the context of those words, and from the linguistic structure of the text. This brief review examines in particular the contributions that linguistic knowledge can make to the task of automatically determining sentiment.

Linguistically, globalization has given rise to both homogenization and differentiation. Imperial conquest and standardization in the European nation-building tradition led to the spread of dominant languages and varieties and the creation of sometimes artificial linguistic boundaries. Standardization associated with nation-building continues, alongside new modes of standardization associated with capitalism. However, the factors that can lead to the reduction of diversity can also lead to its growth. The review sketches processes of linguistic localization associated with political change, tourism, migration and urbanization, and regional identification, and concludes with a detailed look at mechanisms and effects of standardization and localization in the history of Pittsburgh speech.

Since the 1950s, in analyzing the reasons US schools fail African American learners, sociolinguists have located dialectal difference as central to the educational theater of operations. Recognizing that African American Vernacular English (AAVE) was not itself an obstacle to learning, linguists instead identified “ignorance of nonstandard English rules on the part of teachers and text writers” as the culprit (Labov 1969, p. 29). This article reviews the history of academic responses to the vernacular “language problem,” identifying key issues, moments, controversies, and positions relevant to teaching Standard English in African American classrooms. It argues that while linguistic scholars and critical language theorists have developed intricate analyses, knowledge, and responses to vernacular dialects, these advances have not taken root in the schools of America. This article concludes that until standardized assessments require teachers to distinguish pattern in vernacular dialect from error in Standard English, US schools will continue to fail African American learners.

Infants begin learning the phonological structure of their native language remarkably early and use this information to extract word-sized chunks from the speech signal. While acquiring the language-specific segmentation strategies appropriate for their native language, infants are simultaneously beginning to form word–object pairings and learning which sound contrasts are meaningful in the native language. They are also working out how to assign words to word classes, paying attention to the use and placement of function words, and learning how speakers of the language string words together to form sensible grammatical utterances. Amazingly, infants tackle all of these tasks simultaneously, with success in each of these domains dependent on success in the others. This review focuses on infants' discovery of word forms in speech, their construction of a proto-lexicon, and the development of linguistic knowledge during their first year and a half of life. By discussing the development of lexical knowledge in relation to other aspects of linguistic development, I demonstrate the advantages of an integrative approach to understanding early language acquisition.