Vowels vary systematically in F0 based on intrinsic properties of the vowel (e.g. height: VF0) and preceding obstruent (e.g. voicing: CF0). These patterns have been attested in most languages studied, raising the possibility that they stem from physiological sources. However, previous results show a range of variability in maximum effect size and duration. One explanation is that variation could be learned and increased to enhance phonological contrasts or suppressed for functional reasons (e.g. for tone languages). Alternatively, differences could be due to physiological factors such as laryngeal contrast type. We map out the distribution of intrinsic F0 effects across twenty languages, using large corpora of read speech, operationalizing CF0 as the difference between stop series that most closely approximate phonologically ‘voiced’ and ‘voiceless’ stops. We find that both VF0 and CF0 effects are present and are in the same direction in all languages examined, but languages vary greatly in effect size. While some of this variability may be due to phonological properties of the languages (e.g. tone), and some variability in CF0 may be due to the diverse phonetic realizations of ‘voicing’ across languages, much of this variability remains to be explained. We find that the CF0 effect is consistently at least as large as the VF0 effect and is more variable across languages, suggesting a possible explanation for the tendency for CF0 effects to lead to sound change much more often than VF0 effects do. While our results on variability in CF0 effects rely on the validity of ‘lumping’ together diverse phonetic realizations, those on the robustness of CF0 and VF0 and on the variability in VF0 hold regardless. These results motivate further investigation to deepen our understanding of intrinsic F0 effects, their crosslinguistic distribution, and their role as precursors to sound change.

Certain varieties in the Dutch and German language area distinguish two groups of words on the basis of different tonal melodies, a phenomenon called (Franconian) tonal accent. Despite extensive efforts, scholars are still far from reaching consensus on how this contrast may have developed. We show how both the accent genesis and distributional variation across dialects can be modeled by assuming that intrinsic durational differences between two sets of vowels formed the basis of the opposition. A key insight emerging from our scenario is that vowel-lengthening processes in open syllables must have been ongoing when apocope was completed, counter to the received view in Germanic historical phonology. Our analysis also addresses various broader issues concerning mechanisms of prosodic change, the interaction of sound changes, and the reliability of manuscript evidence.

Most formal research on the imperative sentence type has focused on canonical imperatives, forms like Leave!, which are often characterized across languages by properties such as bare verbal morphology and omission of the subject. Noncanonical imperatives therefore offer an opportunity to investigate imperative properties from a different perspective. This article argues that negation-licensed commands, forms like No smoking!, first introduced in Iatridou 2021, contain an unusual combination of properties that offer a unique insight into the nature of canonical imperatives. This article has two main findings: (i) negation-licensed commands have a morphosyntax similar to that of existential declaratives, but a speech-act update similar to that of canonical imperatives, and (ii) their speech-act update is subtly different from canonical imperatives in a way that motivates a reevaluation of the speech-act operator in canonical imperatives. This article therefore demonstrates that noncanonical constructions are worth studying not only because of their interesting properties, but also because they offer insights into canonical constructions that could not be gathered otherwise.

Sentences with syntactic movement out of sentential complements of manner-of-speaking (MoS) verbs (e.g. whisper, shout) are degraded in acceptability, an effect called the manner-of-speaking (MoS) island effect. Accounts variably attribute the MoS island effect to the violation of the subjacency condition, to the low frequency of MoS verbs taking sentential complements, or to a general information-structural constraint that discourse-backgrounded constituents cannot be extracted. In five acceptability judgment experiments, we find that the MoS island effect can be modulated by foregrounding or backgrounding the extracted constituent, suggesting a causal relationship between discourse backgroundedness and this effect. Our findings challenge syntactic and frequency accounts of the MoS island effect.

Phrases in a number of syntactic contexts are required, in a variety of languages, to end in their heads. This article offers a unified theory of the relevant properties of these contexts and of why the phenomena in question, while widespread, are not completely universal. The theory makes use of proposals made independently in CONTIGUITY THEORY (Richards 2010, 2016): the relevant syntactic contexts are argued to involve a prosodically dependent element that must attach prosodically to the head of the phrase to its immediate left, and this attachment is often blocked if the phrase in question is not head-final.

While it is known that speakers tend to use more reduced forms for more predictable words or phrases, it is unclear whether the same happens at the referential level: the influence of referent predictability on pronoun production remains a contentious issue, with divergent findings reported in the literature. To address this inconsistency, we carried out a Bayesian meta-analysis of the current literature on the relationship between referent predictability and pronoun production. Our meta-analysis covers twenty primary peer-reviewed studies, encompassing twenty-six experiments across eight languages. We find stronger evidence for a small positive effect of referent predictability on pronoun usage, as opposed to the alternative hypothesis of no effect or a negative effect. As the first comprehensive synthesis of available evidence on this topic, our study offers insights into pronoun production and identifies promising avenues for future research: focus on typologically diverse languages, on conditions where a variety of referring expressions are expected or where the effect of predictability is more likely to appear, and others. Finally, we also advocate for the use of meta-analysis as a tool for theoretical linguistics.

The German gender system is known for its complexity, and there is a persistent misconception that it is largely arbitrary, and hence a challenge for the typology of gender systems. In response, we construct a database of more than 30,000 German nouns and show that a boosting tree model achieves a predictive success of 96%. Even more surprising, the model performs at 87% when trained on just the 100 most frequent nouns. We thus demonstrate that the complex German system fits into a typologically well-known scheme, being a combination of semantic and formal assignment principles. In addition to our success with the specific problem, we show the value of statistical modeling for typologists and reflect on what exactly we can learn from these techniques.

A central task in sound change research is to account for the emergence of new phonological categories from patterns of synchronic phonetic variation. To this end, a comparison was made of two English dialects differing in their phonologization of coarticulatory vowel nasalization using real-time magnetic resonance imaging of the velum and tongue and their synchronization with the glottal signal. This was done for coda vowel-nasal (VN) sequences preceding voiced and voiceless stops. The results showed that a later phasing of N’s oral gesture and an earlier onset of aperiodicity in the nasal consonant were two of the main physiological factors likely to lead to sound change in voiceless VNC̥ clusters. They were also consistent with models of compensatory lengthening by which the vowel is lengthened and the oral gesture of the nasal consonant is shortened as sound change progresses.

In many Malayic languages (western Austronesian), subjects can undergo A’-movement (relativization, WH-movement, focus movement, and topicalization) without any special qualification, whereas nonsubject nominals can A’-move only if the verb appears without a voice prefix. We propose a novel account for the syntax of voice alternations in Malayic languages, synthesizing insights from the prior proposals of Aldridge 2008b and Nomoto 2015, 2021. Adopting the view that the predicate constitutes a phase (Chomsky 2000, 2001), with word order determined at each phase level (Fox & Pesetsky 2005), our proposal derives both the general subject-only A’-extraction restriction and its limited exception and associated morphological restrictions. The proposal is motivated by our original data on voice and extraction in Suak Mansi Desa, a previously undescribed Malayic language of western Borneo, which we then extend to Standard Indonesian and Standard Malay, as well as various other dialects and languages of the region.

This article examines the English phrase-as-lemma (PAL) construction, which treats phrases syntactically as if they were words (e.g. a don't-mess-with-me driver). We argue that it is important to acknowledge and represent the construction's unique syntax directly rather than trying to shoehorn it into a more familiar grammatical category such as Noun or Adjective. PALs do not share the same distribution as other categories, and critically, their unique syntax influences their interpretation in predictable ways, which we demonstrate with survey data (N = 685). In particular, PALs convey the type of meaning associated with individual English words—lemmas—and thus evoke semantic frames that are presumed shared common knowledge. We further predict that the shared common knowledge and the use of quotes encourages PALs to be interpreted as witty and sarcastic. We show that a full analysis of PALs requires a family of constructions that includes certain conventional instances and productive subtypes. Because the construction's special form and function are intimately related, we predict that comparable PAL constructions should appear in other, unrelated languages. While the PAL construction is not terribly frequent in any language, the implications we draw are quite broad: our knowledge of language is rich and complex, providing subtle means for language users to indicate familiarity with listeners while conveying their message.

Knowledge of the mechanisms of phonetically motivated sound changes is incomplete and based primarily on apparent—rather than real-time—evidence. How exactly does an innovative variant become the new norm within individual and population-level grammars? Here, eighteen L1 English speakers were recorded in 2014 and again in 2022. Acoustic analysis of sibilants in /str/ shows (i) coarticulatory /s/-retraction for all speakers in 2014 and (ii) a group-level increase in the degree of /s/-retraction in 2022. Perception data show that increased /s/-retraction was audible to listeners. Increased /s/-retraction also matched the outcome of simulated interaction between the speaker participants in an agent-based model. However, the effects of sound change are not predictable within idiolects: some speakers’ production targets shifted incrementally in the opposite direction of the group-level trend.

In many languages, finite-clause-embedding verbs vary in whether they allow WH-dependencies to cross from the embedded to the matrix clause—a phenomenon we call ‘bridge effects’. Why bridge effects exist has been the subject of much debate; we argue that contributing to the lack of consensus are the relatively small samples of verbs (from twelve to seventy-five for English) previously tested in the literature. To resolve this issue, we report two new data sets of bridge effects covering a nearly exhaustive sample of 640 English verbs. We use these data sets to address three research questions: Are there bridge effects at all? How well do leading theories of bridge effects explain observed variation across the full range of verbs? And are there new patterns emerging from our data that could lead to a better theory? We ultimately argue in favor of a multivariate approach, drawing upon existing ideas while including a novel morphosyntactic licensing component identified from our data. We also discuss implications for theories of locality and explore how context might affect the acceptability of WH-dependencies.

Cross-category harmony is one of the most well-known typological universals. It describes a trend of consistent alignment of different syntactic categories across phrases within a language. Explanations for this universal vary as to whether cognitive factors play a role or the tendency is instead due to mechanisms of language change alone. In this article we report a series of artificial language learning experiments that aim to test a hypothesized link between cognition and cross-category harmony. As with the typological tendency itself, we find mixed evidence for harmony across different types of phrases. Specifically, learners are biased in favor of consistent alignment of the verb in the verb phrase and the adposition in the adpositional phrase. However, the bias for consistent alignment of the verb in the verb phrase and the adjective in the noun phrase depends on the semantic similarity between adjectives and verbs. When adjectives are active and therefore more verb-like (e.g. broken), we find harmony; when they are stative and therefore less verb-like (e.g. blue), we do not. These results suggest that the bias for cross-category harmony is not purely based on the syntactic notions of head and dependent, but reflects the interaction between a general cognitive bias favoring consistent order and cross-category similarity.

This article documents an evolutionary pathway from parataxis via the correlative construction to finite subordination in the Ugric languages. The first step of the emergence of finite subordination is the appearance of a paratactic precorrelative sentence pair, with an indefinite pronoun in the first sentence that is anaphorically resumed in the second sentence. This pattern, witnessed in the SOV Ob-Ugric languages, developed into a full-fledged correlative construction in Hungarian by the end of the twelfth century. In Hungarian, drifting to head-initial grammar, the correlative construction is shown to have been both the source of the evolution of finite relative clauses and the source of the grammaticalization of finite complement clauses. The path to finite relativization involved the reversal of clause order and the reanalysis of the relative clause as an adjunct of its main-clause correlate. Finite complementation evolved in the context of say-verbs by the integration of their paratactic propositional complement with the use of the correlative pattern of subordination. The relative pronoun was recategorized as a complementizer. The complement clause, first construed as an adjunct of the main clause, later came to be subordinated to the VP, eliciting verbal agreement when functioning as an object. The developmental path pointed out in the Ugric languages bears on theoretical debates concerning the status of parataxis and of nonfinite and finite subordination in the synchrony and diachrony of human language, and provides typological parallels for the interpretation of controversial data from the early history of Indo-European languages.

This study considers how lexical frequency effects and collocations help shape a case of language change in progress. The case of change involves the sentential negative markers point and pas in Acadian French. While point has been lost from most contemporary spoken varieties of French, it has been retained in conservative varieties of Acadian French. Quantitative analyses of one such variety show that point is replacing the pas variant, with the youngest speakers making almost exclusive use of point. Furthermore, the analysis shows how lexical frequency effects and collocations help us understand the evolution of constraints operating on the variable.

In French, the members of a subclass of anticausative verbs are optionally marked with the clitic se, traditionally considered to be a reflexive marker. We show that this optionality is not a case of free variation. Rather, the presence or absence of se is influenced by lexical-pragmatic considerations: while by default, both variants are equally acceptable, in the context of a human subject, cooperative speakers strongly prefer the variant that in certain cases avoids and in other cases maintains ambiguity with the semantically reflexive interpretation that arises in parallel with the intended (anticausative) interpretation. Understanding these preferences requires taking into account the agent bias—that is, the tendency to interpret human nouns in subject position as agents whenever possible—and the multifunctionality of se, which is used to form both (nonagentive) anticausative predicates and (agentive) semantically reflexive ones. The preference for the presence vs. absence of se is predicted based on whether the alternative reflexive interpretation is in line with shared assumptions about the event. The interaction between the choice of form by the cooperative language user and individual verb subclasses is an example of what we call lexical-pragmatic effects.

A few days ago I found myself back at my old alma mater, Stanford University, to celebrate the fiftieth anniversary of the founding of the Department of Linguistics. Much was familiar to me since I left nineteen years ago—the building, the offices, a few of the faculty, the random musical equipment lying around—but there were some changes. The most notable was in the first floor break room: the built-in bookshelves that had long stood at the end of the room had been removed, replaced by a standing table and bar seating. It was a small matter of remodeling, and to be fair the contents of the bookshelf had mostly been moved around the corner into the copy room. But it was a bit of a shock to see, because I remember so vividly the wall full of old journals that adorned those shelves. There stood more peer-reviewed, scholarly studies in linguistics than I had previously seen amassed in one place without a trip to the library, all accessible at my fingertips to just flip through any time I stepped out of my office. It was an inspiring sight to make a young graduate student feel like they were part of something bigger, and much older, than whatever they were working on at the time. I sometimes pulled a random journal off the shelf just to see what was in it, part of a larger goal I had of somehow expanding my mind beyond the barriers of my own work. At that stage of my career I cannot say I understood everything or even much of what I read.

Like most people who have done linguistic field work for thirty years or so, I have worked on languages which are now extinct, eight of them in my case, and I have studied, and continue to study, many languages which are seriously imperiled. My experience is far from unusual, and the testimony of field workers alone would amply illustrate the extent of language loss in the world of the present era.

The organization of taking turns to talk is fundamental to conversation, as well as to other speech-exchange systems. A model for the turn-taking organization for conversation is proposed, and is examined for its compatibility with a list of grossly observable facts about conversation. The results of the examination suggest that, at least, a model for turn-taking in conversation will be characterized as locally managed, party-administered, interactionally controlled, and sensitive to recipient design. Several general consequences of the model are explicated, and contrasts are sketched with turn-taking organizations for other speech-exchange systems.

Transitivity involves a number of components, only one of which is the presence of an object of the verb. These components are all concerned with the effectiveness with which an action takes place, e.g., the punctuality and telicity of the verb, the conscious activity of the agent, and the referentiality and degree of affectedness of the object. These components co-vary with one another in language after language, which suggests that Transitivity is a central property of language use. The grammatical and semantic prominence of Transitivity is shown to derive from its characteristic discourse function: high Transitivity is correlated with foregrounding, and low Transitivity with backgrounding.