Natural Language & Linguistic Theory最新文献

Universal quantifiers differ in whether they are restricted to distributive interpretations, like English every, or permit non-distributive interpretations, like English all. This interpretational difference is traditionally captured by positing two unrelated lexical entries for distributive and non-distributive quantification. But this lexical approach does not explain why distributivity correlates with number: cross-linguistically, distributive universal quantifiers typically take singular complements, while non-distributive quantifiers consistently take plural complements. We derive this correlation by proposing a single lexical meaning for the universal quantifier, which derives a non-distributive interpretation if the restrictor predicate is closed under sum, but a distributive interpretation if it is quantized. Support comes from languages in which the same lexical item expresses distributive or non-distributive quantification depending on the number of the complement. For languages like English that have different expressions for non-distributive and distributive quantification, we propose that the distributive forms contain an additional morphosyntactic element that is semantically restricted to combine with a predicate of atomic individuals. This is motivated by the fact that in several languages, the distributive form is structurally more complex than the non-distributive form and sometimes even contains it transparently. We further show that in such languages, there are empirical advantages to taking the choice between distributive and non-distributive quantifier forms to be driven by semantic properties of the restrictor predicate, rather than morphosyntactic number.

In many languages, reciprocal meanings are expressed either by grammatical means or by using lexical predicates. The grammatical strategy is productive and may involve derivational affixes (Swahili -an) or pronouns (English each other) with transitive forms, whereas lexical reciprocity is expressed by a restricted class of intransitive predicates like English kiss or meet. The situation is more complex in Romance languages, where reciprocal verbal constructions often require a se clitic without a clear separation between transitive and intransitive forms. Addressing this puzzle, we propose that Romance languages involve a grammatical/lexical distinction as in other languages. We show that numerous Romance constructions systematically allow se omission with certain reciprocals, exhibiting parallel properties to those of lexical intransitives in other languages. A similar observation is made in relation to the distinction between grammatical reflexivity (e.g., English oneself) and lexical reflexives (wash, shave). Furthermore, we show that the se requirement may also be relaxed with transitive verbs, when reciprocity or reflexivity is conveyed by an overt reciprocal/reflexive item (e.g., Spanish mutuamente 'mutually'). The emerging theoretical picture supports an analysis of se as a head projection that licenses arity-reduction, though language-specific conditions allow se omission when arity reduction is achieved by a lexical reciprocal item or by another overt reciprocal element.

Whether the lexical semantics of property concepts (words canonically expressed as adjectives in languages with that category; Dixon 1982, Thompson 1989) show variation is a matter of recent debate. At one end of the analytical spectrum, Francez & Koontz-Garboden (2017) contend that their meanings may vary in a way revealed by superficial morphosyntactic behavior. At the other end, Menon & Pancheva (2014) argue that they are universally built on abstract mass-denoting roots, a commonality that can be obscured by (covert) morphosyntax introducing possessive meaning. On the basis of differing strategies for property concept verb formation in Wá⋅šiw (isolate/Hokan, USA), we argue in this paper that there is evidence for variation in the lexical semantics of property concept roots, with some denoting predicates of individuals and others having abstract mass-type meanings, contrary to universalist assumptions. Crucially, the behavior of property concept verb formation in Wá⋅šiw lends itself to an analysis in which possessive semantics is implicated only when it is morphologically observable. By drawing an analogy to canonical possession in the language, we argue moreover that this extra morphology in property concept verbs is best understood as a light verb that both directly categorizes property concept roots and introduces a possessive semantics. These observations provide evidence for the claim that at least some variation in this domain is underpinned by variation in lexical semantics, and more generally for the idea that variation in the lexical semantics of open-class elements drives at least some variation in morphosyntax.

Manner/result polysemy describes a phenomenon where a single root can encode both manner and result meaning components of an eventive verbal predicate. It therefore poses a challenge to (i) the hypothesis of manner/result complementarity as a fundamental constraint on verb/root meaning and (ii) a strict one-to-one mapping between roots and meaning. Examining novel data from the Oceanic language Daakaka, I provide further evidence that polysemous verbs like tiwiye 'press manually, break' only apparently violate manner/result complementarity, as manner and result meaning components are in complementary distribution. As both meaning components are sensitive to their morphosyntactic environment, I develop an account of contextual root allosemy, in which manner and result interpretations are associated with designated syntactic positions in relative configuration to an event-introducing verbalizer v. In particular, I argue that a single root may be associated with two non-compositional entries in the encyclopaedia, an eventive and a stative one, which allows the root to be merged in either the manner or result position. Independent support comes from suppletive verb forms in the paradigm of polysemous roots in Daakaka, where the spell-out conditions of contextual allomorphy and contextual allosemy overlap. Finally, I discuss theoretical and empirical challenges for alternative accounts of manner/result polysemy, including accounts based on derivation, coercion, and homophony.

This paper provides new evidence that syntactic principles that are proposed to explain the (un)grammaticality of a sentence can also hold in sociolinguistic variation. In particular, we argue that two puzzling frequency patterns involving negation in French-the proximity effect on negative concord and the polarity effect on future temporal reference-are deeply related and are both derived from the sensitivity of syntactic agreement to "soft" locality constraints. Recent quantitative studies of future temporal reference reveal that, although all negative items are subject to the polarity effect in Laurentian French, pas does not give rise to the polarity effect in Parisian French. We argue that this dialectal difference can be explained by minor variations in the syntactic and semantic properties of the negative marker pas, given an appropriate analysis of the syntax of negative concord. Our paper therefore shows that incorporating sociolinguistic variation into syntactic theory helps refine our understanding of general syntactic principles, such as locality constraints, and argues that frequency/preference patterns should be included in the full theory of syntactic competence and performance of speakers.

Samoan deverbal nominalizations show a crosslinguistically rare tripartite-inactive alignment where unaccusative, unergative, and transitive subjects are distinguished by inalienable genitive, alienable genitive, and ergative case, respectively, with objects being marked like unaccusative subjects (Mosel 1992). In addition, subject clitics exhibit a marked unergative alignment, where only unergative subject clitics are distinctly marked by alienable genitive case, whereas all other arguments receive inalienable genitive case. In this study, we demonstrate that these alignments follow naturally from a language-specific combination of independently established phenomena, including (i) prepositional ergativity (Polinsky 2016), (ii) split (in)alienability (Myler 2016; Alexiadou 2003), (iii) split-intransitivity, (iv) the unaccusative restriction on nominalizations (Imanishi 2014; Alexiadou 2001), and (v) a nonuniform nature of clitic pronouns (Bleam 2000), and therefore provides novel evidence for each of these phenomena. Comparing the distribution of ergative case in nominalizations crosslinguistically, we argue that the source of ergativity varies across languages and suggest that the split between syntactic and morphological ergativity cannot be reduced to a category-split of ergative subjects.

Complementizer agreement in minority and nonstandard West Germanic languages is well-known and frequently studied, but there is little agreement on its analysis. In this paper, I add to this debate by presenting novel and underdiscussed data from Frisian and Limburgian on intervention effects: what happens to complementizer agreement when the complementizer and the subject are separated by an intervening element. In Frisian, intervention leads to ungrammaticality, and in Limburgian, it leads to the realization of complementizer agreement between the intervener and the subject. These effects cannot be accounted for by existing Agree and PF analyses of complementizer agreement. Instead, I argue that the complementizer agreement morpheme is a clitic. Adopting the approach to clitic doubling of van Craenenbroeck and van Koppen (2008), I develop an analysis of complementizer agreement as clitic doubling. The intervention effects in Frisian and Limburgian follow from an interplay of the structural size of the clitic and restrictions on movement. Specifically, the ungrammaticality of intervention in Frisian is the result of competition between the clitic and the intervener for the same structural position, and the subject-internal realization of complementizer agreement in Limburgian is the result of movement of the clitic below the intervener.

This work investigates the morphosyntax of nominal expressions in Standard Italian that have multiple adjectives in “split coordination,” which permits a plural noun to be modified by singular adjectives, for example le mani destra e sinistra (the hand.pl left.sg and right.sg). The proposal is (i) that these expressions are built from multidominant structures, with a constituent shared by the conjuncts, and (ii) that plural marking on the noun reflects “summative” feature resolution on the nP comparable to coordination resolution. This proposal captures various properties of split-coordinated expressions, including the availability of adjective stacking and of feature-mismatched conjuncts, as well as agreement with a class of nouns that “switch” gender in the plural. Taking agreement with resolving features to be a form of semantic agreement, which has been argued to be possible only in certain syntactic configurations (Smith 2015, 2017, 2021), the account captures prenominal-postnominal adjective asymmetries in split coordination. The work offers a coherent account of coordination and semantic agreement in the nominal domain, connects split coordination to related phenomena such as nominal right node raising and adjectival hydras, and, more broadly, evinces the unity of nominal and verbal agreement, pace analyses of nominal concord (Norris 2014).

This article discusses verbs of change that allow a formally transitive construal that, nevertheless, has anticausative semantics. Verbs forming such “transitive anticausatives” (e.g., The water raised its temperature) also form canonical anticausatives (cf. The temperature of the water rose). Such verbs differ from verbs that only form canonical anticausatives (cf. The water warmed) in that they do not lexicalize a fixed scale along which they measure change, so that the DP merged in the internal argument position of these verbs (a DP denoting a property concept like the temperature) can determine the actual scale of change. When these verbs form canonical anticausatives, the entity undergoing change along this scale is realized as the possessor of this internal argument DP. When these verbs form transitive anticausatives, the entity undergoing the change is realized in the verb’s canonical external argument position, where it is, however, not assigned any external argument role. Instead, as in the canonical anticausative variant, it is interpreted as the possessor of the internal argument DP. This possessive relation is overtly reflected in English and other languages where the subject of the transitive anticausative construal binds a possessive pronoun in the internal argument DP. After an illustration of the phenomenon in typologically different languages, the article lays out the above semantic properties of the transitive anticausative construal and the verbs occurring in it. It then subsumes transitive anticausatives under the theory of the causative alternation in Alexiadou et al. (2006, 2015) and Schäfer (2008). Particular attention is, thereby, given to the morphological marking that sets apart, in many languages, the lexical causative and the anticausative variant of (a subset of) alternating verbs (cf. English raise/rise). Transitive anticausatives show a theoretically challenging but informative behavior here. Even though the transitive anticausative construal expresses anticausative semantics, its verb necessarily features the morphological marking that is canonically associated with its lexical causative use. This suggests that the morphological difference often found between pairs of lexical causative and anticausative verbs is only indirectly related to causative and anticausative semantics but is ultimately determined by more abstract, syntactic properties.