Gradable adjectives denote properties that are relativized to contextual thresholds of application: how long an object must be in order to count as long in a context of utterance depends on what the threshold is in that context. But thresholds are variable across contexts and adjectives, and are in general uncertain. This leads to two questions about the meanings of gradable adjectives in particular contexts of utterance: what truth conditions are they understood to introduce, and what information are they taken to communicate? In this paper, we consider two kinds of answers to these questions, one from semantic theory, and one from Bayesian pragmatics, and assess them relative to human judgments about truth and communicated information. Our findings indicate that Bayesian accounts can model human judgments about what is communicated better than they can model human judgments about truth conditions, but the performance improves if the Bayesian approach is supplemented with the threshold conventions postulated by semantic

In this crowdsourced initiative, independent analysts used the same dataset to test two hypotheses regarding the effects of scientists' gender and professional status on verbosity during group meetings. Not only the analytic approach but also the operationalizations of key variables were left unconstrained and up to individual analysts. For instance, analysts could choose to operationalize status as job title, institutional ranking, citation counts, or some combination. To maximize transparency regarding the process by which analytic choices are made, the analysts used a platform we developed called DataExplained to justify both preferred and rejected analytic paths in real time. Analyses lacking sufficient detail, reproducible code, or with statistical errors were excluded, resulting in 29 analyses in the final sample. Researchers reported radically different analyses and dispersed empirical outcomes, in a number of cases obtaining significant effects in opposite directions for the same research question. A Boba multiverse analysis demonstrates that decisions about how to operationalize variables explain variability in outcomes above and beyond statistical choices (e.g., covariates). Subjective researcher decisions play a critical role in driving the reported empirical results, underscoring the need for open data, systematic robustness checks, and transparency regarding both analytic paths taken and not taken. Implications for organizations and leaders, whose decision making relies in part on scientific findings, consulting reports, and internal analyses by data scientists, are discussed.

We show that the interpretation of sentences like John is not very Adj depends on whether Adj is vague. We argue that this follows from a constraint on the interaction between vagueness and conversational implicature, a domain that has not been studied extensively. The constraint states that implicatures are not drawn if they lead to borderline contradictions (see Ripley 2011; Alxatib & Pelletier 2011; a.o.), a natural extension of the idea that implicatures should not contradict assertions (Hackl 2006; Fox 2007; a.o.). Experiment 1 establishes that not very Adj gives rise to the implicature Adj for the non-vague absolute adjective late, but not for the vague relative adjective tall (in the terminology of Kennedy & McNally 2005a). Experiment 2 generalizes this result to three relative adjectives in the positive form (tall, hot, fast), against those same adjectives in their (non-vague) comparative forms (taller/hotter/faster than the average X). We also constructed quantitative meaning representations for complex predicates of the form Adj very Adj, using fuzzy logic to model the contribution of boolean connectives and our experimental data to represent the meanings of adjectives. The results of these analyses suggest that strengthening not very Adj with Adj leads to a more contradictory interpretation when Adj is vague than when it is not, as expected on our theory. While our results apply directly to only a specific set of lexical items, we hypothesize that they reflect a more general pattern among gradable predicates. This motivates more systematic investigation into the role that vagueness can play in the derivation of conversational implicatures.

Exposure to media content can shape public opinions about tobacco. Accurately describing content is a first step to showing such effects. Historically, content analyses have hand-coded tobacco-focused texts from a few media sources which ignored passing mention coverage and social media sources, and could not reliably capture over-time variation. By using a combination of crowd-sourced and automated coding, we labeled the population of all e-cigarette and other tobacco-related (including cigarettes, hookah, cigars, etc.) 'long-form texts' (focused and passing coverage, in mass media and website articles) and social media items (tweets and YouTube videos) collected May 2014-June 2017 for four tobacco control themes. Automated coding of theme coverage met thresholds for item-level precision and recall, event validation, and weekly-level reliability for most sources, except YouTube. Health, Policy, Addiction and Youth themes were frequent in e-cigarette long-form focused coverage (44%-68%), but not in long-form passing coverage (5%-22%). These themes were less frequent in other tobacco coverage (long-form focused (13-32%) and passing coverage (4-11%)). Themes were infrequent in both e-cigarette (1-3%) and other tobacco tweets (2-4%). Findings demonstrate that passing e-cigarette and other tobacco long-form coverage and social media sources paint different pictures of theme coverage than focused long-form coverage. Automated coding also allowed us to code the amount of data required to estimate reliable weekly theme coverage over three years. E-cigarette theme coverage showed much more week-to-week variation than did other tobacco coverage. Automated coding allows accurate descriptions of theme coverage in passing mentions, social media, and trends in weekly theme coverage.

We show that the interpretation of sentences like John is not very ADJ depends on whether ADJ is vague. We argue that this follows from a constraint on the interaction between vagueness and conversational implicature, a domain that has not been studied extensively. The constraint states that implicatures are not drawn if they lead to “borderline contradictions” (see Ripley 2011; Alxatib & Pelletier 2011; a.o.), a natural extension of the idea that implicatures should not contradict assertions (Fox 2007; Fox & Hackl 2006; a.o.). Experiment 1 establishes that not very ADJ gives rise to the implicature ADJ for the non-vague absolute adjective late, but not for the vague relative adjective tall (in the terminology of Kennedy & McNally 2005a). Experiment 2 generalizes this result to three relative adjectives in the positive form (tall, hot, fast), against those same adjectives in their (non-vague) comparative forms (taller/hotter/faster than the average X). We also constructed quantitative meaning representations for complex predicates of the form ADJ∧¬ very ADJ, using fuzzy logic to model the contribution of boolean connectives and our experimental data to represent the meanings of adjectives. The results of these analyses suggest that strengthening not very ADJ with ADJ leads to a more contradictory interpretation when ADJ is vague than when it is not, as expected on our theory. While our results apply directly to only a specific set of lexical items, we hypothesize that they reflect a more general pattern among gradable predicates. This motivates more systematic investigation into the role that vagueness can play in the derivation of conversational implicatures. *Acknowledgments: We would like to thank David Barner, Emmanuel Chemla, Julian Grove, Emily Hanink, Chris Kennedy, Yaron McNabb, Stephanie Solt, Ming Xiang, and four anonymous reviewers, as well as audiences at the LSA Annual Meeting in 2016, UCL, University of Chicago, and SIASSI for helpful comments on various stages of the project. This work was supported by the Alexander von Humboldt-Stiftung and the Andrew W. Mellon Foundation as part of the SIAS Summer Institute “The Investigation of Linguistic Meaning,” by the European Research Council under the European Union’s Seventh Framework Programme (FP/2007-2013) / ERC Grant Agreement n.31361, by the Deutsche Forschungsgemeinschaft (DFG) as part of the Xprag.de Initiative (Grant Nr. BE 4348/4-1), and by the Netherlands Organisation for Scientific Research (NWO). †Supplemental Materials: Experimental data and associated analysis and visualization code can be found at this repository: https://github.com/lefft/not very adj

Gradable adjectives denote properties that are relativized to context dependent thresholds of application: how long an object must be in order to count as long in a context of utterance depends on what the threshold is. In the typical case, thresholds are uncertain, a feature that distinguishes them from other kinds of contextual parameters. Threshold uncertainty has played a prominent role in discussions of vagueness in philosophy of language, but there has been relativelly little attention to this feature of adjective meaning in linguistic semantics and pragmatics, and correspondingly little attention to a fundamental question about communication with gradable adjectives: what are the principles that govern decisions about threshold values in context, and so determine the communicative content of gradable adjectives in a speech situation? In this paper, we consider two recent answers to this question, which differ in the role that formal properties of adjective meaning play in determining threshold values. On one view, decisions about thresholds are based entirely on probabilisitic prior knowledge of how the objects in a gradable adjective’s domain distribute relative to the scalar property it expresses (Lassiter & Goodman 2014, 2015); according to the second view, a formal property of adjective meaning — scale structure — is primarily responsible for threshold determination (Kennedy 2007, Potts 2008, Qing & Franke 2014a,b). We identify different predictions of the two approaches and present the results of an experimental study that supports the representational approach over the distributional one.

Gradable adjectives (GAs) provide an ideal domain for evaluating theories of the interface betwen semantic interpretation and context: relative and absolute GAs are both context dependent, but absolute adjectives can have precise meanings in a way that relative adjectives cannot. We provide processing evidence for the hypothesis that imprecision in absolute adjectives is a pragmatic phenomenon: absolute thresholds are semantically scalar endpoints, and imprecise uses are derived by a global pragmatic threshold, which specifies how much deviation from literal meaning is pragmatically ignorable in a context (see e.g. Lasersohn 1999). The evidence comes from a set of visual world eye-tracking experiments adapted from the design of Sedivy, Tanenhaus, Chambers & Carlson (1999) and Aparicio, Xiang & Kennedy (2015). We measured the time-course of reference resolution in phrases of the form the Adj Noun while varying whether Adj is relative or absolute, as well as the degree to which the intended referent (target) and a competing object exemplified Adj. We found a contrast-based processing advantage for absolute GAs when the target exemplified Adj imprecisely but the competitor exemplified Adj precisely. However, this advantage was found to be sensitive to the relative/absolute distinction, as well as to the degree to which objects in the display exemplified the relevant scalar property. Collectively, our results provide evidence for a difference in the mental processes underlying relative versus absolute threshold fixing, which we argue is best modeled by a pragmatic theory of imprecision.

Recent research on the brain mechanisms underlying language processing has implicated the left anterior temporal lobe (LATL) as a central region for the composition of simple phrases. Because these studies typically present their critical stimuli without contextual information, the sensitivity of LATL responses to contextual factors is unknown. In this magnetoencephalography (MEG) study, we employed a simple question-answer paradigm to manipulate whether a prenominal adjective or determiner is interpreted restrictively, i.e., as limiting the set of entities under discussion. Our results show that the LATL is sensitive to restriction, with restrictive composition eliciting higher responses than non-restrictive composition. However, this effect was only observed when the restricting element was a determiner, adjectival stimuli showing the opposite pattern, which we hypothesise to be driven by the special pragmatic properties of non-restrictive adjectives. Overall, our results demonstrate a robust sensitivity of the LATL to high level contextual and potentially also pragmatic factors.

This paper presents a description and syntactic analysis of the focus marker lo and exclusive particle de ‘only’ in Masalit. In particular, I show that marking with lo is restricted to nominative case subjects, but does not involve a bi-clausal cleft structure. Lexical focus morphemes with asymmetric distributions are found in a wide variety of African languages, but lo is unique in that its distribution is most easily described in terms of grammatical case; whether it can be described as subject/non-subject asymmetric depends on the particular definition of “subject.” The syntactic properties of lo motivate a systematic investigation of case-focus interactions cross-linguistically. I further show that focus is not morphosyntactically marked on non-nominal constituents. The particle de is semantically similar to English only, but has adnominal syntax; as a result, apparently “projected” focus readings can arise in configurations where de takes as its semantic argument a constituent that contains it. Assuming that the verbal suffix -ti is in fact an irrealis marker, the use of de in universal modal statements, as well as the use of -ti in future tense statements, are direct consequences of the analysis of de. The interaction of -ti and de in Masalit strongly parallels the interaction of -het/-hat ‘may’ and exhaustive focus in Hungarian. Further investigation of connections between modality and focus in Masalit—and cross-linguistically—is motivated by this parallelism.