My Google Scholar page

Published Work

in progress: Explaining 'EARLIEST'
The semantics of degree constructions has motivated the implementation of a MAX operator, a function from a set of degrees to its maximal member (von Stechow 1985, Rullmann 1995, a.o.). This operator is unsatisfying: it's arbitrary (cf. MIN), and therefore not explanatory. There have thus been several calls to reduce MAX to a more pragmatic principle of maximal informativity (Dayal 1996, Beck & Rullmann 1999, Fox & Hackl 2007, von Fintel et al. 2014).

Intriguing differences between before and after have caused some to posit an EARLIEST operator in the temporal domain (Beaver & Condoravdi 2003, Condoravdi 2010). This operator is unsatisfying for similar reasons (cf. LATEST), and some have suggested it, too, can be redefined in terms of informativity (Rett 2015). However, recent cross-linguistic evidence (reported here) complicates the reduction of EARLIEST to 'maximize informativity': while counterparts of before and after across languages share many foundational semantic properties, they appear to differ in a principled way in how certain before constructions are interpreted. I discuss this and other related observations with respect to the future of a domain-general 'maximize informativity' program.

under review: The semantics of many, much, few, and little (not final)
The words many, much, few, and little (and their cross-linguistic counterparts) are quite unusual semantically. They have traditionally been characterized as quantifiers (like every) or adjectives (like tall); however, these analyses can only account for instances of these terms in which they encode information about an individual or set of individuals (as they do when they occur prenominally, in e.g. much traffic). Recent degree-semantic analyses (Rett 2007, Solt 2014) instead characterize the meaning of these words in terms of scales, or sets of degrees, which can account for their canonical uses as well as uses in which they don't appear to be ranging over individuals (as in their differential use, e.g. much taller than).

revised for Semantics & Pragmatics: The semantics of emotive markers and other illocutionary content (not final)
I coin the term 'emotive markers' to describe words like alas which encode not-at-issue information about the speaker's emotive attitude towards the content of the utterances they occur in. I argue that there are important differences emotive markers and encoders of canonical not-at-issue content, like the utterance modifier frankly or the evidential adverb apparently. In contrast to the latter, emotive markers are only compatible with declarative mood and can result in Moore's Paradox. I conclude that the contribution of emotive markers should thus be treated as 'illocutionary content', on par with the sincerity conditions encoded in illocutionary mood. I present a formal analysis of illocutionary content in which it differs from other not-at-issue content in restricting the speaker's Discourse Commitments (Gunlogson 2001) rather than the Common Ground. [handout]

2017 (with A. Brasoveanu): Evaluativity across adjective and construction types: an experimental study [Journal of Linguistics 1-67. doi:10.1017/S0022226717000123]
An adjectival construction is evaluative if and only if it conveys that the property associated with the adjective exceeds a relevant threshold. The questions of which adjectival constructions are evaluative and why have formed the foundation for semantic theories of these constructions and of adjectives themselves (Klein 1980, von Stechow 1984), although it’s been alleged that these theories are based on an incomplete picture of the phenomenon of evaluativity (Bierwisch 1989, Rett 2008a).

We present the first experimental tests of the scope and nature of evaluativity across adjectival constructions and adjective types. These studies confim that evaluativity is conditioned by adjective type (relative or absolute, Kennedy and McNally 2005) and is not restricted to the positive construction. However, they also show several new and surprising aspects of evaluativity: that it is perhaps better characterized as a gradable property than a binary one; that the ways in which relative and absolute adjectives differ in their evaluativity vary across construction; and that, contrary to standard intuitions, subjects are willing to attribute evaluativity to the subject position of comparative constructions like Sue is taller than Bill. We show that this last particularly surprising result reveals a lot about how subjects interpret contextually-sensitive constructions, and we discuss its consequences for experimental studies and semantic theories of adjectival evaluativity as well as context-sensitive phenomena more generally.

2016: On a shared property of deontic and epistemic modals (preprint draft) [Deontic Modality, eds. N. Charlow and M. Chrisman, Oxford University Press]
Epistemic modals encode an evidential restriction, requiring that the speaker have inferential evidence for the prejacent (Karttunen 1972). Stone (1994) and von Fintel and Gillies (2010) encode this restriction lexically in e.g. must, which (given a unified treatments of modals, Kratzer 1991) raises the question: what happens to this restriction when must receives a deontic interpretation?

I claim that both deontic and epistemic modals have in common a requirement that their prejacent be inferred from some premises (Glass 2013). I argue, following Lance and Little 2006, that this is a property of moral reasoning quite generally; in epistemic modal bases, it amounts to an inferential evidence requirement. Deontic and epistemic modals form a natural class with respect to this property to the exclusion of other modal bases; I argue that it is what prevents their acceptability in certain exclamatives (cf. *Wow, must Sue be the murderer!). It also offers insight into why languages like English sometimes lexicalize these two modal bases to the exclusion of others.

2015: Antonymy in space and other strictly-ordered domains [Perspectives on Spatial Cognition, vol. 10 of The Baltic International Yearbook of Cognition, Logic and Communication, eds. M. Glanzberg, J. Skilters, and P. Svenonius, pp 1-33]
Natural language references different types of entities. Some of these entities (e.g. degrees, locations, times) are strictly ordered with respect to one another; others (e.g. individuals, possible worlds) are not. The empirical goal of this paper is to show that some linguistically-encoded relations across these domains (e.g. under, slower than) display a polar asymmetry, while others do not. The theoretical goal of this paper is to argue that this asymmetry – and its restriction to only certain relations – is due to intrinsic properties of strictly-ordered domains, coupled with a bias in how language users perceive these domains. [handout]

2015 (with L. Winans, N. Hyams and L. Kalin): Children’s comprehension of syntactically encoded evidentiality [Proceedings of NELS 45, eds. Thuy Bui & Deniz Özyildiz, vol. 3 189–202]
This paper reports on a comprehension study of children's acquisition of syntactically encoded evidentiality (Rett et al. 2013). We demonstrate that children acquiring syntactically encoded evidentiality show no production delay relative to children acquiring morphologically encoded evidentiality (Ozturk & Papafragou 2008, de Villers et al. 2009). We argue that this delay cannot be due to any delay in the acquisition of evidentiality. We propose that the cross-linguistic production/comprehension asymmetry is due to a task effect; children's performance on felicity tasks involving evidentiality are delayed relative to their ability to make general felicity judgments (Dudley et al. 2013).

2015: The Semantics of Evaluativity [Oxford University Press]
It's generally assumed that the simplest form in which gradable adjectives are used - positive constructions, like John is tall - carry an additional meaning, evaluativity, that is not part of the adjective's lexicalized meaning. Evaluative constructions require that an entity instantiate a gradable predicate to a significantly high degree. This property holds of John is tall, but it fails to hold of other adjectival constructions, like John is taller than Bill or How tall is John?.

The source of this evaluativity has posed a challenge for semantic theories of adjectives and adjectival constructions. These accounts assume that evaluativity is contributed by a null morpheme, either POS (Cresswell, 1976, Kennedy 1999) or EVAL (Rett 2008). The uses of these null morphemes raise complicated questions for a compositional system: does the null morpheme have an overt counterpart? What is its distribution? Why does it have that meaning, and not some other meaning?

Rett's (2008) characterization of EVAL capitalizes on the notions of antonymy and markedness to account for the distribution of evaluativity. This book frames these notions in a neo-Gricean framework of conversational implicature (Horn 1984, Levinson 2000). In constructions, like equatives or degree questions, in which evaluativity is associated only with negative antonyms, evaluativity arises as a manner implicature associated with the use of a marked element. In positive constructions, evaluativity arises as an informativity-based quantity implicature, strengthening the otherwise trivial meaning of the construction. Based in part on observations in Bolinger (1972), I argue that evaluativity is the only available such implicature for degree constructions. The consequence of this proposal for theories of degree semantics is direct and substantial: it allows for a straightforward, compositional semantics of adjectival constructions and a unified treatment of evaluativity with other issues in the semantics of degree. [handout]

2014: Modified numerals and measure phrase equatives [Journal of Semantics 32, 425–475]
In English, equatives can be formed with a numeral or measure phrase (MP) standard (e.g. John can dive as deep as 500m; Rett 2010). These 'measure phrase equatives (MPEs)' differ semantically from their clausal counterparts (e.g. John can dive as deep as Sue can) in two important ways. First, while clausal standards set lower bounds, resulting in an 'at least' interpretation of the equative, MP standards tend to set upper bounds, resulting in an 'at most' interpretation. Second, MPEs are restricted in their distribution relative to clausal equatives: they are only acceptable when the subject is associated with a range of values or when the value they're associated with is significantly high (or both). The main goal of this paper is a unified analysis of the equative morpheme that accounts for these semantic and distributional differences between clausal and MP equatives. I attribute these differences to the fact that MPEs can trigger two different conversational implicatures: a quantity implicature (because they are less informative than MP comparatives); and a manner implicature, because they are more marked than MP constructions like John can dive 500m (deep). I end by suggesting an expansion of this account of MPEs to modified numerals generally, in particular to the differences between Class A and Class B modifiers (Geurts and Nouwen 2007, Nouwen 2010). [handout]

2014: The polysemy of measurement [Lingua 143, 242–266]
The goal of this paper is to gain a better understanding of measurement in natural language by investigating the possible interpretations of measurement DPs like four pizzas and many pizzas. I begin by observing a phenomenon I dub 'individual/degree polysemy': these phrases can denote either a plural individual or a degree corresponding to some measure of that individual. I then show that this polysemy is constrained in a particular way depending on whether or not the relevant measure is monotonic on the part-whole structure of the measured individual (Schwarzschild 2006). I provide an account of this polysemy and its monotonicity restriction by postulating a null measurement operator that can measure degrees as well as individuals (Cresswell 1976, Rett 2007). The paper concludes by drawing parallels to 'how many' ambiguities (Romero 1998), amount relatives (Grosu & Landman 1998) and quantity adjectives (Corver 1997).

2014 (with N. Hyams): The acquisition of syntactically encoded evidentiality [Language Acquisition 21:2, 173–198]
This paper presents several empirical studies of syntactically encoded evidentiality in English. The first part of our study consists of an adult online experiment that confirms claims in Asudeh & Toivonen (2012) that raised Perception Verb Similatives (PVSs) encode direct evidentiality. We then present the results of an acquisition study based on an exhaustive examination of the corpora of 45 American English-speaking children in the Childes database (McWhinney & Snow 1985). The result of this production study are consistent with the hypothesis that children as young as two behave like adults in their ability to correlate the syntax of these constructions with the type of evidence they have. We supplement this claim by directly comparing children's PVS utterances in these corpora to adult utterances in some of the same corpora. In addition to providing some insight into the kind of PVS input children receive, the results of our adult production study provide additional support for our claim that children's use of PVSs is adult-like. The production studies constitute a first step in an ongoing acquisition project which includes a comprehension study (a felicity judgment task) currently in progress.

2013 (with S. Murray): A semantic account of mirative evidentials [SALT 23: 453–472]
Indirect evidentials in Cheyenne and other languages can have a mirative interpretation in specific (extra-)linguistic contexts. We present a semantics for these mirative evidentials based on the three-tiered semantics for evidentials presented in Murray (2010) and the illocutionary analysis of exclamation in Rett (2011). [abstract] [handout]

2013 (with N. Hyams and L. Winans): The effects of syntax on the acquisition of evidentiality [BUCLD 37: 345–357]
Some authors have argued that children acquire indirect evidential markers later than direct evidential markers, but these studies involve languages in which the distinction is tied up with other semantic properties like aspect (Ozturk & Papafragou 2008, de Villiers et al. 2009) or tense (Papafragou et al. 2007). In this study we examine the phenomenon of English copy-raising -- in which a copy-raised construction indicates direct evidence, while a non-copy-raised construction is compatible with direct or indirect evidence (Asudeh & Toivonen 2012) -- in adult comprehension and child production to argue that children seem to be able to acquire the direct/indirect distinction quite early and that early attention to evidential source does not depend on having a language with obligatory, morphologically encoded evidentiality.

2013: Similatives and the degree arguments of verbs [Natural Language and Linguistic Theory 31(4): 1101–1137]
I begin with the observation in Haspelmath and Buchholz (1998) that languages tend to use the same morpheme to mark the standard of comparison across equation constructions. In English, it's the morpheme as, in similatives like John danced as Sue (did) and equatives like John is as tall as Sue (is). The first goal of this paper is to provide an analysis of as that accounts for its distribution across these constructions. The second goal of this paper is to provide an account of Haspelmath and Buchholz's second observation, which is that while languages can form equatives with parameter markers (the first as in John is as tall as Sue (is)), languages generally do not form similatives with parameter markers. I argue that Haspelmath and Buchholz's data suggest that equation constructions are a novel test for lexicalized argumenthood: in particular, that the equation of a non-lexicalized argument prohibits the presence of a PM, and, for English, vice-versa. This conclusion provides evidence that, contrary to recent claims (Piñón 2008, Bochnak forthcoming), verbs, unlike adjectives, do not lexicalize degree arguments.

2012: On modal subjectivity [UCLA WPL 16: 131–150]
Epistemic modals have long been observed to be ambiguous between a speaker-oriented interpretation (Lyon's 1977 'subjective' interpretation) and an interpretation that reflects general knowledge (the 'objective' interpretation). In recent work, Papafragou (2006) argues that subjectively interpreted epistemics are non-truth conditional, and that this results from them being indexical to the speaker and time of utterance. I argue against Papafragou's claim and end by presenting an alternative perspective of the facts based on a distinction between determinate and indeterminate modal bases.

2011: Exclamatives, degrees and speech acts [Linguistics & Philosophy 34(5): 411–442]
The goal of this paper is an account of the semantics and pragmatics of exclamation. I focus on two key observations: first, that sentence exclamations like Wow, John bakes delicious desserts! and exclamatives like What delicious desserts John bakes! express that a particular proposition has violated the speaker's expectations; and second, that exclamatives are semantically restricted in a way that sentence exclamations are not. In my account of these facts, I propose a characterization of illocutionary force of exclamation, a function from propositions to speech acts of exclamation. The difference in meaning between sentence exclamations and exclamatives has consequences for the type of violated expectation. I end with a comparison to some previous approaches and a tentative extension of parts of the analysis to other constructions.

2010: Equatives, measure phrases and NPIs [2009 Amsterdam Colloquium Proceedings: 364–373]
Standard semantic accounts of the equative ascribe it an 'at least' meaning, deriving an 'exactly' reading when necessary via scalar implicature. I argue for a particular formulation of this scalar implicature account which considers that (i) equatives license NPIs in their internal arguments, and (ii) equatives whose internal arguments are measure phrases (MPs) are, in contrast to clausal equatives, ambiguous between `at most' and 'exactly' interpretations. The analysis employs particular assumptions about MPs, scalar implicature and the notion of set complementation to enable 'at least' readings to be sensitive to the direction of a scale, thereby becoming `at most' readings in certain constructions. [abstract] [handout]

2009: A degree account of exclamatives [SALT 18: 601–618]
This paper calls for a unified semantic approach to exclamations formed from wh-clauses, inversion constructions, and definite DPs based on the observation that all three types of utterances are interpretively restricted in the same way. In particular, I argue that they are all subject to a degree restriction (they can only be used to exclaim the degree to which a property is instantiated) and an evaluativity restriction (which requires that degree be high relative to a contextual standard). To account for these restrictions, I propose that the strings used to form an exclamative denote degree properties (not propositions or sets of propositions, as others have proposed). [abstract] [handout]

2008: Degree Modification in Natural Language [my dissertation]
This dissertation is a study of the roles played by degree modifiers -- functions from sets of degrees to sets of degrees -- across different constructions and languages. The immediate goal of such a project is a better understanding of the distribution of these morphemes and how they contribute to the meaning of an expression. More broadly, a study of the semantics of degree modifiers is of interest because it helps demonstrate parallels between the degree and individual domains.

2007: Evaluativity and antonymy [SALT 17: 210–227]
In the semantics of adjectival constructions, the component of meaning which requires that a degree exceed a contextually-given standard ('evaluativity') is typically analyzed as an exclusive attribute of the positive construction. I argue that evaluativity has a much wider distribution, that of a degree modifier (type dt,dt). I then show how its distribution is constrained to account for the fact that it occurs differently with different antonyms (e.g. tall vs. short) and in different degree constructions (e.g. the comparative vs. the equative). [abstract] [handout]

2006: How many maximizes in the Balkan Sprachbund [SALT 16: 190–207]
In Romanian, Bulgarian and Macedonian, the monomorphemic word used to ask how many can optionally be combined with mult ('many'). Constructions with this bimorphemic phrase exhibit stronger maximality effects than those with the monomorphemic phrase. This indicates that mult works as a maximizer over sets of degrees, a fact explained by the assumption that many/mult is a predicate over sets of degrees (Schwarzschild 2002) rather than an individual quantifier (cf. Romero 1998, Hackl 2000). [abstract] [handout]

2006: Pronominal vs. determiner wh-words: evidence from the copy construction [CSSP 6: 355–374]
This paper presents an analysis of the wh-copy construction in some dialects of German and other languages that explains its similarity to extraction constructions while accounting for its incompatibility with wh+NP phrases. Essentially, wh-phrases without an NP complement can be copied because they're non-quantificational (introducing only a free variable into the derivation). Wh- phrases with an NP complement cannot be copied because these wh-phrases are quantificational, and interpreting them twice in the derivation leads to vacuous quantification. [handout]

2006 Context, compositionality and calamity [Mind & Language 21: 541–552]
This paper examines an attempt made in a series of articles (Stanley, 2002, et al.) to create a syntactic placeholder for contextual information. The initial shortcoming of Stanley's proposal is that it does not easily integrate these placeholders with domain-restricting information syntactically encoded elsewhere in the utterance. Thus, Stanley makes erroneous predictions in the case of sentences in which quantifier-restricting information encoded in (for example) a prepositional phrase conflicts with quantifier restriction valued by context is internally incoherent. I explore the space of possible solutions that are available to Stanley, demonstrating how each results in its own interpretation problem and, ultimately, fails. In doing so, I suggest that Stanley's syntactic approach to contextual restriction is untenable.

Unpublished Work

Handout, 2012: Mirativity across constructions and languages [CUSP 5, UCSD]
This paper discusses the phenomenon of mirativity across a variety of constructions. Mirativity is the expression of exceeded expectation. While the term was introduced to describe a secondary meaning of some evidentials (DeLancey 1997), it is also canonically associated with exclamations (Rett 2011) and morphemes like conjunctions (Malchukov 2004).

I will survey theories of these varied constructions and propose a unified description of the discourse effect of mirativity based on (as above) the notion of the expression of exceeded expectation. I will argue that the semantic contribution of mirativity is not-at-issue: it is backgrounded, unembeddable, invariably speaker oriented. And I will argue that a number of theories of related constructions (like evidentials and exclamatives) can do equally well in accounting for these properties. In particular, I argue that speech-act oriented theories like those proposed for evidentials in Faller (2002) and for exclamatives in Rett (2011) are compatible with dynamic accounts proposed for exclamatives in (Murray 2010, Koev 2011).

Ms., 2012: Group terms and the meaning of meet
I caution against Barker's (1992) conclusion that group terms like committee cannot be treated as denoting Linkian plurals. I argue that his only evidence for this claim is based on the mistaken identification of the verb meet as extensional.

Ms., 2005: "Different agreement morphemes for different agreement configurations: evidence from complementizer agreement in Germanic"
Many West Germanic languages exhibit complementizer agreement, a phenomenon in which the complementizer is inflected in a way covaries in phi-features with the subject. In some of these languages, the agreement morpheme on the complementizer differs from the one on the verb. Baker (2008) argues number and gender features can be checked at a distance under ccommand, while person features can only be checked under direct merge (which often, but not always, means that the goal and probe are in a spec-head relationship). I extend this analysis to complementizer agreement, which I argue accounts for 1) why the two morphemes differ phonologically (the verb marker reflects number, gender and person agreement while the complementizer marker reflects only number and gender agreement), and 2) why they have the distribution they do (the verb marker is assigned whenever the probe and goal are in a spec-head relationship; the complementizer marker is assigned whenever the probe merely c-commands the goal). [handout]

Ms., 2005: "Do Numeral Classifiers Influence Similarity Judgments?" (with M. Shatz)
In this paper, we compare speakers of Japanese (a classifier language) with speakers of English (a language without classifiers) to investigate the possibility that a person's knowledge of a language with numeral classifiers contributes to his/her judgments of the similarity of items. If speakers of Japanese judge items sharing a numeral classifier as more similar than do speakers of English, it would be evidence that classifier systems influence similarity judgments, and by inference, conceptual category assignment. We examine this question while addressing theoretical and methodological limitations in a previous classifier study. Our results provide the basis of a discussion of the factors implicated in similarity judgments, conceptual categories, and the nature of numeral classifier assignment.