Abstract Certain Amharic verbal constructions include a main verb and an auxiliary that are both inflected for subject agreement. However, in certain φ-feature combinations, one of the agreement markers is absent. We distinguish two patterns of agreement disappearance and develop an analysis of both on which full agreement in the syntax can be obscured during postsyntactic Vocabulary Insertion. To that end, we propose that Vocabulary Insertion be broken down into two ordered steps. First, Select chooses the best exponent to realize the features of a given node. Second, Link pairs the node and selected exponent. We associate the two types of agreement disappearance in Amharic with each of these steps: main verb agreement is realized as zero during Select via blocking, whereas auxiliary agreement goes unrealized due to a failure of Link. The failure of Link is driven by a novel anti-homophony constraint we dub Uniqueness. Finally, we show that our analysis accounts for cross-Ethiosemitic variation in the realization of agreement and is more empirically successful than alternatives.

Abstract We develop a stratal analysis of Spanish stress in which morphosyntactic conditioning and metrical nonuniformity are tightly constrained. Morphosyntactic conditioning is limited to cyclicity, the distinction between stems and affixes, and the affiliation of affixes to the stem or word level. Nonuniformity arises from lexical prespecification and interstratal constraint reranking. This framework yields several successful empirical predictions: Richness of the Base requires that prespecification be unpredictably distributed across all morph types; Chung’s Generalization demands identical metrical patterns in mono- and polycyclic stem-level constructions; and the Monomorpheme Generalization entails that monomorphemic and overtly affixed stem-level items also exhibit the same patterns.

We present an account of why disjoint reference effects obtain in verbal but not in adjectival passives. Passives in child language are independently argued to always be adjectival, which allows us to use a natural experiment in child grammar that is not available in the adult grammar: predicting the lack of a disjoint reference effect in even those passives that prima facie seem verbal. We conduct our discussion against the background of the difference between adjectival and verbal passives in general. Our account is based on (grammatical) Implicature Theory. Since the initiator in the semantic representation of adjectival passives stays at a kind level, it cannot introduce a discourse referent, hence cannot trigger a disjointness implicature, unlike the initiator in verbal passives (Gehrke 2013, 2015). We show in two experiments that children’s passives do not exhibit disjoint reference, unlike adults’ verbal passives, even though children have no trouble computing disjointness implicatures elsewhere. Our contribution thus confirms with a novel kind of evidence the syntactic nature of young children’s difficulty with verbal passives. It offers a new perspective on the external argument difference between verbal and adjectival passives based on Reinhart’s (2016) Theta System, while also offering additional evidence for a grammatical, rather than general pragmatic, theory of implicatures.

I argue that the well-known islandhood of adjunct prepositional phrases does not substantially derive from their adjuncthood. Instead, islandhood of these domains derives from various factors that are orthogonal to the argument/adjunct distinction, including PP-internal structure, lexical properties of prepositions, and semanticopragmatic construal. To show this, I demonstrate that PP islandhood crosscuts the argument/adjunct distinction. In particular, (a) PPs with NP complements are generally not islands, (b) PPs with tensed clausal complements are generally (strong) islands, and (c) PPs with gerundive complements are generally (weak) islands. These generalizations hold whether or not the relevant PP has a prototypical adjunct function.

Abstract Recent work (Branan and Erlewine 2022) propose an analysis of certain restrictions on Ā-extraction, and suggest that certain Ā-probes may be required to target the closest DP. Newman (2023) presents arguments against their analysis. This paper serves as a defense of Branan & Erlewine’s approach against Newman’s rejoinder.

Theoretical approaches to null subjects standardly focus on the narrow syntactic module. I offer a postsyntactic alternative in terms of PF deletion applying to salvage an unlinearizable derivation. Support for this alternative comes from microvariation in three closely related West Slavic pro-drop languages spoken primarily in Poland (Kashubian, Polish, and Silesian), in which the distribution of null subjects is sensitive to the postsyntactic realization of [PERSON/NUMBER] agreement. I argue that this microvariation indicates that in these languages at least, the licensing of null subjects takes place at PF, which in turn suggests that PF has a bigger role to play in a general theory of null arguments than has so far been assumed.

Bruening and Al Khalaf (2020) show that selectional violations in coordination are extremely limited (there are exactly two) and exactly match those that are permitted in ellipsis and displacement. Patejuk and Przepiórkowski (2023) criticize Bruening and Al Khalaf’s article on numerous fronts. They do successfully show that conjuncts do not need to match in syntactic category, but their dismissal of the selectional violation data does not succeed. I present additional data, including the results of three large-scale acceptability surveys, that show that the two violations of selectional restrictions are real and are fully general. The two patterns that need an analysis are coordinations of NP&CP appearing where CPs are banned, and Adv&AP appearing in prenominal position where adverbs are banned. I propose a variation on Bruening and Al Khalaf’s analysis that accounts for all of the facts and meets Patejuk and Przepiórkowski’s objections.

The typology of the logical vocabulary in natural language is highly skewed. In the domain of logical connectives, AND and OR are often lexicalized, lexicalizations of NOR are less common and tend to be structurally complex, and no other logical connective is ever lexicalized. Existing accounts fail to fully derive this major crosslinguistic pattern, and moreover resort to otherwise unwarranted assumptions. The goal of this article is to provide an account that is less stipulative and has wider empirical coverage than previous accounts, based on a novel notion of communicative stability. Using a model of a rational speaker, we observe that attested languages are stable languages—that is, languages in which the optimal message for a speaker to choose when they want to convey a particular state they are in is not affected by which states they take to be more likely and which ones less. We argue that Stability can explain both why AND and OR are the only simple connectives lexicalized and why NOR is the only complex connective lexicalized.

This article examines a puzzle pertaining to the distribution of covalued nominals in two understudied Mayan languages, Chuj and Ch’ol. While Ch’ol behaves as expected with regard to the binding conditions, Chuj appears to consistently tolerate violations of Condition C. The Chuj data thus cast doubt on the widely held view that the binding conditions reflect a universal property of human language (e.g., Grodzinsky and Reinhart 1993, Reuland 2011). I argue that the difference between Chuj and Ch’ol can be largely explained if, contrary to Ch’ol, Chuj exhibits “high-absolutive” syntax, independently proposed by Coon, Mateo Pedro, and Preminger (2014) to explain a constellation of morphosyntactic properties shared by a subset of Mayan languages. This syntax bleeds otherwise expected binding relations from the subject into the object, explaining the apparent violations. I further show that linear precedence plays a fundamental role in regulating the distribution of covalued nominals across Mayan, which I argue is due to a general ban on cataphoric “free” pronouns. Thus, the Mayan data not only are consistent with the binding conditions, but also provide further evidence in favor of a deep typological parameter within the Mayan language family.