Paper

Non-Agreeing Degree (NAD) Constructions have the distinguishing property that the agreement pattern between subjects and degree predicates is optionally disrupted, even in languages (like Spanish) where verbs commonly agree with their subjects. We show that the agreeing vs. non-agreeing alternation comes with important semantic differences for the interpretation of the degree construction. We provide a first systematic description of this type of constructions and postulate a formal syntactic and semantic analysis. We argue that NAD constructions are characterized by degree predicates that introduce a non-conventional nominal scale and by subjects that denote equally non-conventional units of measurement. We postulate an intensionalization process on the subject of NAD constructions that we capture via a general nominalization function.

Paper      Preprint

This paper presents three new arguments that the matrix subject in English tough- constructions cannot reconstruct into the embedded gap. The first two arguments reexamine data in the literature purported to show such reconstruction. Upon closer scrutiny, we argue that these data in fact involve short reconstruction below a modal or generic operator in the matrix clause, and not genuine long reconstruction into the embedded gap. The third argument is that property positions, which independently require reconstruction, are unable to host tough-gaps. This ban on long reconstruction in English tough-constructions follows without further ado on a base-generation analysis.

Paper      Preprint

This paper provides an investigation of Ignorance Inferences by looking at the superlative modifier at least. The formal properties of these inferences are characterized in terms of the epistemic conditions that they impose on the speaker, thereby establishing how much can and must be inferred about what the speaker is ignorant about. The paper makes two main contributions. First, it argues that the form of these inferences depends solely on the structural properties of the expression that at least is modifying, which do not necessarily coincide with semantic entailment: with totally ordered associates, at least triggers Ignorance Inferences that may be formally different than those obtained with partially ordered associates (see Mendia 2016a). Second, it builds on neo-Gricean double alternative generation mechanisms (e.g. Schwarz 2016) by showing that one of them must be provided by focus alternative semantics.

Paper      Preprint

Degree Neuter Relatives (DNRs) in Spanish are relative clauses formed by a gradable predicate of any sytactic category and the neuter definite determiner lo. The analysis proposed in this paper aims at achieving two things. First, it avoids certain well-known compositionality problems due to sortal mismatches between degrees and entities (see Mendia 2022; NELS52). Second, based on this analysis, it is suggested that the cross-linguistic rarity of DNRs can be explained by specific aspects of the morphological inventory of Spanish allowing the construction of degree denoting Free Relatives as well as Free Relatives formed by a definite article in this language.

Paper     

This snippet points out that the oddness of sentences like some dogs have four legs does not receive a straightforward explanation in terms of pragmatic competition with a more informative alternative statement, precisely because of the difficulty to identify what the competiting more informative alternative statement should be.

Paper      Preprint

Spanish comparatives have two morphemes that can introduce the standard of comparison: the complementizer "que" ("that") and the preposition "de" ("of"/"from"). This paper defends the idea that comparatives introduced by "de" always express overt comparison to a degree (type d). On this view, a simple analysis where everything is interpreted as it surfaces suffices to account for all the properties of de comparatives. On the other hand, comparatives introduced by the standard que are compatible with a wider range of constructions, suggesting they involve additional covert structure. These findings in turn point to an unnoticed locus of cross-linguistic variation with respect to the role of the standard morpheme.

Paper      Preprint

Although there is no consensus about what kinds are, there is a common understanding that kinds can be regarded as collections of objects that share certain properties. What these properties exactly are is often left unspecified. This paper explores the semantics of ad hoc kind-referring terms, where the determination of the relevant set of shared properties does not rely on “natural” properties or world knowledge. Rather, information provided by a nominal modifier, typically a relative clause, is used to impute the required regular behavior on the kind-referring NP. Building on Carlson’s (1977) disjointness condition, I show that we can not only account for the ubiquity of these expressions, but we can also extend the analysis to other constructions that have traditionally not been taken to be kind-referring, such as Amount and Degree Relative constructions.

Paper

This paper was writen in 2009/2010 as one of my MA qualifying papers. It provides evidence that the so-called Taub's Generalization (Taub 1989) holds not only for a variety of QPs, but also in a variety of languages. In the light of these facts, the paper argues against a cover-based semantics for all (pace Brisson 1998, 2003).

Preprint

In this paper we examine a number of constructions that show lack of agreement between the subject and either the verb or some constituent within the predicate. We focus on Non-Agreeing Degree constructions (Mendia and Espinal 2024), which we compare with so-called Pancake Sentences (Enger 2004, Wechsler 2011, Haugen and Enger 2019) and Topic Categorical Sentences (Britto 2000). All three constructions show a non-standard agreement pattern that nevertheless signals some form of semantic shift in interpretation. We argue that, despite surface similarities, these constructions do not belong to the same category..

Preprint

Expressions of sufficiency like enough, suffice, etc. have long attracted attention due to their complex syntactic and semantic properties. Two of the common questions concerning such constructions include whether they obligatorily require an infinitival complement, or whether sufficiency predicates are intrinsically modal. In this paper we contribute to this body of work by studying the subjects of predicates of sufficiency, a hitherto unnoticed source of variation in constructions of sufficiency. Focusing mainly on Spanish data, we show that definite and quantificational expressions that are otherwise ungrammatical subjects of certain sufficiency predicates may nevertheless be "rescued" by the preposition con 'with', thereby revealing that the class of expressions that may appear in subject position of sufficiency predicates is rather heterogeneous.

Preprint

The so-called satisfaction theories of presupposition propose weak presuppositions that are then pragmatically strengthened under to-be-determined conditions. The opposite view, first pioneered by Gazdar (1979) and van der Sandt (1988), contends instead that presuppositions are semantically strong but can nevertheless be canceled under to-be-determined conditions. This paper explores two ideas: (i) one such case of cancellation can be restated in terms of pragmatic weaking, in the sense that presuppositions do not project if, in doing so, the speaker would declare that they hold an inconsistent epistemic state; (ii) presuppositions that fail to be inherited wholesale by the sentence may nevertheless project conditionally, as suggested by satisfaction theory, provided certain contextual conditions are met.

Preprint

Spanish allows an unusual relative clause construction involving a relative clause seemingly headed by a gradable predicate and the neuter determiner lo. Previous analyses have put forth the idea that neuter relatives denote maximal degrees, but while intuitive, straightforward implementations of the idea run into trouble when combining the resulting definite degree (type d) with verbal predicates that do not take d-type arguments. This paper proposes an analysis wherein the DP containing the relative clause is of predicative type, but contains within it a maximal degree that serves as a Measure Phrase within an Adjective Phrase.

Paper

Epistemic numbers are syntactically complex numbers formed by some combination of number and (non-numerical) indefinite expressions that convey uncertainty, vagueness or ignorance. I examine these constructions in a number of languages and provide an analysis that capitalizes on independently motivated properties of their two sub-components: epistemic indefinites and complex numerals. I argue that variability in these two components explain the cross-linguistics variability in the nature and interpretation of epistemic numbers across languages.

Paper

Under certain circumstances, speakers are subject to so-called spurious NPI licensing effects, whereby they perceive that NPIs without a c–commanding licensor are in fact licensed and grammatical. Previous studies have all involved the presence of a licensor in a position that linearly precedes, but does not c–command the NPI. In this paper, we show that spurious NPI licensing can occur in the outright absence of a licensor, in contexts that force an exhaustive parse. We reason that at least these instances of spurious NPI licensing might be reduced to the EXH operator pragmatically “rescuing” the NPI, in the sense of Giannakidou (1998, 2006).

Preprint

Emphatic Relative Constructions in Spanish (ERCs) are constructions that superficially resemble definite DPs modified by a restrictive relative clause. However, despite their surface appearance, they differ from these in two important respects: (i) they can be embedded under a great variety of wh-embedding predicates. This paper argues that despite their surface appearance, ERCs are subordinate questions: syntactically, they involve an interrogative core that combines with a definite determiner; semantically, the interrogative core yields a question meaning, and the definite determiner returns the most informative proposition from the Hamblin-set delivered by the CP (e.g. Dayal 1996).

Paper

Amount Relatives (ARs) differ from restrictive relative clauses in that they do not refer to a particular object denoted by the head of the relative clause, but to an amount of such objects (Carlson 1977a, Heim 1987). Traditionally, ARs have been regarded as degree expressions. In this paper I argue against this view and propose instead that amount interpretations of relative clauses are in fact a special case of kind interpretation.

Paper

In this short paper I discuss a particular construction of Spanish where numeral expressions (like "twenty-three") contain some kind of vague quantifier (like 'some', 'many' or 'few'; cf. 'twenty-some' in English). I present an sketch of a general system to interpret cardinal numbers flexible enough to account for the variety of quantifiers that can appear in a variety of languages.

Paper      Preprint

It has long been noted that Superlative Modifiers ("at least" and "at most") can be associated with focus (e.g., Krifka 1999). By using a number of diagnostics from Beaver and Clark (2008) and others, this paper argues that SMs have a lexically-encoded, conventionalized dependency on focus. In this respect, they are unlike other superlatives ("−est"), whose Association with Focus is Free. These claims are substantiated by extensive data from a variety of unrelated languages. The results lend support to Beaver & Clark's (2008) claim that different kinds of focus-sensitive elements interact with the meaning of focus in different ways (i.e., the disctinction between “conventional” vs. "free" association with focus).

Paper

Experimental evidence is presented showing that when "at least" takes a complement that constitutes a Total Order (like numerals), the exhaustive interpretation of the prejacent must obligatorily constitute an epistemic possibility for the speaker. In contrast, when the complement is Partially Ordered, the exhaustive interpretation of the prejacent can but need not be a live possibility for the speaker. These facts are taken to suppor previous neo-Gricean accounts of the Ignorance Inferences that Superlative Modifiers come with (see the paper Structural Effects on Implicature Calculation below).

Preprint

In this paper it is argued that computing implicatures with Superlative Modifiers SMs like at least and at most requires to factor in both Horn Set alternatives and focus alternatives. The account formulates a classical neo–Gricean implementation elaborating on the following ideas: that SMs are conventionally associated with focus (Krifka 1999), and that the alculation of implicatures with SMs requires two sets of alternatives (Mayr 2013, Schwarz 2016). In addition, I propose in the paper that the set of alternatives relevant for the Gricean calculus is provided by two independent mechanisms: the familiar substitution method within elements of a Horn Set (Horn 1972, Sauerland 2004) and Association with Focus, whereby a set of alternatives is obtained by replacing the focus–bearing constituent with contextually relevant alternatives (Rooth 1985; cf. Fox and Katzir 2011).

Paper

This paper suggests that non–atomic distributive interpretations of plural quantified NPs are better explained by means of some notion of “typicality”. Sentences of the form all the X are P are predicted to have a non–atomic reading only if X can form a typical group. I suggest that typical groups are conceptually salient sub–pluralities of atomic individuals that, most importantly, are claimed to be relevant for the grammar. Two consequences follow: (i) quantifier–determiners have access to typical groups and (ii) non–atomic distribution over typical groups is context–independent.

Paper

This paper investigates the inferences of ignorance that come with superlative numerals. I argues that these are better understood as primary implicatures that are derived in a neo–Gricean framework together with a basic epistemic logic. In doing so, the paper supports the view of superlative numerals first advanced by Büring (2007), where their ignorance inferences are equal to those of a disjunctive expression, of the form exactly n or more/less than n.

Manuscript

Basque displays what is known as the Strong Person Case Constraint (PCC): an absolutive DP may generally not be 1st or 2nd person if it is c-commanded by a dative DP. We make the novel observation that this restriction is obviated under verbal ellipsis, even if this ellipsis does not affect the DPs whose cooccurrence is normally ruled out. We argue that this generalization indicates that the PCC arises from properties of the verbal agreement, not of the DP arguments. This indicates that a comprehensive account of the Basque PCC must be sensitive to both narrow-syntactic and PF properties (in particular whether or not the verb agreement is pronounced). We then develop an account of the Basque PCC based on Coon and Keine’s (2021) feature-gluttony proposal, whereby the PCC results from an irresolvable conflict that arises in the morphological realization of a probe that has agreed with two DPs.

Manuscript

In Spanish, certain DPs that superficially resemble definite descriptions modified by a relative clause show the syntactic and semantic behavior of interrogatives. I show that unlike concealed questions, these constructions possess full interrogative syntax. I propose a syntactic and semantic analysis of these constructions on which they are garden-variety wh-constructions that are selected by a D-head that is argued to be the overt exponent of an Answer-hood operator à la Dayal (1996).

Handout

In this paper we address the question of whether characterizing generic statements, such as Birds fly and Sue smokes can be given a unified analysis. We argue for a new point of view, namely the stance that a cognitive agent A has on the exceptions to the generically predicated property P. We argue that A’s stance on exceptions to P systematically correlates (i) with different types of charaterizing generics (Rules & Regularions vs. Inductive; Carlson 1995), and (ii) with different natural language expressions (Q-Adverbs, affixes, i.a.), signaling the type of characterizing generic statement it expresses. Our main evidence comes from the Czech verb suffix -VA-: we argue that -VA- is a morphological marker of inductive generalizations, and, due to its specific lexical semantic properties, it only allows two epistemic attitudes wrt. to exceptions: either A knows for certain that a generalization P has exceptions (hence is not normative), or A is ignorant about whether P has no exceptions. If this hypothesis is right, then this would seem to suggest that we need recognize different subtypes of characterizing statements, providing evidence against a unified treatment (with Krifka 2009, Pelletier 2010, i.a.).

Poster

This talk provides a first approximation to the semantics of the so-called habitual periphrastic expression in Spanish soler + INF. It is argued that soler is a dedicated marker of habituality (not genercity or frequentative in general). To be felicitous, soler requires a plurality of certain minimal size, be it a plurality of situations/events or a plurality of individuals. However, the verb soler itself is argued not to contribute any quantificational force of its own, and thus such constructions must be bound by other Q-expressions and are always evaluated wrt. the actual world.

Slides

In some contexts DPs have the ability to refer outside the nominal domain, denoting instead amounts or degrees rather than individuals. For instance, four pizzas is enough. Examples like these have been used to argue in favor of a so-called degree/individual polysemy, where the locus of this variation lies in the dual senses of the nominals themselves (Brasoveanu 2009, Rett 2014). This talk explores the syntactic and semantic properties of these DPs, arguing against a polysemy analysis of such cases.

Handout

This handout discusses some data that calls into question the notion that the conditions governing the distribution of standard markers across languages is solely syntactic. Very much work in progress.

Handout

This handout discusses the main empirical issues posed by characterizing generic statements and advocates the impossibility of a uniform analysis. Empirical data supporting the grammatical reality of so-called weak (or inductive) generic statements is provided by discussing the behavior of the Czech particle -va.

Slides

This paper investigates a puzzling pattern in Bangla whereby only concealed questions that denote identity questions about a number/degree/amount are grammatical in the language, while ordinary individual-referring extensions cannot form concealed questions. We provide a domain-restricted individual concept analysis of concealed questions whereby measure DPs are DPs whose denotation makes reference to a nominalized degree, a point in an ordering of sets of entities. Bangla is argued to have a much more restricted type-shifter than its very well-studied English counterpart, and only measure DPs satisfy the restriction. This analysis effectively establishes a categorical difference between ordinary and restricted concealed questions.

Slides

This talk suggests that a key aspect in providing a semantics of characterizing genericity is to factor in the point of view of the speaker's stance on exceptions to the generically predicated property. We argue that the speaker's stance on exceptions systematically correlates (i) with different types of generalizations, and (ii) with different natural language expressions (Q-Adverbs, affixes, i.a.) signaling the type of generalization a sentence expresses.