Nunberg et al. (1994) classified idioms into two groups: idiomatically combining expressions (ICEs), “whose meanings—while conventional—are distributed among their parts” and idiomatic phrases (IPs), “which do not distribute their meanings to their components” (p. 491). This distinction generally correlates with morphosyntactic (in)flexibility in idioms. However, Espinal & Mateu (2010) and others have argued that a sharp ICE/IP distinction is questionable. Here, I highlight a different sort of problematic example for a neat ICE/IP distinction: some putative IPs such as to chew the rag have yielded compounds such as rag(-)chewing, even though the meanings of their wholes do not distribute to their parts. I argue that existing analyses of unexpected flexibility of IPs fail to extend to compounding, formulate a preliminary account using the notions of morphosemantic alignment (Bonami & Strnadová 2019) and (morphological) motivation (e.g., Booij 2017) from morphological theory, and discuss the implications for the morphosyntax/ semantics interface.
This paper investigates the grammaticalization of the numeral ‘one’ in Slavic languages at the syntax-semantics interface. While extensively documented in Romance and Germanic, Slavic has received limited attention despite exhibiting varying grammaticalization degrees. We propose a unified analysis combining three theoretical frameworks: a semantic account based on bleaching, capturing the progressive loss of cardinality and the shift from referentiality to non-referentiality; a syntactic analysis tracing the movement of ‘one’ from NumP to SpecDP and D, correlating with the interpretive stages in semantics; and a nanosyntactic representation that explains how a single form functions across these positions through containment relations. This unified approach reconciles the formal identity of ‘one’ across its uses with its changing semantic and syntactic properties, contributing to grammaticalization theory and cross-linguistic typology of indefinite marking.
In this paper, I discuss Polarity Items (PIs) occurring in superlative contexts (ever-words) and their evolution in Catalan and Portuguese. First, I discuss these items in these languages: mai in Catalan, jamais (as compared to nunca ‘never’) in Portuguese, and the more recently use of já in Brazilian Portuguese and de sempre in European Portuguese. Following that, I present the diachronic evolution that these expressions underwent as reported in the literature: some items disappeared in superlative contexts, others developed homophony and are used in semantic and in syntactic dependencies, and the more recent ones appear only as dependent on the semantic context of superlatives. Finally, I comment on the contribution that these data may bring to the debate on the directionality of change.
This article presents a case study on two Spanish modifiers, la verdad and de verdad, which share the lexical base verdad ‘truth’ but demonstrate distinct distributions and functions. We argue that these modifiers operate on different layers of the speech act structure, adopting the recent model by Krifka (2023, 2024). Corpus data show la verdad is less frequent, restricted to peripheral positions, and exclusively modifies assertions. De verdad is more frequent, syntactically mobile, and appears in various speech acts (assertions, questions, exclamations). We propose that de verdad is a modifier of the commitment operator, specifying the commitment event as truthful, resulting in a strengthening effect, while la verdad is a modifier of the assertion operator, specifying the act of asserting as truthful, which often relates to politeness. This distinction in the proposed analysis accounts for their differing distributional profiles and semantic effects.
Taxonomic nouns like kind, type, and sort appear to refer to abstract entities, rather than particular objects. Focusing on the Romanian tip ‘type’, we provide evidence based on agreement patterns in favor of Wilkinson’s (1995) hypothesis that the two interpretations—taxonomic and existential— of taxonomic nouns correspond to distinct syntactic structures. We propose a formal analysis of existential readings employing a pseudo-partitive syntax with an intransitive-predicate denotation for the taxonomic noun. Our analysis clarifies the distinction between genuine taxonomic Ns, including tip ‘type’, which refer to abstract entities, and the Romanian noun fel ‘kind’, which seems unable to do so. We show that despite lexical variety, a unified treatment of taxonomic expressions remains viable without multiplying abstract entity sorts beyond Carlsonian kinds. Our proposal accommodates the syntactic diagnostics while avoiding unnecessary ontological commitments.
