Sentences with negated adjectives receive a stronger interpretation than given by their semantics, a phenomenon called negative strengthening. It has been reported that inherently positive adjectives display a higher degree of negative strengthening than inherently negative adjectives. We investigate two possible causes of this asymmetry: intrinsic adjectival polarity and face considerations. Results of an experiment where face-related factors were manipulated suggest that both polarity and face contribute to the asymmetry. Extending a prob- abilistic RSA model of polite speech, we formalize the listener’s reasoning about a speaker’s use of negated adjectives as a tradeoff between expecting a speaker to maximize both an utterance’s social and informational utility, while avoiding inherently costly adjectives.

abstract = { Sentences with negated adjectives receive a stronger interpretation than given by their semantics, a phenomenon called negative strengthening. It has been reported that inherently positive adjectives display a higher degree of negative strengthening than inherently negative adjectives. We investigate two possible causes of this asymmetry: intrinsic adjectival polarity and face considerations. Results of an experiment where face-related factors were manipulated suggest that both polarity and face contribute to the asymmetry. Extending a prob- abilistic RSA model of polite speech, we formalize the listener’s reasoning about a speaker’s use of negated adjectives as a tradeoff between expecting a speaker to maximize both an utterance’s social and informational utility, while avoiding inherently costly adjectives. } ,
author = { Jeong, Sarang and Potts, Christopher and Degen, Judith } ,
booktitle = { Proceedings of the 46th Annual Conference of the Cognitive Science Society } ,
title = { Is the asymmetry in negative strengthening the result of adjectival polarity or face considerations? } ,
year = { to appear }
}

This paper explores how speakers choose between two ut- terance alternatives with similar syntactic properties and distinct yet related meanings. We consider the interaction of two speaker pressures: to mention accessible lexical items early in the utterance and to mention informative content early in the utterance, the latter of which is explicitly predicted by an incremental Rational Speech Act (IRSA) model. In Exp. 1, we observed a significant effect of accessibility on utterance choice in an online spoken production task, which elicited descrip- tions of the relationship between two entities using a provided verb. We found that making entities more accessible via foregrounding led speakers to mention them earlier. In Exp. 2, an interactive production task, both informativity and foregrounding were manipulated. While IRSA predicts more informative content to be mentioned earlier in the sentence, we observed neither significant effects of informativity nor of accessibility. Consistent with recent work on Good-Enough theories of production, we conclude that even when two sentences are not entirely meaning-equivalent, production choices can be affected by lexical accessibility; the pressure to mention informative material early, however, should be investigated further.

abstract = { This paper explores how speakers choose between two ut- terance alternatives with similar syntactic properties and distinct yet related meanings. We consider the interaction of two speaker pressures: to mention accessible lexical items early in the utterance and to mention informative content early in the utterance, the latter of which is explicitly predicted by an incremental Rational Speech Act (IRSA) model. In Exp. 1, we observed a significant effect of accessibility on utterance choice in an online spoken production task, which elicited descrip- tions of the relationship between two entities using a provided verb. We found that making entities more accessible via foregrounding led speakers to mention them earlier. In Exp. 2, an interactive production task, both informativity and foregrounding were manipulated. While IRSA predicts more informative content to be mentioned earlier in the sentence, we observed neither significant effects of informativity nor of accessibility. Consistent with recent work on Good-Enough theories of production, we conclude that even when two sentences are not entirely meaning-equivalent, production choices can be affected by lexical accessibility; the pressure to mention informative material early, however, should be investigated further. } ,
author = { Goodwin, Emily and Degen, Judith } ,
booktitle = { Proceedings of the 46th Annual Conference of the Cognitive Science Society } ,
title = { Production of syntactic alternations displays accessibility but not informativity effects } ,
year = { to appear }
}

Variation in the use of syntactic alternations has long been an explanatory target of language production theories. In this work, we test the predictions of several semantic, pragmatic and psycholinguistic theories of language use for the English dative alternation. We first experimentally test the role of incremental constituent informativity in the dative alternation, and find that contrary to information structural and RSA models of production, informativity has little effect on production preferences. We then more rigorously focus on accessibility effects, demonstrating that a lossy-context automatic policy can recover a key pattern of accessibility. Ultimately, we conclude that audience design pressures likely do not influence incremental production, but simply may affect planning at a broader scope.

abstract = { Variation in the use of syntactic alternations has long been an explanatory target of language production theories. In this work, we test the predictions of several semantic, pragmatic and psycholinguistic theories of language use for the English dative alternation. We first experimentally test the role of incremental constituent informativity in the dative alternation, and find that contrary to information structural and RSA models of production, informativity has little effect on production preferences. We then more rigorously focus on accessibility effects, demonstrating that a lossy-context automatic policy can recover a key pattern of accessibility. Ultimately, we conclude that audience design pressures likely do not influence incremental production, but simply may affect planning at a broader scope. } ,
author = { Rathi, Neil and Waldon, Brandon and Degen, Judith } ,
booktitle = { Proceedings of the 46th Annual Conference of the Cognitive Science Society } ,
title = { Informativity and accessibility in incremental production of the dative alternation } ,
year = { to appear }
}

Understanding referring expression generation has long been of interest to psycholinguistics, pragmatics, and sociolinguistics. Experimental data in the former two has shown that referring expression generation is modulated by both pragmatic and cognitive considerations, and the latter suggests that referring expressions have social meaning beyond their literal referential utility. This project integrates these three accounts by extending Burnett (2017)’s socially-enriched implementation of the Rational Speech Act (RSA) framework to account for variation in referring expressions used to denote transgender women in two politically opposed media corpora. Our findings highlight the utility of the RSA framework in explaining socially-modulated variation while also accounting for pragmatic and cognitive considerations. Finally, this paper contributes to growing literatures that address the relationship between (alt-)right ideologies about gender and language by highlighting the use of bioessentialist language such as biological male in the propagation of anti-trans rhetoric in the United States.

abstract = { Understanding referring expression generation has long been of interest to psycholinguistics, pragmatics, and sociolinguistics. Experimental data in the former two has shown that referring expression generation is modulated by both pragmatic and cognitive considerations, and the latter suggests that referring expressions have social meaning beyond their literal referential utility. This project integrates these three accounts by extending Burnett (2017)’s socially-enriched implementation of the Rational Speech Act (RSA) framework to account for variation in referring expressions used to denote transgender women in two politically opposed media corpora. Our findings highlight the utility of the RSA framework in explaining socially-modulated variation while also accounting for pragmatic and cognitive considerations. Finally, this paper contributes to growing literatures that address the relationship between (alt-)right ideologies about gender and language by highlighting the use of bioessentialist language such as biological male in the propagation of anti-trans rhetoric in the United States. } ,
author = { Papineau, Brandon and Degen, Judith } ,
booktitle = { Proceedings of the 46th Annual Conference of the Cognitive Science Society } ,
title = { ‘Biological males’ and ‘trans(gender) women’: social considerations in the production of referring expressions } ,
year = { to appear }
}

Presuppositions are taken to typically project out of entailment-canceling environments like the scope of negation (e.g., Chierchia and McConnell-Ginet 1990). While projection is often characterized as categorical (that is, content either typically projects or not), there is mounting empirical evidence that projection is gradient, with content being more or less projective (e.g., Karttunen 1971, Xue and Onea 2011, de Marneffe et al. 2019, Tonhauser et al. 2018, Degen and Tonhauser 2022). Mandelkern et al. 2020 critically evaluated the inference rating measures on which this empirical evidence is based and claimed that a different measure, namely naturalness ratings in explicit ignorance contexts, provides support for categorical projection and distinguishes presuppositions from nonpresuppositions. This paper presents the results of an experiment designed to investigate Mandelkern et al.’s 2020 claim for factive predicates (presumed presupposition triggers) and nonfactive ones (nontriggers). The results do not support Mandelkern et al.’s 2020 claim and rather align with Degen and Tonhauser’s 2022 result that there is no evidence for a categorical distinction between factive and nonfactive predicates.

abstract = { Presuppositions are taken to typically project out of entailment-canceling environments like the scope of negation (e.g., Chierchia and McConnell-Ginet 1990). While projection is often characterized as categorical (that is, content either typically projects or not), there is mounting empirical evidence that projection is gradient, with content being more or less projective (e.g., Karttunen 1971, Xue and Onea 2011, de Marneffe et al. 2019, Tonhauser et al. 2018, Degen and Tonhauser 2022). Mandelkern et al. 2020 critically evaluated the inference rating measures on which this empirical evidence is based and claimed that a different measure, namely naturalness ratings in explicit ignorance contexts, provides support for categorical projection and distinguishes presuppositions from nonpresuppositions. This paper presents the results of an experiment designed to investigate Mandelkern et al.’s 2020 claim for factive predicates (presumed presupposition triggers) and nonfactive ones (nontriggers). The results do not support Mandelkern et al.’s 2020 claim and rather align with Degen and Tonhauser’s 2022 result that there is no evidence for a categorical distinction between factive and nonfactive predicates. } ,
author = { Tonhauser, Judith and Degen, Judith } ,
title = { I don’t know if projection is categorical. Did Mandelkern et al. 2020 discover that it is? } ,
year = { submitted }
}

Sentences with syntactic movement out of sentential complements of manner-of- speaking (MoS) verbs (e.g., whisper, shout) are degraded in acceptability, an effect called the “manner-of-speaking (MoS) island effect”. Accounts variably attribute the MoS island effect to the violation of the subjacency condition, to the low frequency of MoS verbs taking sentential complements, or to a general information structural constraint that discourse-backgrounded constituents cannot be extracted. In three acceptability judgment experiments, we find that the MoS island effect can be modulated by foregrounding or backgrounding the extracted constituent, suggesting a causal relationship between discourse backgroundedness and the MoS island effect. Our findings challenge syntactic and frequency accounts of the MoS island effect.

abstract = { Sentences with syntactic movement out of sentential complements of manner-of- speaking (MoS) verbs (e.g., whisper, shout) are degraded in acceptability, an effect called the “manner-of-speaking (MoS) island effect”. Accounts variably attribute the MoS island effect to the violation of the subjacency condition, to the low frequency of MoS verbs taking sentential complements, or to a general information structural constraint that discourse-backgrounded constituents cannot be extracted. In three acceptability judgment experiments, we find that the MoS island effect can be modulated by foregrounding or backgrounding the extracted constituent, suggesting a causal relationship between discourse backgroundedness and the MoS island effect. Our findings challenge syntactic and frequency accounts of the MoS island effect. } ,
author = { Lu, Jiayi and Pan, Dingyi and Degen, Judith } ,
title = { Evidence for a discourse account of manner-of-speaking islands } ,
year = { submitted }
}

The softmax function is a ubiquitous helper function, frequently used as a probabilistic link function for unordered categorical data, in different kinds of models, such as regression, artificial neural networks, or probabilistic cognitive models. To fully understand the models in which the softmax function occurs, different levels of understanding of the softmax function itself are necessary. For input-output oriented models, like regression or neural network models, mathematical properties are crucial. For models with interpretable and meaningful internal representations like probabilistic cognitive models, we also require a thorough conceptual understanding of the motivation for using the softmax function (instead of something else). This tutorial provides an in-depth exposition of the informal, mathematical and conceptual properties of the softmax function. It also provides two mathematical derivations (as a stochastic choice model, and as maximum entropy distribution), together with three conceptual interpretations that can serve as rationale for using the softmax function in models that require explainability of modeling choices.

abstract = { The softmax function is a ubiquitous helper function, frequently used as a probabilistic link function for unordered categorical data, in different kinds of models, such as regression, artificial neural networks, or probabilistic cognitive models. To fully understand the models in which the softmax function occurs, different levels of understanding of the softmax function itself are necessary. For input-output oriented models, like regression or neural network models, mathematical properties are crucial. For models with interpretable and meaningful internal representations like probabilistic cognitive models, we also require a thorough conceptual understanding of the motivation for using the softmax function (instead of something else). This tutorial provides an in-depth exposition of the informal, mathematical and conceptual properties of the softmax function. It also provides two mathematical derivations (as a stochastic choice model, and as maximum entropy distribution), together with three conceptual interpretations that can serve as rationale for using the softmax function in models that require explainability of modeling choices. } ,
author = { Franke, Michael and Degen, Judith } ,
title = { The softmax function: Properties, motivation, and interpretation } ,
year = { submitted }
}

Syntactic adaptation may be a key mechanism underlying children’s learning of novel words. Havron et al. (2019) exposed French-speaking children (ages 3 to 4) to a speaker biased toward using either familiar verbs or familiar nouns in a syntactic context which permitted both structures. This prime later influenced participants’ interpretations of ambiguous novel words presented in the same syntactic frame. In Experiment 1, we successfully replicated Havron et al. with 77 French-speaking adults, using a web-based eye-tracking paradigm. Experiment 2 adapted the paradigm to English, finding that repeated exposure to a syntactic structure induced 102 English-speaking adults to update their expectations about the meanings of novel words. Experiment 3 found similar evidence of syntactic adaptation in 74 three- to five-year-old English-speaking children. Participants adapted to the specific linguistic structure used, not just the speaker’s tendency to mention actions or objects. These findings support the role of rapid adaptation during word learning and demonstrate the feasibility of conducting eye-tracking studies through online platforms.

abstract = { Syntactic adaptation may be a key mechanism underlying children’s learning of novel words. Havron et al. (2019) exposed French-speaking children (ages 3 to 4) to a speaker biased toward using either familiar verbs or familiar nouns in a syntactic context which permitted both structures. This prime later influenced participants’ interpretations of ambiguous novel words presented in the same syntactic frame. In Experiment 1, we successfully replicated Havron et al. with 77 French-speaking adults, using a web-based eye-tracking paradigm. Experiment 2 adapted the paradigm to English, finding that repeated exposure to a syntactic structure induced 102 English-speaking adults to update their expectations about the meanings of novel words. Experiment 3 found similar evidence of syntactic adaptation in 74 three- to five-year-old English-speaking children. Participants adapted to the specific linguistic structure used, not just the speaker’s tendency to mention actions or objects. These findings support the role of rapid adaptation during word learning and demonstrate the feasibility of conducting eye-tracking studies through online platforms. } ,
author = { Swanson, Elizabeth and Frank, Michael C. and Degen, Judith } ,
doi = { https://doi.org/10.34842/0ef7-5497 } ,
issue = { 1 } ,
journal = { Language Development Research } ,
title = { Syntactic adaptation and word learning in children and adults } ,
volume = { 4 } ,
year = { 2024 }
}

Sentence acceptability judgments are often affected by a pervasive phenomenon called satiation: native speakers give increasingly higher ratings to initially degraded sentences after repeated exposure. Various studies have investigated the satiation effect experimentally, the vast majority of which focused on different types of island-violating sentences in English (sentences with illicit long-distance syntactic movements). However, mixed findings are reported regarding which types of island violations are affected by satiation and which ones are not. This paper presents a meta-analysis of past experimental studies on the satiation of island effects in English, with the aim to provide accurate estimates of the rate of satiation for each type of island, test whether different island effects show different rates of satiation, explore potential factors that contributed to the heterogeneity in past results, and spot possible publication bias. The meta-analysis shows reliable satiation for adjunct islands, the Complex NP Constraint (CNPC), subject islands, the that-trace effect, the want-for construction, and whether-islands undergo satiation, albeit at different rates. No evidence for satiation is found for the Left Branch Condition (LBC). Whether context sentences were presented in the original acceptability judgment experiments predicts the differences in the rates of satiation reported across studies. Potential publication bias is found among studies testing the CNPC and whether-islands. These meta-analytic results can be used to inform debates regarding the nature of island effects, and serve as a proof of concept that meta-analysis can be a valuable tool for linguistic research.

abstract = { Sentence acceptability judgments are often affected by a pervasive phenomenon called satiation: native speakers give increasingly higher ratings to initially degraded sentences after repeated exposure. Various studies have investigated the satiation effect experimentally, the vast majority of which focused on different types of island-violating sentences in English (sentences with illicit long-distance syntactic movements). However, mixed findings are reported regarding which types of island violations are affected by satiation and which ones are not. This paper presents a meta-analysis of past experimental studies on the satiation of island effects in English, with the aim to provide accurate estimates of the rate of satiation for each type of island, test whether different island effects show different rates of satiation, explore potential factors that contributed to the heterogeneity in past results, and spot possible publication bias. The meta-analysis shows reliable satiation for adjunct islands, the Complex NP Constraint (CNPC), subject islands, the that-trace effect, the want-for construction, and whether-islands undergo satiation, albeit at different rates. No evidence for satiation is found for the Left Branch Condition (LBC). Whether context sentences were presented in the original acceptability judgment experiments predicts the differences in the rates of satiation reported across studies. Potential publication bias is found among studies testing the CNPC and whether-islands. These meta-analytic results can be used to inform debates regarding the nature of island effects, and serve as a proof of concept that meta-analysis can be a valuable tool for linguistic research. } ,
author = { Lu, Jiayi and Frank, Michael C. and Degen, Judith } ,
doi = { https://doi.org/10.5070/G60111425 } ,
issue = { 1 } ,
journal = { Glossa: Psycholinguistics } ,
title = { A meta-analysis of syntactic satiation in extraction from islands } ,
volume = { 3 } ,
year = { 2024 }
}

The syntactic log-odds ratio (SLOR), a surprisal-based measure estimated from pretrained language models (LMs) has been proposed as a linking function for human sentence acceptability judgments, a widespread measure of linguistic knowledge in experimental linguistics. We test this proposal in three steps: by examining whether SLOR values estimated by GPT-2 Small predict human acceptability judgments; by asking whether satiation effects observed in human judgments are also exhibited by fine-tuned LMs; and by testing whether satiation effects generalize in qualitatively similar ways in the model compared to humans. We show that SLOR in general predicts acceptability, but there is a significant amount of variance in acceptability data that SLOR fails to capture. SLOR also fails to capture certain patterns of satiation and generalization. Our results challenge the idea that surprisal alone, via a SLOR linking function, constitutes an accurate model for human acceptability judgments.

abstract = { The syntactic log-odds ratio (SLOR), a surprisal-based measure estimated from pretrained language models (LMs) has been proposed as a linking function for human sentence acceptability judgments, a widespread measure of linguistic knowledge in experimental linguistics. We test this proposal in three steps: by examining whether SLOR values estimated by GPT-2 Small predict human acceptability judgments; by asking whether satiation effects observed in human judgments are also exhibited by fine-tuned LMs; and by testing whether satiation effects generalize in qualitatively similar ways in the model compared to humans. We show that SLOR in general predicts acceptability, but there is a significant amount of variance in acceptability data that SLOR fails to capture. SLOR also fails to capture certain patterns of satiation and generalization. Our results challenge the idea that surprisal alone, via a SLOR linking function, constitutes an accurate model for human acceptability judgments. } ,
author = { Lu, Jiayi, Merchan, Jonathan, Wang, Lian, and Degen, Judith } ,
booktitle = { Society for Computation in Linguistics } ,
doi = { https://doi.org/10.7275/scil.2125 } ,
issue = { 1 } ,
pages = { 10-19 } ,
title = { Can syntactic log-odds ratio predict acceptability and satiation? } ,
volume = { 7 } ,
year = { 2024 }
}

The past decade has seen the rapid development of a new approach to pragmatics that attempts to integrate insights from formal and experimental semantics and pragmatics, psycholinguistics, and computational cognitive science in the study of meaning: probabilistic pragmatics. The most influential probabilistic approach to pragmatics is the Rational Speech Act (RSA) framework. In this review, I demonstrate the basic mechanics and commitments of RSA as well as some of its standard extensions, highlighting the key features that have led to its success in accounting for a wide variety of pragmatic phenomena. Fundamentally, it treats language as probabilistic, informativeness as gradient, alternatives as context-dependent, and subjective prior beliefs (world knowledge) as a crucial facet of interpretation. It also provides an integrated account of the link between production and interpretation. I highlight key challenges for RSA, which include scalability, the treatment of the boundedness of cognition, and the incremental and compositional nature of language.

abstract = { The past decade has seen the rapid development of a new approach to pragmatics that attempts to integrate insights from formal and experimental semantics and pragmatics, psycholinguistics, and computational cognitive science in the study of meaning: probabilistic pragmatics. The most influential probabilistic approach to pragmatics is the Rational Speech Act (RSA) framework. In this review, I demonstrate the basic mechanics and commitments of RSA as well as some of its standard extensions, highlighting the key features that have led to its success in accounting for a wide variety of pragmatic phenomena. Fundamentally, it treats language as probabilistic, informativeness as gradient, alternatives as context-dependent, and subjective prior beliefs (world knowledge) as a crucial facet of interpretation. It also provides an integrated account of the link between production and interpretation. I highlight key challenges for RSA, which include scalability, the treatment of the boundedness of cognition, and the incremental and compositional nature of language. } ,
author = { Degen, Judith } ,
journal = { Annual Review of Linguistics } ,
pages = { 519-540 } ,
title = { The Rational Speech Act Framework } ,
volume = { 9 } ,
year = { 2023 }
}

Scalar inferences (SI) are a signature example of how humans interpret language based on unspoken alternatives. While empirical studies have demonstrated that human SI rates are highly variable – both within instances of a single scale, and across different scales – there have been few proposals that quantitatively explain both cross- and within-scale variation. Furthermore, while it is generally assumed that SIs arise through reasoning about unspoken alternatives, it remains debated whether humans reason about alternatives as linguistic forms, or at the level of concepts. Here, we test a shared mechanism explaining SI rates within and across scales: context-driven expectations about the unspoken alternatives. Using neural language models to approximate human predictive distributions, we find that SI rates are captured by the expectedness of the strong scalemate as an alternative. Crucially, however, expectedness robustly predicts cross-scale variation only under a meaning-based view of alternatives. Our results suggest that pragmatic inferences arise from context-driven expectations over alternatives, and these expectations operate at the level of concepts.

abstract = { Scalar inferences (SI) are a signature example of how humans interpret language based on unspoken alternatives. While empirical studies have demonstrated that human SI rates are highly variable – both within instances of a single scale, and across different scales – there have been few proposals that quantitatively explain both cross- and within-scale variation. Furthermore, while it is generally assumed that SIs arise through reasoning about unspoken alternatives, it remains debated whether humans reason about alternatives as linguistic forms, or at the level of concepts. Here, we test a shared mechanism explaining SI rates within and across scales: context-driven expectations about the unspoken alternatives. Using neural language models to approximate human predictive distributions, we find that SI rates are captured by the expectedness of the strong scalemate as an alternative. Crucially, however, expectedness robustly predicts cross-scale variation only under a meaning-based view of alternatives. Our results suggest that pragmatic inferences arise from context-driven expectations over alternatives, and these expectations operate at the level of concepts. } ,
author = { Hu, Jennifer and Levy, Roger and Degen, Judith and Schuster, Sebastian } ,
booktitle = { Transactions of the Association for Computational Linguistics } ,
doi = { https://doi.org/10.1162/tacl_a_00579 } ,
pages = { 885--901 } ,
title = { Expectations over unspoken alternatives predict pragmatic inferences } ,
volume = { 11 } ,
year = { 2023 }
}

The world’s languages differ in how they order adjectives and nouns relative to each other. We ask whether cross-linguistic variation and systematicity in adjective-noun order can be explained by the iterated pressure for pragmatic referential communication. To this end, we apply the Rational Speech Act framework with an an iterated learning mechanism to study how cooperative pressures may shape typological regularities in referential communication. First, we show that the less informative adjectives are relative to nouns, the more likely they are to occur post-nominally. This is the case when informativeness is manipulated via the composition of the lexical space (i.e., changing the relative number of adjectives vs. nouns that are available for reference), and via the inherent referential utility of adjectives vs. nouns. Secondly, we show that under the assumption that nouns are on average more informative than adjectives, the model predicts a cross-linguistic distribution of ordering preferences that qualitatively resembles the empirical one, with these biases becoming further entrenched with iterated language use. Taken together, these results suggest a possible pathway for syntactic preferences to be calcified over time as the result of pragmatic communicative pressures on language.

abstract = { The world’s languages differ in how they order adjectives and nouns relative to each other. We ask whether cross-linguistic variation and systematicity in adjective-noun order can be explained by the iterated pressure for pragmatic referential communication. To this end, we apply the Rational Speech Act framework with an an iterated learning mechanism to study how cooperative pressures may shape typological regularities in referential communication. First, we show that the less informative adjectives are relative to nouns, the more likely they are to occur post-nominally. This is the case when informativeness is manipulated via the composition of the lexical space (i.e., changing the relative number of adjectives vs. nouns that are available for reference), and via the inherent referential utility of adjectives vs. nouns. Secondly, we show that under the assumption that nouns are on average more informative than adjectives, the model predicts a cross-linguistic distribution of ordering preferences that qualitatively resembles the empirical one, with these biases becoming further entrenched with iterated language use. Taken together, these results suggest a possible pathway for syntactic preferences to be calcified over time as the result of pragmatic communicative pressures on language. } ,
author = { Yu, Dhara and Waldon, Brandon, and Degen, Judith } ,
booktitle = { Proceedings of the 45th Annual Conference of the Cognitive Science Society } ,
title = { The cross-linguistic order of adjectives and nouns may be the result of iterated pragmatic pressures on referential communication } ,
year = { 2023 }
}

We present a large-scale conceptual replication of an experiment that provided evidence of false consensus biases in legal interpretation: when reading a legal contract, individuals tend to over-estimate the extent to which others would agree with their interpretation of that contract (Solan, Rosenblatt, & Osherson, 2008). Our results are consistent with this previous finding. We also observe substantial unexplained item-level variation in the extent to which individuals agree on contract interpretation, as well as unexplained variation in the extent towhich the false consensus bias holds across different contexts.In a first step towards understanding the source(s) of this variability,we show that a state-of-the-art large language model (LLM) with zero-shot prompting does not robustly predict the degree to which interpreters will exhibit consensus in a given context. However, performance improves when the model is exposed to data of the form collected in our experiment, suggesting a path forward for modeling and predicting variabilityin the interpretation of legally-relevant natural language.

abstract = { We present a large-scale conceptual replication of an experiment that provided evidence of false consensus biases in legal interpretation: when reading a legal contract, individuals tend to over-estimate the extent to which others would agree with their interpretation of that contract (Solan, Rosenblatt, & Osherson, 2008). Our results are consistent with this previous finding. We also observe substantial unexplained item-level variation in the extent to which individuals agree on contract interpretation, as well as unexplained variation in the extent towhich the false consensus bias holds across different contexts.In a first step towards understanding the source(s) of this variability,we show that a state-of-the-art large language model (LLM) with zero-shot prompting does not robustly predict the degree to which interpreters will exhibit consensus in a given context. However, performance improves when the model is exposed to data of the form collected in our experiment, suggesting a path forward for modeling and predicting variabilityin the interpretation of legally-relevant natural language. } ,
author = { Waldon, Brandon and Brodsky, Madigan and Ma, Megan and Degen, Judith } ,
booktitle = { Proceedings of the 45th Annual Conference of the Cognitive Science Society } ,
title = { Predicting consensus in legal document interpretation } ,
year = { 2023 }
}

Projection inferences are inferences about speaker commitment to a content embedded under an entailment-canceling operator, for example in polar interrogatives with clauseembedding predicates (Does John know that Julian dances Salsa?). Speaker commitment to embedded content is modulated by multiple factors, including the predicate, interlocutors’prior beliefs about the content, and its at-issueness. We propose an RSA model of projection inferences in such environments. Crucially, we take the interpretive procedure to involve inferring a speaker’s and attitude holder’s belief in the content. In a behavioral study, we investigate inferred beliefs about contents embedded under the predicates “think” and “know” that listeners ascribe to the speaker and a potential attitude holder. We use the empirical data to parametrize themodel. The resulting predictions mirror some, but not all, ofthe qualitative empirical patterns. This is a first step towards a systematic analysis of projection inferences using probabilistic pragmatic models.

abstract = { Projection inferences are inferences about speaker commitment to a content embedded under an entailment-canceling operator, for example in polar interrogatives with clauseembedding predicates (Does John know that Julian dances Salsa?). Speaker commitment to embedded content is modulated by multiple factors, including the predicate, interlocutors’prior beliefs about the content, and its at-issueness. We propose an RSA model of projection inferences in such environments. Crucially, we take the interpretive procedure to involve inferring a speaker’s and attitude holder’s belief in the content. In a behavioral study, we investigate inferred beliefs about contents embedded under the predicates “think” and “know” that listeners ascribe to the speaker and a potential attitude holder. We use the empirical data to parametrize themodel. The resulting predictions mirror some, but not all, ofthe qualitative empirical patterns. This is a first step towards a systematic analysis of projection inferences using probabilistic pragmatic models. } ,
author = { Pan, Dingyi and Degen, Judith } ,
booktitle = { Proceedings of the 45th Annual Conference of the Cognitive Science Society } ,
title = { Towards a computational account of projection inferences in polar interrogatives with clause-embedding predicates } ,
year = { 2023 }
}

Using evidential expressions to indicate one’s source of information for an utterance tends to convey uncertainty on the speaker’s part. Previous accounts of this uncertainty inference attribute it to either extralinguistic reasoning about evidence directness, or to pragmatic reasoning about alternative utterances. Here we present a novel hybrid account, and introduce a set of utterances which allows us to tease apart the three accounts’ predictions. We test these predictions in two studies by manipulating the directness of evidence indicated by an evidential expression. Exp. 1 shows that listeners infer more uncertainty with extreme values of directness. Exp. 2 shows that speakers are more likely to indicate evidence in contexts where the evidence is unreliable. We argue that these findings support an account which involves both extralinguistic and pragmatic reasoning, and develop a formal implementation of such an account within the Rational Speech Act framework.

abstract = { Using evidential expressions to indicate one’s source of information for an utterance tends to convey uncertainty on the speaker’s part. Previous accounts of this uncertainty inference attribute it to either extralinguistic reasoning about evidence directness, or to pragmatic reasoning about alternative utterances. Here we present a novel hybrid account, and introduce a set of utterances which allows us to tease apart the three accounts’ predictions. We test these predictions in two studies by manipulating the directness of evidence indicated by an evidential expression. Exp. 1 shows that listeners infer more uncertainty with extreme values of directness. Exp. 2 shows that speakers are more likely to indicate evidence in contexts where the evidence is unreliable. We argue that these findings support an account which involves both extralinguistic and pragmatic reasoning, and develop a formal implementation of such an account within the Rational Speech Act framework. } ,
author = { Fishman, Alon and Degen, Judith } ,
booktitle = { Proceedings of the 45th Annual Conference of the Cognitive Science Society } ,
title = { Evidential uncertainty involves both pragmatic and extralinguistic reasoning: a computational account } ,
year = { 2023 }
}

The context dependence of artifact noun category boundaries is underexplored relative to interpretational context dependence in other linguistic domains (e.g., gradable adjectives). Taking inspiration from a normative debate over the role of context in legal interpretation (in particular, the role of legislators’ policy goals), we show experimentally that contextual information as to a rule’s purpose systematically modulates interpreter beliefs about the category boundaries of artifact nouns contained within the rule. We propose a Bayesian pragmatic model of the context-dependent resolution of artifact noun extensions, which we compare against context-independent baselines. Our experimental and modeling results suggest the need for an explicitly context-sensitive, multi-dimensional degree semantics for artifact nouns.

abstract = { The context dependence of artifact noun category boundaries is underexplored relative to interpretational context dependence in other linguistic domains (e.g., gradable adjectives). Taking inspiration from a normative debate over the role of context in legal interpretation (in particular, the role of legislators’ policy goals), we show experimentally that contextual information as to a rule’s purpose systematically modulates interpreter beliefs about the category boundaries of artifact nouns contained within the rule. We propose a Bayesian pragmatic model of the context-dependent resolution of artifact noun extensions, which we compare against context-independent baselines. Our experimental and modeling results suggest the need for an explicitly context-sensitive, multi-dimensional degree semantics for artifact nouns. } ,
author = { Waldon, Brandon and Condoravdi, Cleo and Levin, Beth and Degen, Judith } ,
booktitle = { Proceedings of Sinn and Bedeutung } ,
doi = { https://doi.org/10.18148/sub/2023.v27.1093 } ,
title = { On the context dependence of artifact noun interpretation } ,
volume = { 27 } ,
year = { 2023 }
}

This paper is about epistemic must and might. A sizeable proportion of the literature on epistemic modality takes as its point of departure data such as the following: (1) (In the context of direct observation of rain): a. # It must be raining outside. b. It is raining outside. Karttunen (1972) remarks that must φ is degraded when compared to bare unmodalized assertions of φ in ‘direct-observation’ contexts like (1). I’ll briefly address this ‘directness’ problem for the semantics of must, but my point of departure is a less-discussed issue. I’ll call it the ‘argumentation’ problem, discussed at length by Stone (1994) and illustrated by exchanges such as the following: (2) a. A: What’s the weather like outside? b. B: ? It must be raining. Stone remarks that in order for (2b) to be felicitous, Speaker A has to be able to identify the particular chain of reasoning that licenses the conclusion that it’s raining. If A cannot reconstruct B’s argument for rain, then (2b) is degraded. Like the directness problem, the argumentation problem has to do with an asymmetry in the felicity conditions of must φ and its bare counterpart. In (2), if B simply responds It’s raining, then B’s argument for rain need not be salient in context. A solution to the argumentation problem explains this asymmetry. I review two perspectives on the problem: the Semantic account, whereby the ‘argumentation’ property of must is encoded in the semantics of the modal; and the Pragmatic account, which posits that this property is contextually conditioned. Next, I present experimental evidence against the Pragmatic account. Finally, I present a solution to the argumentation problem and offer an account of must’s putative dual, might.

abstract = { This paper is about epistemic must and might. A sizeable proportion of the literature on epistemic modality takes as its point of departure data such as the following: (1) (In the context of direct observation of rain): a. # It must be raining outside. b. It is raining outside. Karttunen (1972) remarks that must φ is degraded when compared to bare unmodalized assertions of φ in ‘direct-observation’ contexts like (1). I’ll briefly address this ‘directness’ problem for the semantics of must, but my point of departure is a less-discussed issue. I’ll call it the ‘argumentation’ problem, discussed at length by Stone (1994) and illustrated by exchanges such as the following: (2) a. A: What’s the weather like outside? b. B: ? It must be raining. Stone remarks that in order for (2b) to be felicitous, Speaker A has to be able to identify the particular chain of reasoning that licenses the conclusion that it’s raining. If A cannot reconstruct B’s argument for rain, then (2b) is degraded. Like the directness problem, the argumentation problem has to do with an asymmetry in the felicity conditions of must φ and its bare counterpart. In (2), if B simply responds It’s raining, then B’s argument for rain need not be salient in context. A solution to the argumentation problem explains this asymmetry. I review two perspectives on the problem: the Semantic account, whereby the ‘argumentation’ property of must is encoded in the semantics of the modal; and the Pragmatic account, which posits that this property is contextually conditioned. Next, I present experimental evidence against the Pragmatic account. Finally, I present a solution to the argumentation problem and offer an account of must’s putative dual, might. } ,
author = { Waldon, Brandon } ,
booktitle = { Proceedings of the 56th Meeting of the Chicago Linguistic Society (CLS 56) } ,
title = { Epistemic must and might: evidence that argumentation is semantically encoded } ,
year = { 2022 }
}

While widely used in psycholinguistics, the linking hypothesis for eye movements in the visual world paradigm is still poorly understood. Recent work on linking hypotheses for referential tasks in particular has found mixed support for the Referential Belief Link: that the proportion of looks to a referent in a time window reflects participants’ degree of belief that the referent is the intended target in that time window. Here we test the hypothesis that participants’ expectations for the utterances observed in an experiment modulate the extent to which the Referential Belief Link holds. This hypothesis is motivated by a simple idea: when utterances are unexpected, listeners engage in additional reasoning to make sense of the observed signal. In a re-analysis of a previous eye movement and incremental decision task dataset, in conjunction with two novel production experiments, we find that the more surprising an observed utterance is, the smaller the correlation between explicit and implicit beliefs is. We discuss the importance of participants’ production expectations in research using the visual world paradigm.

abstract = { While widely used in psycholinguistics, the linking hypothesis for eye movements in the visual world paradigm is still poorly understood. Recent work on linking hypotheses for referential tasks in particular has found mixed support for the Referential Belief Link: that the proportion of looks to a referent in a time window reflects participants’ degree of belief that the referent is the intended target in that time window. Here we test the hypothesis that participants’ expectations for the utterances observed in an experiment modulate the extent to which the Referential Belief Link holds. This hypothesis is motivated by a simple idea: when utterances are unexpected, listeners engage in additional reasoning to make sense of the observed signal. In a re-analysis of a previous eye movement and incremental decision task dataset, in conjunction with two novel production experiments, we find that the more surprising an observed utterance is, the smaller the correlation between explicit and implicit beliefs is. We discuss the importance of participants’ production expectations in research using the visual world paradigm. } ,
author = { Degen, Judith and Pophristic, Stefan } ,
booktitle = { Proceedings of the 44th Annual Meeting of the Cognitive Science Society } ,
keywords = { psycholinguistics; experimental pragmatics; scalar implicature; linking hypothesis; visual world paradigm; eye-tracking } ,
title = { The role of production expectations in visual world paradigm linking hypotheses } ,
year = { 2022 }
}

author = { Degen, Judith and Tonhauser, Judith } ,
booktitle = { The Open Handbook of Linguistic Data Management } ,
doi = { 10.7551/mitpress/12200.003.0052 } ,
isbn = { 9780262366076 } ,
month = { 01 } ,
publisher = { The MIT Press } ,
title = { Managing Web Experiments for Psycholinguistics: An Example from Experimental Semantics/Pragmatics } ,
year = { 2022 }
}

The ordering of morphemes in a word displays well-documented regularities across languages. Previous work has explained these in terms of notions such as semantic scope, relevance, and productivity. Here, we test a recently formulated processing theory of the ordering of linguistic units, the efficient tradeoff hypothesis (Hahn et al., 2021). The claim of the theory is that morpheme ordering can partly be explained by the optimization of a tradeoff between memory and surprisal. This claim has received initial empirical support from two languages. In this work, we test this idea more extensively using data from four additional agglutinative languages with significant amounts of morphology, and by considering nouns in addition to verbs. We find that the efficient tradeoff hypothesis predicts ordering in most cases with high accuracy, and accounts for cross-linguistic regularities in noun and verb inflection. Our work adds to a growing body of work suggesting that many ordering properties of language arise from a pressure for efficient language processing.

abstract = { The ordering of morphemes in a word displays well-documented regularities across languages. Previous work has explained these in terms of notions such as semantic scope, relevance, and productivity. Here, we test a recently formulated processing theory of the ordering of linguistic units, the efficient tradeoff hypothesis (Hahn et al., 2021). The claim of the theory is that morpheme ordering can partly be explained by the optimization of a tradeoff between memory and surprisal. This claim has received initial empirical support from two languages. In this work, we test this idea more extensively using data from four additional agglutinative languages with significant amounts of morphology, and by considering nouns in addition to verbs. We find that the efficient tradeoff hypothesis predicts ordering in most cases with high accuracy, and accounts for cross-linguistic regularities in noun and verb inflection. Our work adds to a growing body of work suggesting that many ordering properties of language arise from a pressure for efficient language processing. } ,
author = { Hahn, Michael and Mathew, Rebecca and Degen, Judith } ,
doi = { 10.1162/opmi_a_00051 } ,
issn = { 2470-2986 } ,
journal = { Open Mind } ,
pages = { 208-232 } ,
title = { Morpheme Ordering Across Languages Reflects Optimization for Processing Efficiency } ,
volume = { 5 } ,
year = { 2022 }
}

Redundant modification in referring expression production varies both within language (e.g., English speakers produce more redundant color than size modifiers) and cross- linguistically (e.g., English speakers produce more redundant color modifoers than Spanish speakers). It is an open question whether these asymmetries are the result of asymmetries in the general referential utility of color and size modifiers or of incremental language processing pressures. Cross-linguistic investigations of redundant modification are important to this debate: similar cross-linguistic rates of redundant modification would suggest a strong role for general referential utility. In contrast, lower prevalence of redundant modification in languages with post-nominal modification suggests a strong role for incrementality. Here, we test whether differences in redundant adjective use are systematic for a particularly interesting language: Central Taurus Sign Language. As a language in its infancy, CTSL has no established conventions, and therefore provides us with a unique opportunity to explore how redundancy emerges in the initial stages of language formation. We evaluate different computational models of referring expression that each make different assumptions regarding the source of asymmetries in the production of redundant modifiers.

abstract = { Redundant modification in referring expression production varies both within language (e.g., English speakers produce more redundant color than size modifiers) and cross- linguistically (e.g., English speakers produce more redundant color modifoers than Spanish speakers). It is an open question whether these asymmetries are the result of asymmetries in the general referential utility of color and size modifiers or of incremental language processing pressures. Cross-linguistic investigations of redundant modification are important to this debate: similar cross-linguistic rates of redundant modification would suggest a strong role for general referential utility. In contrast, lower prevalence of redundant modification in languages with post-nominal modification suggests a strong role for incrementality. Here, we test whether differences in redundant adjective use are systematic for a particularly interesting language: Central Taurus Sign Language. As a language in its infancy, CTSL has no established conventions, and therefore provides us with a unique opportunity to explore how redundancy emerges in the initial stages of language formation. We evaluate different computational models of referring expression that each make different assumptions regarding the source of asymmetries in the production of redundant modifiers. } ,
author = { Kursat, Leyla and Waldon, Brandon and Ergin, Rabia and Degen, Judith } ,
booktitle = { Proceedings of the 44th Annual Meeting of the Cognitive Science Society } ,
keywords = { experimental pragmatics; redundancy; village sign language; Rational Speech Act; Bayesian Data Analysis } ,
title = { Evaluating models of referring expression production on an emerging sign language } ,
year = { 2022 }
}

A recent proposal of syntactic satiation claims that it is driven by adaptation: comprehenders track and update their beliefs about the probability of observing certain sentences, leading to subsequent increases in the acceptability of those sentences. This leaves open what the representational targets of satiation are, that is: what is the tracked information that belief update is based on? In two acceptability judgment experiments, we show that exposure to one type of island violation can lead to the satiation of another island type, suggesting that island type-general representations are tracked by comprehenders in addition to island type-specific representations. The same experimental paradigm can be used for further exploration of the representational targets of satiation.

abstract = { A recent proposal of syntactic satiation claims that it is driven by adaptation: comprehenders track and update their beliefs about the probability of observing certain sentences, leading to subsequent increases in the acceptability of those sentences. This leaves open what the representational targets of satiation are, that is: what is the tracked information that belief update is based on? In two acceptability judgment experiments, we show that exposure to one type of island violation can lead to the satiation of another island type, suggesting that island type-general representations are tracked by comprehenders in addition to island type-specific representations. The same experimental paradigm can be used for further exploration of the representational targets of satiation. } ,
author = { Lu, Jiayi and Wright, Nicholas and Degen, Judith } ,
booktitle = { Proceedings of the 44th Annual Meeting of the Cognitive Science Society } ,
keywords = { psycholinguistics; island effect; adaptation; satiation; acceptability judgments } ,
title = { Satiation effects generalize across island types } ,
year = { 2022 }
}

Language and gender are inextricably linked; we regularly make reference to the genders of individuals around us, and the language used to do so recursively feeds the biases we hold about gender in the social world. What has been left under-investigated is the role that individual, rather than societally-held, ideologies about gender play in the linguistic system. In two web-based studies, we investigate the pro- cessing and production of gender-neutral role nouns such as congressperson as a function of individual gender ideology and political alignment. Our results indicate an asymmetry between the processing and production of such nouns: while individuals’ gender ideologies do not modulate processing, they do interact with political party in production tasks such that Democratic participants with more progressive gender ideologies produce more gender-neutral role nouns. We argue that these forms have become linguistic resources for indexing social progressiveness, leading to their use by Democrats and avoidance by Republicans.

abstract = { Language and gender are inextricably linked; we regularly make reference to the genders of individuals around us, and the language used to do so recursively feeds the biases we hold about gender in the social world. What has been left under-investigated is the role that individual, rather than societally-held, ideologies about gender play in the linguistic system. In two web-based studies, we investigate the pro- cessing and production of gender-neutral role nouns such as congressperson as a function of individual gender ideology and political alignment. Our results indicate an asymmetry between the processing and production of such nouns: while individuals’ gender ideologies do not modulate processing, they do interact with political party in production tasks such that Democratic participants with more progressive gender ideologies produce more gender-neutral role nouns. We argue that these forms have become linguistic resources for indexing social progressiveness, leading to their use by Democrats and avoidance by Republicans. } ,
author = { Papineau, Brandon and Podesva, Robert J. and Degen, Judith } ,
booktitle = { Proceedings of the 44th Annual Meeting of the Cognitive Science Society } ,
keywords = { language and gender; language processing; language production; language and politics; morphology } ,
title = { ‘Sally the Congressperson’: The Role of Individual Ideology on the Processing and Production of English Gender-Neutral Role Nouns } ,
year = { 2022 }
}

Properties of the content of the clausal complement have long been assumed to distinguish factive predicates like 'know' from non-factive ones like 'think' (Kiparsky and Kiparsky 1970, i.a.). There is, however, disagreement about which properties define factive predicates as well as uncertainty about whether the content of the complement of particular predicates exhibit the properties attributed to the content of the complement of factive predicates. This has led to a lack of consensus about which predicates are factive, a troublesome situation given the central role that factivity plays in linguistic theorizing. This paper reports six experiments designed to investigate two critical properties of the content of the complement of clause-embedding predicates, namely projection and entailment, with the goal of establishing whether these properties identify a class of factive predicates. We find that factive predicates are more heterogeneous than previously assumed and that there is little empirical support from these experiments for the assumed categorical distinction between factive and non-factive predicates. We discuss implications of our results for formal analyses of presuppositions, one area where factivity has played a central role. We propose that projection is sensitive to more fine-grained meaning distinctions between clause-embedding predicates than factivity.

abstract = { Properties of the content of the clausal complement have long been assumed to distinguish factive predicates like 'know' from non-factive ones like 'think' (Kiparsky and Kiparsky 1970, i.a.). There is, however, disagreement about which properties define factive predicates as well as uncertainty about whether the content of the complement of particular predicates exhibit the properties attributed to the content of the complement of factive predicates. This has led to a lack of consensus about which predicates are factive, a troublesome situation given the central role that factivity plays in linguistic theorizing. This paper reports six experiments designed to investigate two critical properties of the content of the complement of clause-embedding predicates, namely projection and entailment, with the goal of establishing whether these properties identify a class of factive predicates. We find that factive predicates are more heterogeneous than previously assumed and that there is little empirical support from these experiments for the assumed categorical distinction between factive and non-factive predicates. We discuss implications of our results for formal analyses of presuppositions, one area where factivity has played a central role. We propose that projection is sensitive to more fine-grained meaning distinctions between clause-embedding predicates than factivity. } ,
author = { Degen, Judith and Tonhauser, Judith } ,
doi = { doi:10.1353/lan.2022.0015 } ,
issue = { 3 } ,
journal = { Language } ,
title = { Are there factive predicates? An empirical investigation } ,
volume = { 98 } ,
year = { 2022 }
}

Experimental investigation is fundamental to theory-building in cognitive science, but its value depends on the linking as- sumptions made by researchers about the mapping between empirical measurements and theoretical constructs. We ar- gue that sufficient clarity and justification are often lacking for linking assumptions made in visual world eye-tracking, a widely used experimental method in psycholinguistic research. We test what we term the Referential Belief linking assump- tion: that the proportion of looks to a referent in a time window reflects participants’ degree of belief that the referent is the in- tended target in that time window. We do so by comparing eye-tracking data against explicit beliefs collected in an incre- mental decision task (Exp. 1), which replicates a scalar impli- cature processing study (Exp. 3 of Sun & Breheny, 2020). In Exp. 2, we replicate Sun and Breheny (2020) in a web-based eye-tracking paradigm using WebGazer.js. The results pro- vide support for the Referential Belief link and cautious opti- mism for the prospect of conducting web-based eye-tracking. We discuss limitations on both fronts.

abstract = { Experimental investigation is fundamental to theory-building in cognitive science, but its value depends on the linking as- sumptions made by researchers about the mapping between empirical measurements and theoretical constructs. We ar- gue that sufficient clarity and justification are often lacking for linking assumptions made in visual world eye-tracking, a widely used experimental method in psycholinguistic research. We test what we term the Referential Belief linking assump- tion: that the proportion of looks to a referent in a time window reflects participants’ degree of belief that the referent is the in- tended target in that time window. We do so by comparing eye-tracking data against explicit beliefs collected in an incre- mental decision task (Exp. 1), which replicates a scalar impli- cature processing study (Exp. 3 of Sun & Breheny, 2020). In Exp. 2, we replicate Sun and Breheny (2020) in a web-based eye-tracking paradigm using WebGazer.js. The results pro- vide support for the Referential Belief link and cautious opti- mism for the prospect of conducting web-based eye-tracking. We discuss limitations on both fronts. } ,
author = { Degen, Judith and Kursat, Leyla and Leigh, Daisy } ,
booktitle = { Proceedings of the 43nd Annual Meeting of the Cognitive Science Society } ,
journal = { Proceedings of the Annual Meeting of the Cognitive Science Society } ,
keywords = { psycholinguistics, experimental pragmatics, scalar implicature, linking functions, visual world, eye-tracking } ,
title = { Seeing is believing: testing an explicit linking assumption for visual world eye-tracking in psycholinguistics } ,
volume = { 43 } ,
year = { 2021 }
}

Beliefs about the world affect language processing and interpretation in several empirical domains. In two experiments, we tested whether subjective prior beliefs about the probability of utterance content modulate projection, that is, listeners’ inferences about speaker commitment to that content. We find that prior beliefs predict projection at both the group and the by-participant level: the higher the prior belief in a content, the more speakers are taken to be committed to it. This result motivates the integration of formal analyses of projection with cognitive theories of language understanding.

abstract = { Beliefs about the world affect language processing and interpretation in several empirical domains. In two experiments, we tested whether subjective prior beliefs about the probability of utterance content modulate projection, that is, listeners’ inferences about speaker commitment to that content. We find that prior beliefs predict projection at both the group and the by-participant level: the higher the prior belief in a content, the more speakers are taken to be committed to it. This result motivates the integration of formal analyses of projection with cognitive theories of language understanding. } ,
author = { Degen, Judith and Tonhauser, Judith } ,
doi = { https://doi.org/10.1162/opmi_a_00042 } ,
journal = { Open Mind } ,
keywords = { experimental semantics, experimental pragmatics, projection } ,
pages = { 59-70 } ,
title = { Prior beliefs modulate projection } ,
volume = { 5 } ,
year = { 2021 }
}

Memory limitations are known to constrain language comprehension and production, and have been argued to account for crosslinguistic word order regularities. However, a systematic assessment of the role of memory limitations in language structure has proven elusive, in part because it is hard to extract precise large-scale quantitative generalizations about language from existing mechanistic models of memory use in sentence processing. We provide an architecture-independent information-theoretic formalization of memory limitations which enables a simple calculation of the memory efficiency of languages. Our notion of memory efficiency is based on the idea of a memory--surprisal tradeoff: a certain level of average surprisal per word can only be achieved at the cost of storing some amount of information about past context. Based on this notion of memory usage, we advance the Efficient Tradeoff Hypothesis: the order of elements in natural language is under pressure to enable favorable memory-surprisal tradeoffs. We derive that languages enable more efficient tradeoffs when they exhibit information locality: when predictive information about an element is concentrated in its recent past. We provide empirical evidence from three test domains in support of the Efficient Tradeoff Hypothesis: a reanalysis of a miniature artificial language learning experiment, a large-scale study of word order in corpora of 54 languages, and an analysis of morpheme order in two agglutinative languages. These results suggest that principles of order in natural language can be explained via highly generic cognitively motivated principles and lend support to efficiency-based models of the structure of human language.

abstract = { Memory limitations are known to constrain language comprehension and production, and have been argued to account for crosslinguistic word order regularities. However, a systematic assessment of the role of memory limitations in language structure has proven elusive, in part because it is hard to extract precise large-scale quantitative generalizations about language from existing mechanistic models of memory use in sentence processing. We provide an architecture-independent information-theoretic formalization of memory limitations which enables a simple calculation of the memory efficiency of languages. Our notion of memory efficiency is based on the idea of a memory--surprisal tradeoff: a certain level of average surprisal per word can only be achieved at the cost of storing some amount of information about past context. Based on this notion of memory usage, we advance the Efficient Tradeoff Hypothesis: the order of elements in natural language is under pressure to enable favorable memory-surprisal tradeoffs. We derive that languages enable more efficient tradeoffs when they exhibit information locality: when predictive information about an element is concentrated in its recent past. We provide empirical evidence from three test domains in support of the Efficient Tradeoff Hypothesis: a reanalysis of a miniature artificial language learning experiment, a large-scale study of word order in corpora of 54 languages, and an analysis of morpheme order in two agglutinative languages. These results suggest that principles of order in natural language can be explained via highly generic cognitively motivated principles and lend support to efficiency-based models of the structure of human language. } ,
archiveprefix = { psyarxiv } ,
author = { Hahn, Michael and Degen, Judith and Futrell, Richard } ,
doi = { 10.31234/osf.io/nu4qz } ,
github = { https://github.com/m-hahn/memory-surprisal } ,
issue = { 4 } ,
journal = { Psychological Review } ,
pages = { 726--756 } ,
slides = { files/osf-CUNY 2019-Tradeoff.pdf } ,
supplement = { files/hahn_psychreview_2021_si.pdf } ,
title = { Modeling word and morpheme order in natural language as an efficient tradeoff of memory and surprisal } ,
volume = { 128 } ,
year = { 2021 }
}

When referring to objects, speakers are often more specific than neces- sary for the purpose of establishing unique reference, e.g., by producing redundant modifiers. A computational model of referring expression production that accounts for many of the key patterns in redundant adjectival modification assumes that adjectives differ in how noisy (reliable), and consequently, how useful they are for reference. Here we investigate one hypothesis about the source of the assumed adjectival noise: that it reflects the perceptual difficulty of establishing whether the property denoted by the adjective holds of the contextually relevant objects. In Exp.1, we collect perceptual difficulty norms for items that vary in color and material. In Exp. 2, we test the highest (material) and lowest (color) perceptual dif- ficulty items in a reference game and find that material is indeed less likely to be mentioned redundantly, replicating previous work. In Exp. 3, we obtain norms for the tested items in a second perceptual difficulty measure with the aim of testing the effect of perceptual difficulty within property type. The overall results provide preliminary support for the hypothesis that the propensity to redundantly use color over material adjectives may be driven by the relative ease of assessing an object’s color, compared to the relative difficulty of assessing its material.

abstract = { When referring to objects, speakers are often more specific than neces- sary for the purpose of establishing unique reference, e.g., by producing redundant modifiers. A computational model of referring expression production that accounts for many of the key patterns in redundant adjectival modification assumes that adjectives differ in how noisy (reliable), and consequently, how useful they are for reference. Here we investigate one hypothesis about the source of the assumed adjectival noise: that it reflects the perceptual difficulty of establishing whether the property denoted by the adjective holds of the contextually relevant objects. In Exp.1, we collect perceptual difficulty norms for items that vary in color and material. In Exp. 2, we test the highest (material) and lowest (color) perceptual dif- ficulty items in a reference game and find that material is indeed less likely to be mentioned redundantly, replicating previous work. In Exp. 3, we obtain norms for the tested items in a second perceptual difficulty measure with the aim of testing the effect of perceptual difficulty within property type. The overall results provide preliminary support for the hypothesis that the propensity to redundantly use color over material adjectives may be driven by the relative ease of assessing an object’s color, compared to the relative difficulty of assessing its material. } ,
author = { Kursat, Leyla and Degen, Judith } ,
booktitle = { Proceedings of the Linguistic Society of America } ,
doi = { https://doi.org/10.3765/plsa.v6i1.5003 } ,
number = { 1 } ,
pages = { 676--688 } ,
title = { Perceptual difficulty differences predict asymmetry in redundant modification with color and material adjectives } ,
volume = { 6 } ,
year = { 2021 }
}

Neural networks have recently successfully learned to predict some pragmatic inferences (e.g., Jeretic et al. (2020); Jiang and de Marneffe (2019)). For instance, Schuster et al. (2020) trained a neuralnetwork to predict human ratings of scalar inference strength from “some” to the negation of a stronger alternative with “all”. However, it remains an open question to what extent these results are specific to the inference from “some” to “not all” or whether they generalize to other types of scalar inferences. We thus explore to what extent a neural network can learn to predict a different widely studied scalar inference: that from “or” to the negation of a stronger alternative with “and”.

abstract = { Neural networks have recently successfully learned to predict some pragmatic inferences (e.g., Jeretic et al. (2020); Jiang and de Marneffe (2019)). For instance, Schuster et al. (2020) trained a neuralnetwork to predict human ratings of scalar inference strength from “some” to the negation of a stronger alternative with “all”. However, it remains an open question to what extent these results are specific to the inference from “some” to “not all” or whether they generalize to other types of scalar inferences. We thus explore to what extent a neural network can learn to predict a different widely studied scalar inference: that from “or” to the negation of a stronger alternative with “and”. } ,
author = { Li, Elissa and Schuster, Sebastian and Degen, Judith } ,
booktitle = { Proceedings of the Society for Computation in Linguistics } ,
doi = { https://doi.org/10.7275/xr01-a852 } ,
title = { Predicting scalar inferences from ``or'' to ``not both'' using neural sentence encoders } ,
volume = { 4 } ,
year = { 2021 }
}

Listeners adapt to variability in language use by updating their expectations over variants, often in speaker-specific ways. We propose that adaptation of this sort contributes to satiation, the phenomenon whereby the acceptability of unacceptable sentences increases after repeated exposure. We provide support for an adaptation account of satiation by showing that the satiation of purportedly unaccept- able island-violating constructions demonstrates speaker- specificity, a key property of adaptation.

abstract = { Listeners adapt to variability in language use by updating their expectations over variants, often in speaker-specific ways. We propose that adaptation of this sort contributes to satiation, the phenomenon whereby the acceptability of unacceptable sentences increases after repeated exposure. We provide support for an adaptation account of satiation by showing that the satiation of purportedly unaccept- able island-violating constructions demonstrates speaker- specificity, a key property of adaptation. } ,
author = { Lu, Jiayi and Lassiter, Dan and Degen, Judith } ,
booktitle = { Proceedings of the 42nd Annual Meeting of the Cognitive Science Society } ,
journal = { Proceedings of the Annual Meeting of the Cognitive Science Society } ,
title = { Syntactic satiation is driven by speaker-specific adaptation } ,
volume = { 43 } ,
year = { 2021 }
}

We present a novel probabilistic model of referring expression production, synthesizing recent analyses proposed within the Rational Speech Act (RSA) framework (Frank and Goodman, 2012). Our model makes incremental utterance choice predictions (Cohn-Gordon et al. 2018a; Cohn-Gordon et al. 2018b) and assumes a non-deterministic semantics for adjectives in referring expressions (Degen et al. 2020). The model captures previously attested production patterns in reference game experiments, including English speakers’ tendency to produce redundant color adjectives more frequently than redundant size adjectives, as well as Spanish speakers’ tendency to employ redundant color adjectives less frequently than English speakers. We report the predictions made by the model under various parameter regimes, motivating future empirical work.

abstract = { We present a novel probabilistic model of referring expression production, synthesizing recent analyses proposed within the Rational Speech Act (RSA) framework (Frank and Goodman, 2012). Our model makes incremental utterance choice predictions (Cohn-Gordon et al. 2018a; Cohn-Gordon et al. 2018b) and assumes a non-deterministic semantics for adjectives in referring expressions (Degen et al. 2020). The model captures previously attested production patterns in reference game experiments, including English speakers’ tendency to produce redundant color adjectives more frequently than redundant size adjectives, as well as Spanish speakers’ tendency to employ redundant color adjectives less frequently than English speakers. We report the predictions made by the model under various parameter regimes, motivating future empirical work. } ,
author = { Waldon, Brandon and Degen, Judith } ,
booktitle = { Proceedings of the Society for Computation in Linguistics } ,
doi = { https://doi.org/10.7275/vsfn-t057 } ,
title = { Modeling cross-linguistic production of referring expressions } ,
volume = { 4 } ,
year = { 2021 }
}

Asking and answering questions is a staple of human communication. To answer a question effectively, a hearer must interpret the speaker's intention given the specific question asked. ‘Wh’-questions like ‘Where can I get coffee?’ are underspecified for (non-)exhaustivity, i.e., how many answers must be provided to resolve the speaker's goal. Intuitions from the semantics literature report that questions are generally exhaustive, and non-exhaustive only in the context of specific linguistic factors (e.g., the modal ‘can’, certain ‘wh’-words). To test these assumptions, we collected question paraphrase ratings for naturally occurring root questions in variable linguistic contexts. In contrast to previous claims, we find that questions are not biased for exhaustivity. However, other prior observations are supported by the data. We argue that a full account of the observed distribution of meanings must integrate discourse factors like the hearer's estimate of the speaker's goal, alongside (or subsuming the effect of) linguistic cues.

abstract = { Asking and answering questions is a staple of human communication. To answer a question effectively, a hearer must interpret the speaker's intention given the specific question asked. ‘Wh’-questions like ‘Where can I get coffee?’ are underspecified for (non-)exhaustivity, i.e., how many answers must be provided to resolve the speaker's goal. Intuitions from the semantics literature report that questions are generally exhaustive, and non-exhaustive only in the context of specific linguistic factors (e.g., the modal ‘can’, certain ‘wh’-words). To test these assumptions, we collected question paraphrase ratings for naturally occurring root questions in variable linguistic contexts. In contrast to previous claims, we find that questions are not biased for exhaustivity. However, other prior observations are supported by the data. We argue that a full account of the observed distribution of meanings must integrate discourse factors like the hearer's estimate of the speaker's goal, alongside (or subsuming the effect of) linguistic cues. } ,
author = { Moyer, Morgan and Degen, Judith } ,
booktitle = { Proceedings of the Annual Meeting of the Cognitive Science Society } ,
title = { Who thinks wh-questions are exhaustive? } ,
volume = { 43 } ,
year = { 2021 }
}

Referring is one of the most basic and prevalent uses of language. How do speakers choose from the wealth of referring expressions at their disposal? Rational theories of language use have come under attack for decades for not being able to account for the seemingly irrational overinformativeness ubiquitous in referring expressions. Here we present a novel production model of referring expressions within the Rational Speech Act framework that treats speakers as agents that rationally trade off cost and informativeness of utterances. Crucially, we relax the assumption that informativeness is computed with respect to a deterministic Boolean semantics, in favor of a non-deterministic continuous semantics. This innovation allows us to capture a large number of seemingly disparate phenomena within one unified framework: the basic asymmetry in speakers' propensity to overmodify with color rather than size; the increase in overmodification in complex scenes; the increase in overmodification with atypical features; and the increase in specificity in nominal reference as a function of typicality. These findings cast a new light on the production of referring expressions: rather than being wastefully overinformative, reference is usefully redundant.

abstract = { Referring is one of the most basic and prevalent uses of language. How do speakers choose from the wealth of referring expressions at their disposal? Rational theories of language use have come under attack for decades for not being able to account for the seemingly irrational overinformativeness ubiquitous in referring expressions. Here we present a novel production model of referring expressions within the Rational Speech Act framework that treats speakers as agents that rationally trade off cost and informativeness of utterances. Crucially, we relax the assumption that informativeness is computed with respect to a deterministic Boolean semantics, in favor of a non-deterministic continuous semantics. This innovation allows us to capture a large number of seemingly disparate phenomena within one unified framework: the basic asymmetry in speakers' propensity to overmodify with color rather than size; the increase in overmodification in complex scenes; the increase in overmodification with atypical features; and the increase in specificity in nominal reference as a function of typicality. These findings cast a new light on the production of referring expressions: rather than being wastefully overinformative, reference is usefully redundant. } ,
author = { Degen, Judith and Hawkins, Robert X D and Graf, Caroline and Kreiss, Elisa and Goodman, Noah D } ,
doi = { 10.1037/rev0000186 } ,
issue = { 4 } ,
journal = { Psychological Review } ,
keywords = { language production, reference, overinformativeness, experimental pragmatics, Bayesian modeling } ,
pages = { 591–621 } ,
title = { When redundancy is useful: A Bayesian approach to `overinformative' referring expressions } ,
volume = { 127 } ,
year = { 2020 }
}

Contrastive inferences, whereby a listener pragmatically infers a speaker’s referential intention of a partial referring expression like the yellow by reasoning about other objects in the context, are notoriously unstable. We report a production-centric model of interpretation couched within the Rational Speech Act framework. Adjective production probabilities a listener expects for objects in a context drive the size of contrastive inferences: the greater the asymmetry in expectation for a speaker to use a pre-nominal adjective for the target rather than for competitors, the greater the listener’s resulting target preference. Modifier production probabilities were collected (Exp. 1) and used to make predictions about comprehension in an incremental decision task (Exp. 2). The model’s interpretation predictions are supported by the data. This account has the potential to explain the fluctuating appearance of contrastive inferences and shifts the explanatory focus away from contrastive inference towards online interpretation of referring expressions more broadly.

abstract = { Contrastive inferences, whereby a listener pragmatically infers a speaker’s referential intention of a partial referring expression like the yellow by reasoning about other objects in the context, are notoriously unstable. We report a production-centric model of interpretation couched within the Rational Speech Act framework. Adjective production probabilities a listener expects for objects in a context drive the size of contrastive inferences: the greater the asymmetry in expectation for a speaker to use a pre-nominal adjective for the target rather than for competitors, the greater the listener’s resulting target preference. Modifier production probabilities were collected (Exp. 1) and used to make predictions about comprehension in an incremental decision task (Exp. 2). The model’s interpretation predictions are supported by the data. This account has the potential to explain the fluctuating appearance of contrastive inferences and shifts the explanatory focus away from contrastive inference towards online interpretation of referring expressions more broadly. } ,
author = { Kreiss, Elisa and Degen, Judith } ,
booktitle = { Proceedings of the 42nd Annual Meeting of the Cognitive Science Society } ,
keywords = { contrastive inference; RSA; typicality; incremental processing } ,
title = { Production Expectations Modulate Contrastive Inference } ,
volume = { 42 } ,
year = { 2020 }
}

Studies addressing the question of whether scalar inferences generally incur a processing cost have yielded conflicting results. Constraint-based accounts, which seek to unify these conflicting results, make a prediction which we test here: the probability of an interpretation and the speed with which it is processed depends on the contextual support it receives. We manipulated contextual support for the scalar inference in two truth-value judgment experiments by manipulating a lexical feature (presence of partitive “of the”) and a pragmatic feature (the implicit Question Under Discussion). Participants’ responder type – whether their majority response was pragmatic (reflecting the inference) or literal (reflecting its absence) – was the main predictor of response times: pragmatic responses were faster than literal responses when generated by pragmatic responders; the reverse was true for literal responders. We interpret this as further evidence against costly inference accounts and in support of constraint-based accounts of pragmatic processing.

abstract = { Studies addressing the question of whether scalar inferences generally incur a processing cost have yielded conflicting results. Constraint-based accounts, which seek to unify these conflicting results, make a prediction which we test here: the probability of an interpretation and the speed with which it is processed depends on the contextual support it receives. We manipulated contextual support for the scalar inference in two truth-value judgment experiments by manipulating a lexical feature (presence of partitive “of the”) and a pragmatic feature (the implicit Question Under Discussion). Participants’ responder type – whether their majority response was pragmatic (reflecting the inference) or literal (reflecting its absence) – was the main predictor of response times: pragmatic responses were faster than literal responses when generated by pragmatic responders; the reverse was true for literal responders. We interpret this as further evidence against costly inference accounts and in support of constraint-based accounts of pragmatic processing. } ,
author = { Kursat, Leyla and Degen, Judith } ,
booktitle = { Proceedings of the 42nd Annual Meeting of the Cognitive Science Society } ,
keywords = { psycholinguistics; experimental pragmatics; scalar inference; Question Under Discussion } ,
title = { Probability and processing speed of scalar inferences is context-dependent } ,
volume = { 42 } ,
year = { 2020 }
}

Vocabulary growth and syntactic development are known to be highly correlated in early child language. What determines when words are acquired and how can this help us understand what drives early language development? We train an LSTM language model, known to detect syntactic regularities that are relevant for predicting the difficulty of words, on child-directed speech. We use the average surprisal of words for the model, which encodes sequential predictability, as a predictor for the age of acquisition of words in early child language. We compare this predictor to word frequency and others and find that average surprisal is a good predictor for the age of acquisition of function words and predicates beyond frequency, but not for nouns. Our approach provides insight into what makes a good model of early word learning, especially for words whose meanings rely heavily on linguistic context.

abstract = { Vocabulary growth and syntactic development are known to be highly correlated in early child language. What determines when words are acquired and how can this help us understand what drives early language development? We train an LSTM language model, known to detect syntactic regularities that are relevant for predicting the difficulty of words, on child-directed speech. We use the average surprisal of words for the model, which encodes sequential predictability, as a predictor for the age of acquisition of words in early child language. We compare this predictor to word frequency and others and find that average surprisal is a good predictor for the age of acquisition of function words and predicates beyond frequency, but not for nouns. Our approach provides insight into what makes a good model of early word learning, especially for words whose meanings rely heavily on linguistic context. } ,
author = { Portelance, Eva and Degen, Judith and Frank, Michael C. } ,
booktitle = { Proceedings of the 42nd Annual Meeting of the Cognitive Science Society } ,
keywords = { Language model; recurrent neural network; LSTM; language acquisition; age of acquisition; child directed speech; word learning. } ,
title = { Predicting Age of Acquisition in Early Word Learning Using Recurrent Neural Networks } ,
volume = { 42 } ,
year = { 2020 }
}

Pragmatic inferences often subtly depend on the presence or absence of linguistic features. For example, the presence of a partitive construction (of the) increases the strength of a so-called scalar inference: listeners perceive the inference that Chris did not eat all of the cookies to be stronger after hearing “Chris ate some of the cookies” than after hearing the same utterance without a partitive, “Chris ate some cookies”. In this work, we explore to what extent neural network sentence encoders can learn to predict the strength of scalar inferences. We first show that an LSTM-based sentence encoder trained on an English dataset of human inference strength ratings is able to predict ratings with high accuracy (r = 0.78). We then probe the model’s behavior using manually constructed minimal sentence pairs and corpus data. We find that the model inferred previously established associations between linguistic features and inference strength, suggesting that the model learns to use linguistic features to predict pragmatic inferences.

abstract = { Pragmatic inferences often subtly depend on the presence or absence of linguistic features. For example, the presence of a partitive construction (of the) increases the strength of a so-called scalar inference: listeners perceive the inference that Chris did not eat all of the cookies to be stronger after hearing “Chris ate some of the cookies” than after hearing the same utterance without a partitive, “Chris ate some cookies”. In this work, we explore to what extent neural network sentence encoders can learn to predict the strength of scalar inferences. We first show that an LSTM-based sentence encoder trained on an English dataset of human inference strength ratings is able to predict ratings with high accuracy (r = 0.78). We then probe the model’s behavior using manually constructed minimal sentence pairs and corpus data. We find that the model inferred previously established associations between linguistic features and inference strength, suggesting that the model learns to use linguistic features to predict pragmatic inferences. } ,
author = { Schuster, Sebastian and Chen, Yuxing and Degen, Judith } ,
booktitle = { Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics } ,
doi = { 1910.14254 } ,
keywords = { computational pragmatics; scalar implicature; neural networks } ,
pages = { 5387--5403 } ,
title = { Harnessing the linguistic signal to predict scalar inferences } ,
year = { 2020 }
}

Pragmatic theories of utterance interpretation share the assumption that listeners reason about alternative utterances that a speaker could have produced, but didn't. For such reasoning to be successful, listeners must have precise expectations about a speaker's production choices. This is at odds with the considerable variability across speakers that exists at all levels of linguistic representation. This tension can be reconciled by listeners adapting to the statistics of individual speakers. While linguistic adaptation is increasingly widely attested, semantic/pragmatic adaptation is underexplored. Moreover, what kind of representations listeners update during semantic/pragmatic adaptation – estimates of the speaker's lexicon, or estimates of the speaker's utterance preferences – remains poorly understood. In this work, we investigate semantic/pragmatic adaptation in the domain of uncertainty expressions like might and probably. In a series of web-based experiments, we find 1) that listeners vary in their expectations about a generic speaker's use of uncertainty expressions; 2) that listeners rapidly update their expectations about the use of uncertainty expressions after brief exposure to a speaker with a specific usage of uncertainty expressions; and 3) that listeners' interpretations of uncertainty expressions change after being exposed to a specific speaker. We present a novel computational model of semantic/pragmatic adaptation based on Bayesian belief updating and show, through a series of model comparisons, that semantic/pragmatic adaptation is best captured by listeners updating their beliefs both about the speaker's lexicon and their utterance preferences. This work has implications for both semantic theories of uncertainty expressions and psycholinguistic theories of adaptation: it highlights the need for dynamic semantic representations and provides evidence against accounts that cast adaptation as simple low-level priming.

abstract = { Pragmatic theories of utterance interpretation share the assumption that listeners reason about alternative utterances that a speaker could have produced, but didn't. For such reasoning to be successful, listeners must have precise expectations about a speaker's production choices. This is at odds with the considerable variability across speakers that exists at all levels of linguistic representation. This tension can be reconciled by listeners adapting to the statistics of individual speakers. While linguistic adaptation is increasingly widely attested, semantic/pragmatic adaptation is underexplored. Moreover, what kind of representations listeners update during semantic/pragmatic adaptation – estimates of the speaker's lexicon, or estimates of the speaker's utterance preferences – remains poorly understood. In this work, we investigate semantic/pragmatic adaptation in the domain of uncertainty expressions like might and probably. In a series of web-based experiments, we find 1) that listeners vary in their expectations about a generic speaker's use of uncertainty expressions; 2) that listeners rapidly update their expectations about the use of uncertainty expressions after brief exposure to a speaker with a specific usage of uncertainty expressions; and 3) that listeners' interpretations of uncertainty expressions change after being exposed to a specific speaker. We present a novel computational model of semantic/pragmatic adaptation based on Bayesian belief updating and show, through a series of model comparisons, that semantic/pragmatic adaptation is best captured by listeners updating their beliefs both about the speaker's lexicon and their utterance preferences. This work has implications for both semantic theories of uncertainty expressions and psycholinguistic theories of adaptation: it highlights the need for dynamic semantic representations and provides evidence against accounts that cast adaptation as simple low-level priming. } ,
author = { Schuster, Sebastian and Degen, Judith } ,
doi = { 10.1016/j.cognition.2020.104285 } ,
journal = { Cognition } ,
keywords = { adaptation; language comprehension; experimental pragmatics; Bayesian cognitive modeling; uncertainty expressions } ,
pages = { 104285 } ,
title = { I know what you’re probably going to say: Listener adaptation to variable use of uncertainty expressions } ,
volume = { 203 } ,
year = { 2020 }
}

Two contents of evaluative adjective sentences, like Kim was smart to watch the movie, are the prejacent (that Kim watched the movie) and the generalization (that the degree to which Kim watching the movie is smart was higher than the contextual standard of smart). The prejacent is standardly analyzed as a presupposition (e.g., Norrick 1978; Barker 2002; Oshima 2009; Kertz 2010). This paper argues against such analyses of the prejacent because, among other things, they do not capture an interaction between the prejacent and the generalization that has not yet been observed for projective content: when the prejacent projects, the generalization does not, and when the prejacent does not project, the generalization does. We develop an analysis according to which the prejacent is not a lexically specified presupposition but is projective to the extent that it is not at-issue with respect to the question addressed by the utterance of the evaluative adjective sentence. In addition to capturing the interaction between the prejacent and the generalization, our question-based projection analysis extends previous such analyses (e.g., Beaver & Clark 2008; Beaver et al. 2017; Simons et al. 2017) by incorporating a novel constraint on the question addressed by an utterance: the more the interpreter takes the truth of content c to follow from the common ground a priori, the less likely the question is about c. We provide experimental evidence for the analysis and argue that it improves on that of Karttunen et al. (2014), according to which evaluative adjectives are ambiguous.

abstract = { Two contents of evaluative adjective sentences, like Kim was smart to watch the movie, are the prejacent (that Kim watched the movie) and the generalization (that the degree to which Kim watching the movie is smart was higher than the contextual standard of smart). The prejacent is standardly analyzed as a presupposition (e.g., Norrick 1978; Barker 2002; Oshima 2009; Kertz 2010). This paper argues against such analyses of the prejacent because, among other things, they do not capture an interaction between the prejacent and the generalization that has not yet been observed for projective content: when the prejacent projects, the generalization does not, and when the prejacent does not project, the generalization does. We develop an analysis according to which the prejacent is not a lexically specified presupposition but is projective to the extent that it is not at-issue with respect to the question addressed by the utterance of the evaluative adjective sentence. In addition to capturing the interaction between the prejacent and the generalization, our question-based projection analysis extends previous such analyses (e.g., Beaver & Clark 2008; Beaver et al. 2017; Simons et al. 2017) by incorporating a novel constraint on the question addressed by an utterance: the more the interpreter takes the truth of content c to follow from the common ground a priori, the less likely the question is about c. We provide experimental evidence for the analysis and argue that it improves on that of Karttunen et al. (2014), according to which evaluative adjectives are ambiguous. } ,
author = { Tonhauser, Judith and de Marneffe, Marie-Catherine and Degen, Judith } ,
doi = { 10.5334/gjgl.701 } ,
journal = { Glossa: a journal of general linguistics } ,
number = { 1 } ,
pages = { 87 } ,
title = { Evaluative adjective sentences: A question-based analysis of projection } ,
volume = { 5 } ,
year = { 2020 }
}

Modern semantic analyses of epistemic language (incl. the modal must) can be characterized by the ‘credence assumption’: speakers have full certainty regarding the propositions that structure their epistemic states. Intuitively, however: a) speakers have graded, rather than categorical, commitment to these propositions, which are often never fully and explicitly articulated; b) listeners have higher- order uncertainty about this speaker uncertainty; c) must φ is utensed to communicate speaker commitment to some conclusion φ and to indicate speaker commitment to the premises that condition the conclusion. I explore the consequences of relaxing the credence assumption by extending the argument system semantic framework first proposed by Stone (1994) to a Bayesian probabilistic framework of modeling pragmatic interpretation (Goodman and Frank, 2016).

abstract = { Modern semantic analyses of epistemic language (incl. the modal must) can be characterized by the ‘credence assumption’: speakers have full certainty regarding the propositions that structure their epistemic states. Intuitively, however: a) speakers have graded, rather than categorical, commitment to these propositions, which are often never fully and explicitly articulated; b) listeners have higher- order uncertainty about this speaker uncertainty; c) must φ is utensed to communicate speaker commitment to some conclusion φ and to indicate speaker commitment to the premises that condition the conclusion. I explore the consequences of relaxing the credence assumption by extending the argument system semantic framework first proposed by Stone (1994) to a Bayesian probabilistic framework of modeling pragmatic interpretation (Goodman and Frank, 2016). } ,
author = { Waldon, Brandon } ,
booktitle = { Proceedings of Probability and Meaning (PaM2020) } ,
pages = { 34--40 } ,
title = { Linguistic interpretation as inference under argument system uncertainty: the case of epistemic must } ,
year = { 2020 }
}

Scalar inferences are commonly assumed to involve both literal semantic interpretation and social cognitive reasoning. However, the precise way to characterize listeners’ representation of context - including the space of possible utterance alternatives as well as the space of possible conventional meanings associated with linguistic forms - is a matter of ongoing debate. We report a partial replication of a scalar inference priming study by Rees and Bott (2018), introducing a novel baseline condition against which to compare behavior across different priming treatments. We also investigate the effect of raising participants’ awareness of communicatively stronger alternatives that explicitly encode an exhaustive meaning (e.g. some but not all with respect to some). Our results suggest that exhaustive alternatives (which are ‘symmetric’ to canonical alternatives) can modulate the availability and strength of scalar inferences, and that semantic uncertainty is an independent channel through which scalar inferences are modulated. We discuss implications for theories of pragmatic competence.

abstract = { Scalar inferences are commonly assumed to involve both literal semantic interpretation and social cognitive reasoning. However, the precise way to characterize listeners’ representation of context - including the space of possible utterance alternatives as well as the space of possible conventional meanings associated with linguistic forms - is a matter of ongoing debate. We report a partial replication of a scalar inference priming study by Rees and Bott (2018), introducing a novel baseline condition against which to compare behavior across different priming treatments. We also investigate the effect of raising participants’ awareness of communicatively stronger alternatives that explicitly encode an exhaustive meaning (e.g. some but not all with respect to some). Our results suggest that exhaustive alternatives (which are ‘symmetric’ to canonical alternatives) can modulate the availability and strength of scalar inferences, and that semantic uncertainty is an independent channel through which scalar inferences are modulated. We discuss implications for theories of pragmatic competence. } ,
author = { Waldon, Brandon and Degen, Judith } ,
booktitle = { Proceedings of the 42nd Annual Meeting of the Cognitive Science Society } ,
keywords = { experimental pragmatics; implicature; priming; adaptation; computational pragmatics } ,
title = { Symmetric alternatives and semantic uncertainty modulate scalar inference } ,
volume = { 42 } ,
year = { 2020 }
}

Truth Value Judgment Task experiments (TVJTs) are a common means of investigating pragmatic competence, particularly with regards to scalar inference. We present a novel quantitative linking function from pragmatic competence to participant behavior on TVJTs, based upon a Bayesian probabilistic model of linguistic production. Our model captures a range of observed phenomena on TVJTs, including intermediate responses on a non-binary scale, population and individual-level variation, participant endorsement of false utterances, and variation in response due to so-called scalar diversity.

abstract = { Truth Value Judgment Task experiments (TVJTs) are a common means of investigating pragmatic competence, particularly with regards to scalar inference. We present a novel quantitative linking function from pragmatic competence to participant behavior on TVJTs, based upon a Bayesian probabilistic model of linguistic production. Our model captures a range of observed phenomena on TVJTs, including intermediate responses on a non-binary scale, population and individual-level variation, participant endorsement of false utterances, and variation in response due to so-called scalar diversity. } ,
article = { 3 } ,
author = { Waldon, Brandon and Degen, Judith } ,
booktitle = { Proceedings of the Society for Computation in Linguistics } ,
doi = { 10.7275/sg32-aq30 } ,
keywords = { psycholinguistics; truth value judgement task; pragmatics; linking function } ,
number = { 1 } ,
pages = { 10--19 } ,
title = { Modeling Behavior in Truth Value Judgment Task Experiments } ,
volume = { 3 } ,
year = { 2020 }
}

Processing language requires integrating information from multiple sources, including context, world knowledge, and the linguistic signal itself. How is this information integrated? A range of positions on the issue is possible, spanned by two extreme positions: extreme informational privilege -- certain types of information are processed earlier in online processing and weighted most heavily in the resulting utterance interpretation; and extreme parallelism -- all information is processed in parallel and weighted equally in the resulting interpretation. In reviewing the current empirical landscape on scalar implicature processing, the chapter argues for a constraint-based approach to pragmatic processing, which is closer in spirit to the parallelism account than the informational privilege account. The approach is also extended to other pragmatic phenomena.

abstract = { Processing language requires integrating information from multiple sources, including context, world knowledge, and the linguistic signal itself. How is this information integrated? A range of positions on the issue is possible, spanned by two extreme positions: extreme informational privilege -- certain types of information are processed earlier in online processing and weighted most heavily in the resulting utterance interpretation; and extreme parallelism -- all information is processed in parallel and weighted equally in the resulting interpretation. In reviewing the current empirical landscape on scalar implicature processing, the chapter argues for a constraint-based approach to pragmatic processing, which is closer in spirit to the parallelism account than the informational privilege account. The approach is also extended to other pragmatic phenomena. } ,
author = { Degen, Judith and Tanenhaus, Michael K. } ,
booktitle = { Oxford Handbook of Experimental Semantics and Pragmatics } ,
chapter = { 3 } ,
editor = { Cummins, Chris and Katsos, Napoleon } ,
isbn = { 9780198791768 } ,
publisher = { Oxford University Press } ,
title = { Constraint-based pragmatic processing } ,
year = { 2019 }
}

We present a new experimental paradigm for investigating lexical expressions that conveydifferent strengths of speaker commitment. Specifically, we compare different evidentialcontexts for using modal devices, epistemic discourse particles, and statements with noevidential markers at all, examining the extent to which listeners' interpretations ofcertain types of evidential words and their judgments about speaker commitment differ instrength. We also probe speakers' production preferences for these different devices undervarying evidential circumstances. The results of our experiments shed new light on distinctions and controversies that play a key role in the current theoretical literature on thesemantics and pragmatics of modals and discourse particles. Our paradigm thus contributes to a domain of experimental research on evidential expressions that is only just takingshape at the crossroads of theoretical semantics/pragmatics and psycholinguistics; weprovide a potential starting point for approaching theoretical debates on the nature ofmodal evidential expressions from an experimental and context-oriented perspective.

abstract = { We present a new experimental paradigm for investigating lexical expressions that conveydifferent strengths of speaker commitment. Specifically, we compare different evidentialcontexts for using modal devices, epistemic discourse particles, and statements with noevidential markers at all, examining the extent to which listeners' interpretations ofcertain types of evidential words and their judgments about speaker commitment differ instrength. We also probe speakers' production preferences for these different devices undervarying evidential circumstances. The results of our experiments shed new light on distinctions and controversies that play a key role in the current theoretical literature on thesemantics and pragmatics of modals and discourse particles. Our paradigm thus contributes to a domain of experimental research on evidential expressions that is only just takingshape at the crossroads of theoretical semantics/pragmatics and psycholinguistics; weprovide a potential starting point for approaching theoretical debates on the nature ofmodal evidential expressions from an experimental and context-oriented perspective. } ,
author = { Degen, Judith and Trotzke, Andreas and Scontras, Gregory and Wittenberg, Eva and Goodman, Noah D } ,
doi = { 10.1016/j.pragma.2018.11.015 } ,
file = { :Users/judithdegen/cogsci/papers{\_}misc/2019{\_}DegenTrotzkeEtAl.pdf:pdf } ,
issn = { 0378-2166 } ,
journal = { Journal of Pragmatics } ,
keywords = { discourse particles,english,evidentials,german,modals,psycholinguistics } ,
pages = { 33--48 } ,
publisher = { Elsevier Ltd } ,
title = { Definitely, maybe: A new experimental paradigm for investigating the pragmatics of evidential devices across languages } ,
volume = { 140 } ,
year = { 2019 }
}

The past 15 years have seen increasing experimental investigations of core pragmatic questions in the ever more active and lively field of experimental pragmatics. Within experimental pragmatics, many of the core questions have relied on the operationalization of the theoretical notion of `implicature rate'. Implicature rate based results have informed the work on acquisition, online processing, and scalar diversity, inter alia. Implicature rate has typically been quantified as the proportion of `pragmatic' judgments in two-alternative forced choice truth value judgment tasks. Despite its theoretical importance, this linking hypothesis from implicature rate to behavioral responses has never been extensively tested. Here we show that two factors dramatically affect the `implicature rate' inferred from truth value judgment tasks: a) the number of responses provided to participants; and b) the linking hypothesis about what constitutes a `pragmatic' judgment. We argue that it is time for the field of experimental pragmatics to engage more seriously with its foundational assumptions about how theoretical notions map onto behaviorally measurable quantities, and present a sketch of an alternative linking hypothesis that derives behavior in truth value judgment tasks from probabilistic utterance expectations.

abstract = { The past 15 years have seen increasing experimental investigations of core pragmatic questions in the ever more active and lively field of experimental pragmatics. Within experimental pragmatics, many of the core questions have relied on the operationalization of the theoretical notion of `implicature rate'. Implicature rate based results have informed the work on acquisition, online processing, and scalar diversity, inter alia. Implicature rate has typically been quantified as the proportion of `pragmatic' judgments in two-alternative forced choice truth value judgment tasks. Despite its theoretical importance, this linking hypothesis from implicature rate to behavioral responses has never been extensively tested. Here we show that two factors dramatically affect the `implicature rate' inferred from truth value judgment tasks: a) the number of responses provided to participants; and b) the linking hypothesis about what constitutes a `pragmatic' judgment. We argue that it is time for the field of experimental pragmatics to engage more seriously with its foundational assumptions about how theoretical notions map onto behaviorally measurable quantities, and present a sketch of an alternative linking hypothesis that derives behavior in truth value judgment tasks from probabilistic utterance expectations. } ,
author = { Jasbi, Masoud and Waldon, Brandon and Degen, Judith } ,
doi = { 10.3389/fpsyg.2019.00189 } ,
issn = { 1664-1078 } ,
journal = { Frontiers in Psychology } ,
keywords = { scalar implicature, methodology, linking hypothesis, experimental pragmatics, truth value judgment task } ,
pages = { 189 } ,
title = { Linking hypothesis and number of response options modulate inferred scalar implicature rate } ,
volume = { 10 } ,
year = { 2019 }
}

Transmission of information by means of language is a potentially lossy process. Especially adjunct information, such as the graded degree of evidence, is a piece of information that seems prima facie likely to be distorted by reproduction noise. To investigate this issue, we present the results of a two- step iterated narration study: first, we collected a corpus of 250 crime story reproductions that were produced in parallel reproduction chains of 5 generations in depth, for 5 different seed stories; a second separate large-scale experiment then targeted readers’ interpretation of these reproductions. Crucially, strength of evidence for the guilt of each story’s suspect(s) was manipulated in the initial seed stories. Across generations, readers’ guilt perceptions decreased when the evidence was originally strong, but remained stable when evidence was originally weak. Analysis of linguistic measures revealed that dissimilarity between a seed story and its reproduction, story length, and amount of hedging language affected the readers’ own guilt perception and the readers’ attribution of guilt perception to the author differently. The results provide evidence that evidential information indeed influences guilt perception in complex ways.

abstract = { Transmission of information by means of language is a potentially lossy process. Especially adjunct information, such as the graded degree of evidence, is a piece of information that seems prima facie likely to be distorted by reproduction noise. To investigate this issue, we present the results of a two- step iterated narration study: first, we collected a corpus of 250 crime story reproductions that were produced in parallel reproduction chains of 5 generations in depth, for 5 different seed stories; a second separate large-scale experiment then targeted readers’ interpretation of these reproductions. Crucially, strength of evidence for the guilt of each story’s suspect(s) was manipulated in the initial seed stories. Across generations, readers’ guilt perceptions decreased when the evidence was originally strong, but remained stable when evidence was originally weak. Analysis of linguistic measures revealed that dissimilarity between a seed story and its reproduction, story length, and amount of hedging language affected the readers’ own guilt perception and the readers’ attribution of guilt perception to the author differently. The results provide evidence that evidential information indeed influences guilt perception in complex ways. } ,
author = { Kreiss, Elisa and Franke, Michael and Degen, Judith } ,
booktitle = { Proceedings of the 41st Annual Conference of the Cognitive Science Society } ,
keywords = { experimental pragmatics; iterated narration; transmission chains; uncertain evidence } ,
title = { Uncertain evidence statements and guilt perception in iterative reproductions of crime stories } ,
year = { 2019 }
}

Speakers exhibit variability in their choice between uncertainty expressions such as might and probably. Recent work has found that listeners cope with such variability by updating their expectations about how a specific speaker uses uncertainty expressions when interacting with a single speaker. However, it is still unclear to what extent listeners form speaker-specific expectations for multiple speakers and to what extent listeners are adapting to a situation independent of the speakers. Here, we take a first step towards answering these questions. In Experiment 1, listeners formed speaker-specific expectations after being exposed to two speakers whose use of uncertainty expressions differed. In Experiment 2, listeners who were exposed to two speakers with identical use of uncertainty expressions formed considerably stronger expectations than in Experiment 1. This suggests that listeners form both speaker-specific and situation-specific expectations. We discuss the implications of these results for theories of adaptation.

abstract = { Speakers exhibit variability in their choice between uncertainty expressions such as might and probably. Recent work has found that listeners cope with such variability by updating their expectations about how a specific speaker uses uncertainty expressions when interacting with a single speaker. However, it is still unclear to what extent listeners form speaker-specific expectations for multiple speakers and to what extent listeners are adapting to a situation independent of the speakers. Here, we take a first step towards answering these questions. In Experiment 1, listeners formed speaker-specific expectations after being exposed to two speakers whose use of uncertainty expressions differed. In Experiment 2, listeners who were exposed to two speakers with identical use of uncertainty expressions formed considerably stronger expectations than in Experiment 1. This suggests that listeners form both speaker-specific and situation-specific expectations. We discuss the implications of these results for theories of adaptation. } ,
author = { Schuster, Sebastian and Degen, Judith } ,
booktitle = { Proceedings of the 41st Annual Conference of the Cognitive Science Society } ,
keywords = { psycholinguistics; semantics; pragmatics; adaptation; uncertainty expressions } ,
title = { Speaker-specific adaptation to variable use of uncertainty expressions } ,
year = { 2019 }
}

Scontras et al. (2017) present experimental evidence demonstrating that the best predictor of adjective ordering preferences in the English noun phrase is the subjectivity of the property named by any given adjective: less subjective adjectives are preferred linearly closer to the nouns they modify. The current work builds on this empirical finding by proposing that the reason subjectivity predicts adjective ordering preferences has to do with the hierarchical structure of nominal modification. Adjectives that are linearly closer to the modified noun are often structurally closer, composing with the noun before adjectives that are farther away. Pressures from successful reference resolution dictate that less subjective, more useful adjectives contribute their meaning to the resulting nominal earlier, in an attempt to more effectively limit the reference search space.

abstract = { Scontras et al. (2017) present experimental evidence demonstrating that the best predictor of adjective ordering preferences in the English noun phrase is the subjectivity of the property named by any given adjective: less subjective adjectives are preferred linearly closer to the nouns they modify. The current work builds on this empirical finding by proposing that the reason subjectivity predicts adjective ordering preferences has to do with the hierarchical structure of nominal modification. Adjectives that are linearly closer to the modified noun are often structurally closer, composing with the noun before adjectives that are farther away. Pressures from successful reference resolution dictate that less subjective, more useful adjectives contribute their meaning to the resulting nominal earlier, in an attempt to more effectively limit the reference search space. } ,
author = { Scontras, Gregory and Degen, Judith and Goodman, Noah } ,
doi = { 10.3765/sp.12.7 } ,
journal = { Semantics and Pragmatics } ,
keywords = { adjective ordering, subjectivity, hierarchical structure, modification, reference resolution } ,
title = { On the grammatical source of adjective ordering preferences } ,
year = { 2019 }
}

Across languages, adjectives are subject to ordering restrictions. Recent research shows that these are predicted by adjective subjectivity, but the question remains open why this is the case. We first conduct a corpus study and not only replicate the subjectivity effect, but also find a previously undocumented effect of mutual information between adjectives and nouns. We then describe a rational model of adjective use in which listeners explicitly reason about judgments made by different speakers, formalizing the notion of subjectivity as agreement between speakers. We show that, once incremental processing is combined with memory limitations, our model predicts effects both of subjectivity and mutual information. We confirm the adequacy of our model by evaluating it on corpus data, finding that it correctly predicts ordering in unseen data with an accuracy of 96.2 %. This suggests that adjective ordering can be explained by general principles of human communication and language processing.

abstract = { Across languages, adjectives are subject to ordering restrictions. Recent research shows that these are predicted by adjective subjectivity, but the question remains open why this is the case. We first conduct a corpus study and not only replicate the subjectivity effect, but also find a previously undocumented effect of mutual information between adjectives and nouns. We then describe a rational model of adjective use in which listeners explicitly reason about judgments made by different speakers, formalizing the notion of subjectivity as agreement between speakers. We show that, once incremental processing is combined with memory limitations, our model predicts effects both of subjectivity and mutual information. We confirm the adequacy of our model by evaluating it on corpus data, finding that it correctly predicts ordering in unseen data with an accuracy of 96.2 %. This suggests that adjective ordering can be explained by general principles of human communication and language processing. } ,
author = { Hahn, Michael and Degen, Judith and Goodman, Noah and Jurafsky, Dan and Futrell, Richard } ,
booktitle = { Proceedings of the 40th Annual Conference of the Cognitive Science Society } ,
title = { An Information-Theoretic Explanation of Adjective Ordering Preferences } ,
year = { 2018 }
}

A common dependent measure used in visual-world eye-tracking experiments is the proportion of looks to a visually depicted object in a certain time window after the onset of the critical stimulus. When interpreting such data, a common assumption is that looks to the object reflect the listener’s belief that the object is the intended target referent. While this is intuitively plausible (at least for paradigms in which the task requires selecting a referent), relatively little is known about how exactly the proportion of looks to an object is related to a listener’s current belief about that object. Here, we test a simple, explicit linking hypothesis: the proportion of looks to an object reflects the probability that the listener assigns to the object being the target. To test this hypothesis, we supplement the eye-tracking data from Leffel, Xiang, and Kennedy (2016) with an offline incremental decision task to measure participants’ beliefs about the intended referent at various points in the unfolding sentence, and assess the extent to which these beliefs predict the eye-tracking data. The results suggest that the degree to which an object is believed to be the referent is only one factor that affects eye movements in referential tasks. Preliminary free production data we have collected for the scenes suggests that utterance expectations also play a role. We discuss methodological implications of these results for experimental linguistics.

abstract = { A common dependent measure used in visual-world eye-tracking experiments is the proportion of looks to a visually depicted object in a certain time window after the onset of the critical stimulus. When interpreting such data, a common assumption is that looks to the object reflect the listener’s belief that the object is the intended target referent. While this is intuitively plausible (at least for paradigms in which the task requires selecting a referent), relatively little is known about how exactly the proportion of looks to an object is related to a listener’s current belief about that object. Here, we test a simple, explicit linking hypothesis: the proportion of looks to an object reflects the probability that the listener assigns to the object being the target. To test this hypothesis, we supplement the eye-tracking data from Leffel, Xiang, and Kennedy (2016) with an offline incremental decision task to measure participants’ beliefs about the intended referent at various points in the unfolding sentence, and assess the extent to which these beliefs predict the eye-tracking data. The results suggest that the degree to which an object is believed to be the referent is only one factor that affects eye movements in referential tasks. Preliminary free production data we have collected for the scenes suggests that utterance expectations also play a role. We discuss methodological implications of these results for experimental linguistics. } ,
author = { Qing, Ciyang and Lassiter, Daniel and Degen, Judith } ,
booktitle = { Proceedings of the 40th Annual Conference of the Cognitive Science Society } ,
keywords = { eye-tracking,gradable adjectives,imprecision,linking functions,pragmatics,semantics,vagueness,visual world } ,
title = { What do eye movements in the visual world reflect? A case study from adjectives } ,
year = { 2018 }
}

Projective content is utterance content that a speaker may be taken to be committed to even when the expression associated with the content occurs embedded under an entailment-canceling operator (e.g., Chierchia & McConnell-Ginet, 1990). It has long been observed that projective content varies in how projective it is (e.g., Karttunen, 1971; Simons, 2001; Abusch, 2010), though preliminary experimental research has been able to confirm only some of the intuitions about projection variability (e.g., Smith & Hall, 2011; Xue & Onea, 2011). Given the sparse empirical evidence for projection variability, the first goal of this paper was to investigate projection variability for projective content associated with 19 expressions of American English. The second goal was to explore the hypothesis, called the Gradient Projection Principle, that content projects to the extent that it is not at-issue. The findings of two pairs of experiments provide robust empirical evidence for projection variability and for the Gradient Projection Principle. We show that many analyses of projection cannot account for the observed projection variability and discuss the implications of our finding that projective content varies in its at-issueness for an empirically adequate analysis of projection.

abstract = { Projective content is utterance content that a speaker may be taken to be committed to even when the expression associated with the content occurs embedded under an entailment-canceling operator (e.g., Chierchia & McConnell-Ginet, 1990). It has long been observed that projective content varies in how projective it is (e.g., Karttunen, 1971; Simons, 2001; Abusch, 2010), though preliminary experimental research has been able to confirm only some of the intuitions about projection variability (e.g., Smith & Hall, 2011; Xue & Onea, 2011). Given the sparse empirical evidence for projection variability, the first goal of this paper was to investigate projection variability for projective content associated with 19 expressions of American English. The second goal was to explore the hypothesis, called the Gradient Projection Principle, that content projects to the extent that it is not at-issue. The findings of two pairs of experiments provide robust empirical evidence for projection variability and for the Gradient Projection Principle. We show that many analyses of projection cannot account for the observed projection variability and discuss the implications of our finding that projective content varies in its at-issueness for an empirically adequate analysis of projection. } ,
author = { Tonhauser, Judith and Beaver, David I and Degen, Judith } ,
doi = { 10.1093/jos/ffy007 } ,
journal = { Journal of Semantics } ,
number = { 3 } ,
pages = { 495--542 } ,
title = { How projective is projective content? Gradience in projectivity and at-issueness } ,
volume = { 35 } ,
year = { 2018 }
}

From English to Hungarian to Mokilese, speakers exhibit strong ordering preferences in multi-adjective strings: “the big blue box” sounds far more natural than “the blue big box.” We show that an adjective’s distance from the modified noun is predicted not by a rigid syntax, but by the adjective’s meaning: less subjective adjectives occur closer to the nouns they modify. This finding provides an example of a broad linguistic universal—adjective ordering preferences—emerging from general properties of cognition.

abstract = { From English to Hungarian to Mokilese, speakers exhibit strong ordering preferences in multi-adjective strings: “the big blue box” sounds far more natural than “the blue big box.” We show that an adjective’s distance from the modified noun is predicted not by a rigid syntax, but by the adjective’s meaning: less subjective adjectives occur closer to the nouns they modify. This finding provides an example of a broad linguistic universal—adjective ordering preferences—emerging from general properties of cognition. } ,
annote = { doi: 10.1162/OPMI{\_}a{\_}00005 } ,
author = { Scontras, Gregory and Degen, Judith and Goodman, Noah D } ,
doi = { 10.1162/OPMI_a_00005 } ,
issue = { 1 } ,
journal = { Open Mind } ,
pages = { 53--66 } ,
publisher = { MIT Press } ,
title = { Subjectivity Predicts Adjective Ordering Preferences } ,
volume = { 1 } ,
year = { 2017 }
}

{\textcopyright} 2016 Cognitive Science Society, Inc.Two visual world experiments investigated the processing of the implicature associated with some using a "gumball paradigm." On each trial, participants saw an image of a gumball machine with an upper chamber with orange and blue gumballs and an empty lower chamber. Gumballs dropped to the lower chamber, creating a contrast between a partitioned set of gumballs of one color and an unpartitioned set of the other. Participants then evaluated spoken statements, such as "You got some of the blue gumballs." Experiment 1 investigated the time course of the pragmatic enrichment from some to not all when the only utterance alternatives available to refer to the different sets were some and all. In Experiment 2, the number terms two, three, four, and five were also included in the set of alternatives. Scalar implicatures were delayed relative to the interpretation of literal statements with all only when number terms were available. The results are interpreted as evidence for a constraint-based account of scalar implicature processing.

abstract = { {\textcopyright} 2016 Cognitive Science Society, Inc.Two visual world experiments investigated the processing of the implicature associated with some using a "gumball paradigm." On each trial, participants saw an image of a gumball machine with an upper chamber with orange and blue gumballs and an empty lower chamber. Gumballs dropped to the lower chamber, creating a contrast between a partitioned set of gumballs of one color and an unpartitioned set of the other. Participants then evaluated spoken statements, such as "You got some of the blue gumballs." Experiment 1 investigated the time course of the pragmatic enrichment from some to not all when the only utterance alternatives available to refer to the different sets were some and all. In Experiment 2, the number terms two, three, four, and five were also included in the set of alternatives. Scalar implicatures were delayed relative to the interpretation of literal statements with all only when number terms were available. The results are interpreted as evidence for a constraint-based account of scalar implicature processing. } ,
author = { Degen, Judith and Tanenhaus, Michael K. } ,
doi = { 10.1111/cogs.12227 } ,
issn = { 15516709 } ,
journal = { Cognitive Science } ,
keywords = { Alternatives,Eye-tracking,Pragmatics,Quantifiers,Scalar implicature } ,
number = { 1 } ,
pages = { 172--201 } ,
title = { Availability of Alternatives and the Processing of Scalar Implicatures: A Visual World Eye-Tracking Study } ,
volume = { 40 } ,
year = { 2016 }
}

Nominal reference is very flexible—the same object may be called a dalmatian, a dog, or an animal when all are literally true. What accounts for the choices that speakers make in how they refer to objects? The addition of modifiers (e.g. big dog) has been extensively explored in the literature, but fewer studies have explored the choice of noun, including its level of abstraction. We collected freely produced referring expressions in a multi-player reference game experiment, where we manipulated the object’s context. We find that utterance choice is affected by the contextual informativeness of a description, its length and frequency, and the typicality of the object for that description. Finally, we show how these factors naturally enter into a formal model of production within the Rational Speech-Acts framework, and that the resulting model predicts our quantitative production data.

abstract = { Nominal reference is very flexible—the same object may be called a dalmatian, a dog, or an animal when all are literally true. What accounts for the choices that speakers make in how they refer to objects? The addition of modifiers (e.g. big dog) has been extensively explored in the literature, but fewer studies have explored the choice of noun, including its level of abstraction. We collected freely produced referring expressions in a multi-player reference game experiment, where we manipulated the object’s context. We find that utterance choice is affected by the contextual informativeness of a description, its length and frequency, and the typicality of the object for that description. Finally, we show how these factors naturally enter into a formal model of production within the Rational Speech-Acts framework, and that the resulting model predicts our quantitative production data. } ,
address = { Austin, TX } ,
author = { Graf, Caroline and Degen, Judith and Hawkins, Robert X D and Goodman, Noah D } ,
booktitle = { Proceedings of the 38th Annual Conference of the Cognitive Science Society } ,
editor = { Papafragou, A. and Grodner, D. and Mirman, D. and Trueswell, J.C. } ,
keywords = { intropsychling } ,
mendeley-tags = { intropsychling } ,
pages = { 2261--2266 } ,
publisher = { Cognitive Science Society } ,
title = { Animal, dog, or dalmatian? Level of abstraction in nominal referring expressions } ,
year = { 2016 }
}

Linguistic meaning has long been recognized to be highly context-dependent. Quantifiers like many and some provide a particularly clear example of context-dependence. For example, the interpretation of quantifiers requires listeners to determine the relevant domain and scale. We focus on another type of context-dependence that quantifiers share with other lexical items: talker variability. Different talkers might use quantifiers with different interpretations in mind. We used a web-based crowdsourcing paradigm to study participants' expectations about the use of many and some based on recent exposure. We first established that the mapping of some and many onto quantities (candies in a bowl) is variable both within and between participants. We then examined whether and how listeners' expectations about quantifier use adapts with exposure to talkers who use quantifiers in different ways. The results demonstrate that listeners can adapt to talker-specific biases in both how often and with what intended meaning many and some are used.

abstract = { Linguistic meaning has long been recognized to be highly context-dependent. Quantifiers like many and some provide a particularly clear example of context-dependence. For example, the interpretation of quantifiers requires listeners to determine the relevant domain and scale. We focus on another type of context-dependence that quantifiers share with other lexical items: talker variability. Different talkers might use quantifiers with different interpretations in mind. We used a web-based crowdsourcing paradigm to study participants' expectations about the use of many and some based on recent exposure. We first established that the mapping of some and many onto quantities (candies in a bowl) is variable both within and between participants. We then examined whether and how listeners' expectations about quantifier use adapts with exposure to talkers who use quantifiers in different ways. The results demonstrate that listeners can adapt to talker-specific biases in both how often and with what intended meaning many and some are used. } ,
author = { Yildirim, Ilker and Degen, Judith and Tanenhaus, Michael K. and Jaeger, T. Florian } ,
doi = { 10.1016/j.jml.2015.08.003 } ,
issn = { 0749596X } ,
journal = { Journal of Memory and Language } ,
keywords = { Adaptation,Pragmatics,Quantifiers,Semantics,Talker-specificity } ,
pages = { 128--143 } ,
publisher = { Elsevier Inc. } ,
title = { Talker-specificity and adaptation in quantifier interpretation } ,
volume = { 87 } ,
year = { 2016 }
}

What makes a question useful? What makes an answer appropriate? In this paper, we formulate a family of increasingly sophisticated models of question-answer behavior within the Rational Speech Act framework. We compare these models based on three different pieces of evidence: first, we demonstrate how our answerer models capture a classic effect in psycholinguistics showing that an answerer’s level of informativeness varies with the inferred questioner goal, while keeping the question constant. Second, we jointly test the questioner and answerer components of our model based on empirical evidence from a question-answer reasoning game. Third, we examine a special case of this game to further distinguish among the questioner models. We find that sophisticated pragmatic reasoning is needed to account for some of the data. People can use questions to provide cues to the answerer about their interest, and can select answers that are informative about inferred interests.

abstract = { What makes a question useful? What makes an answer appropriate? In this paper, we formulate a family of increasingly sophisticated models of question-answer behavior within the Rational Speech Act framework. We compare these models based on three different pieces of evidence: first, we demonstrate how our answerer models capture a classic effect in psycholinguistics showing that an answerer’s level of informativeness varies with the inferred questioner goal, while keeping the question constant. Second, we jointly test the questioner and answerer components of our model based on empirical evidence from a question-answer reasoning game. Third, we examine a special case of this game to further distinguish among the questioner models. We find that sophisticated pragmatic reasoning is needed to account for some of the data. People can use questions to provide cues to the answerer about their interest, and can select answers that are informative about inferred interests. } ,
author = { Hawkins, Robert XD and Stuhlmuller, Andreas and Degen, Judith and Goodman, Noah D } ,
booktitle = { Proceedings of the 37th Annual Conference of the Cognitive Science Society } ,
title = { Why do you ask? Good questions provoke informative answers } ,
year = { 2015 }
}

A prevalent, but to date untested, assumption about lexicalized scalar implicatures such as those from 'some' to 'not all', is that they fall into the class of GCIs and as such, constitute a homogeneous class of highly regularized and context-independent implicatures. This paper reports a test of this assumption in which linguistically untrained participants’ implicature strength judgments were collected for naturally occurring utterances containing the word 'some' in a large-scale corpus-based web study. The results indicate that implicature strength is highly variable and systematically dependent on features of the linguistic context such as the partitive, determiner strength, and discourse accessibility. These results call into question the GCI status of scalar implicatures from 'some' to 'not all' and demonstrate the usefulness of corpora and web-based methods for challenging received wisdom, enriching the empirical landscape, and informing theory in pragmatics.

abstract = { A prevalent, but to date untested, assumption about lexicalized scalar implicatures such as those from 'some' to 'not all', is that they fall into the class of GCIs and as such, constitute a homogeneous class of highly regularized and context-independent implicatures. This paper reports a test of this assumption in which linguistically untrained participants’ implicature strength judgments were collected for naturally occurring utterances containing the word 'some' in a large-scale corpus-based web study. The results indicate that implicature strength is highly variable and systematically dependent on features of the linguistic context such as the partitive, determiner strength, and discourse accessibility. These results call into question the GCI status of scalar implicatures from 'some' to 'not all' and demonstrate the usefulness of corpora and web-based methods for challenging received wisdom, enriching the empirical landscape, and informing theory in pragmatics. } ,
author = { Degen, Judith } ,
doi = { 10.3765/sp.8.11 } ,
issn = { 1937-8912 } ,
journal = { Semantics and Pragmatics } ,
number = { 11 } ,
pages = { 1--55 } ,
publisher = { Semantics and Pragmatics } ,
title = { Investigating the distribution of 'some' (but not 'all') implicatures using corpora and web-based methods } ,
volume = { 8 } ,
year = { 2015 }
}

Three experiments investigated the processing of the implicature associated with some using a "gumball paradigm." On each trial, participants saw an image of a gumball machine with an upper chamber with 13 gumballs and an empty lower chamber. Gumballs then dropped to the lower chamber and participants evaluated statements, such as "You got some of the gumballs." Experiment 1 established that some is less natural for reference to small sets (1, 2, and 3 of the 13 gumballs) and unpartitioned sets (all 13 gumballs) compared to intermediate sets (6-8). Partitive some of was less natural than simple some when used with the unpartitioned set. In Experiment 2, including exact number descriptions lowered naturalness ratings for some with small sets but not for intermediate size sets and the unpartitioned set. In Experiment 3, the naturalness ratings from Experiment 2 predicted response times. The results are interpreted as evidence for a Constraint-Based account of scalar implicature processing and against both two-stage, Literal-First models and pragmatic Default models.

abstract = { Three experiments investigated the processing of the implicature associated with some using a "gumball paradigm." On each trial, participants saw an image of a gumball machine with an upper chamber with 13 gumballs and an empty lower chamber. Gumballs then dropped to the lower chamber and participants evaluated statements, such as "You got some of the gumballs." Experiment 1 established that some is less natural for reference to small sets (1, 2, and 3 of the 13 gumballs) and unpartitioned sets (all 13 gumballs) compared to intermediate sets (6-8). Partitive some of was less natural than simple some when used with the unpartitioned set. In Experiment 2, including exact number descriptions lowered naturalness ratings for some with small sets but not for intermediate size sets and the unpartitioned set. In Experiment 3, the naturalness ratings from Experiment 2 predicted response times. The results are interpreted as evidence for a Constraint-Based account of scalar implicature processing and against both two-stage, Literal-First models and pragmatic Default models. } ,
archiveprefix = { arXiv } ,
arxivid = { arXiv:1011.1669v3 } ,
author = { Degen, Judith and Tanenhaus, Michael K. } ,
doi = { 10.1111/cogs.12171 } ,
eprint = { arXiv:1011.1669v3 } ,
isbn = { 9788578110796 } ,
issn = { 03640213 } ,
journal = { Cognitive Science } ,
keywords = { Alternatives,Pragmatics,Quantifiers,Scalar implicature } ,
number = { 4 } ,
pages = { 667--710 } ,
pmid = { 25265993 } ,
title = { Processing scalar implicature A constraint-based approach } ,
volume = { 39 } ,
year = { 2015 }
}

World knowledge enters into pragmatic utterance interpreta- tion in complex ways, and may be defeasible in light of speak- ers' utterances. Yet there is to date a surprising lack of sys- tematic investigation into the role of world knowledge in prag- matic inference. In this paper, we show that a state-of-the-art model of pragmatic interpretation greatly overestimates the in- fluence of world knowledge on the interpretation of utterances like Some of the marbles sank. We extend the model to cap- ture the idea that the listener is uncertain about the background knowledge the speaker is bringing to the conversation. This extension greatly improves model predictions of listeners' in- terpretation and also makes good qualitative predictions about listeners' judgments of how ‘normal' the world is in light of a speaker's statement. Theoretical and methodological implica- tions are discussed. Keywords:

abstract = { World knowledge enters into pragmatic utterance interpreta- tion in complex ways, and may be defeasible in light of speak- ers' utterances. Yet there is to date a surprising lack of sys- tematic investigation into the role of world knowledge in prag- matic inference. In this paper, we show that a state-of-the-art model of pragmatic interpretation greatly overestimates the in- fluence of world knowledge on the interpretation of utterances like Some of the marbles sank. We extend the model to cap- ture the idea that the listener is uncertain about the background knowledge the speaker is bringing to the conversation. This extension greatly improves model predictions of listeners' in- terpretation and also makes good qualitative predictions about listeners' judgments of how ‘normal' the world is in light of a speaker's statement. Theoretical and methodological implica- tions are discussed. Keywords: } ,
address = { Austin, TX } ,
author = { Degen, Judith and Tessler, Michael Henry and Goodman, Noah D } ,
editor = { Noelle, D. C. and Dale, R. and Warlaumont, A. S. and Yoshimi, J. and Matlock, T. and Jennings, C. D. and Maglio, P. P. } ,
journal = { Proceedings of the 37th Annual Conference of the Cognitive Science Society } ,
keywords = { computational pragmatics,experimental pragmatics,how often do you,if not always,in water,liefs,now imagine read-,prior be-,prior beliefs,probably extremely often,scalar implicature,think marbles would sink,world knowledge } ,
number = { 2 } ,
pages = { 548--553 } ,
publisher = { Cognitive Science Society } ,
title = { Wonky worlds: Listeners revise world knowledge when utterances are odd } ,
year = { 2015 }
}

A rarely discussed but important issue in research on pragmatic inference is the choice of dependent measure for estimating the robustness of pragmatic inferences and their sensitivity to contextual manipulations. Here we present the results from three studies exploring the effect of contextual manipulations on scalar implicature. In all three studies we manipulate the salient question under discussion and the perceptual availability of relevant set sizes. The studies differ only in the dependent measure used: Exp. 1 uses truth judgements, Exp. 2 uses word probability ratings, and Exp. 3 uses a direct measure of sentence interpretation. We argue that the first two are effectively measures of production, and find they are sensitive to our contextual manipulations. In contrast the interpretation measure shows no effect of context. We argue that this methodologically troubling finding can be understood and predicted by using the framework of probabilistic pragmatics.

abstract = { A rarely discussed but important issue in research on pragmatic inference is the choice of dependent measure for estimating the robustness of pragmatic inferences and their sensitivity to contextual manipulations. Here we present the results from three studies exploring the effect of contextual manipulations on scalar implicature. In all three studies we manipulate the salient question under discussion and the perceptual availability of relevant set sizes. The studies differ only in the dependent measure used: Exp. 1 uses truth judgements, Exp. 2 uses word probability ratings, and Exp. 3 uses a direct measure of sentence interpretation. We argue that the first two are effectively measures of production, and find they are sensitive to our contextual manipulations. In contrast the interpretation measure shows no effect of context. We argue that this methodologically troubling finding can be understood and predicted by using the framework of probabilistic pragmatics. } ,
author = { Degen, Judith and Goodman, Noah D } ,
journal = { Proceedings of the 36th Annual Conference of the Cognitive Science Society } ,
keywords = { pragmatics,psycholinguistics,scalar implicature } ,
pages = { 397--402 } ,
title = { Lost your marbles? The puzzle of dependent measures in experimental pragmatics } ,
year = { 2014 }
}

In the face of underspecified utterances, listeners routinely and without much apparent effort make the right kinds of pragmatic inferences about a speaker's intended meaning. This dissertation investigates the processing of scalar impli- catures as a way of addressing how listeners perform this remarkable feat. In particular, the role of context in the processing of scalar implicatures from some to not all is explored. Contrary to the widely held assumption that scalar impli- catures are highly regularized, frequent, and relatively context-independent, this dissertation suggests that they are in fact relatively infrequent and highly context- dependent; both the robustness and the speed with which scalar implicatures from some to not all are computed are modulated by the probabilistic support that the implicature receives from multiple contextual cues. Scalar implicatures are found to be especially sensitive to the naturalness or expectedness of both scalar and non-scalar alternative utterances the speaker could have produced, but didn't. A novel contextualist account of scalar implicature processing that has roots in both constraint-based and information-theoretic accounts of language processing is proposed that provides a unifying explanation for a) the varying robustness of scalar implicatures across different contexts, b) the varying speed of scalar implica- tures across different contexts, and c) the speed and efficiency of communication.

abstract = { In the face of underspecified utterances, listeners routinely and without much apparent effort make the right kinds of pragmatic inferences about a speaker's intended meaning. This dissertation investigates the processing of scalar impli- catures as a way of addressing how listeners perform this remarkable feat. In particular, the role of context in the processing of scalar implicatures from some to not all is explored. Contrary to the widely held assumption that scalar impli- catures are highly regularized, frequent, and relatively context-independent, this dissertation suggests that they are in fact relatively infrequent and highly context- dependent; both the robustness and the speed with which scalar implicatures from some to not all are computed are modulated by the probabilistic support that the implicature receives from multiple contextual cues. Scalar implicatures are found to be especially sensitive to the naturalness or expectedness of both scalar and non-scalar alternative utterances the speaker could have produced, but didn't. A novel contextualist account of scalar implicature processing that has roots in both constraint-based and information-theoretic accounts of language processing is proposed that provides a unifying explanation for a) the varying robustness of scalar implicatures across different contexts, b) the varying speed of scalar implica- tures across different contexts, and c) the speed and efficiency of communication. } ,
author = { Degen, Judith } ,
journal = { University of Rochester } ,
title = { Alternatives in Pragmatic Reasoning } ,
year = { 2013 }
}

We present data from three experiments addressing how much theory of mind reasoning is involved in production and interpretation of ambiguous referential expressions in an artificial language task, and how this interacts with the cost and availability of alternative utterances. When an unambiguous alternative is not available, listeners tend to draw simple Quantity inferences reminiscent of scalar implicatures (Grice, 1975). When an unambiguous alternative is available, fewer inferences are observed, but gradiently more as the cost of unambiguous alternatives increase. We outline a novel game theoretic model of pragmatic reasoning based on probabilistic back-and-forth reasoning about interlocutors’ rational choices and beliefs. The model provides a good fit to the data and raises interesting issues for future research.

abstract = { We present data from three experiments addressing how much theory of mind reasoning is involved in production and interpretation of ambiguous referential expressions in an artificial language task, and how this interacts with the cost and availability of alternative utterances. When an unambiguous alternative is not available, listeners tend to draw simple Quantity inferences reminiscent of scalar implicatures (Grice, 1975). When an unambiguous alternative is available, fewer inferences are observed, but gradiently more as the cost of unambiguous alternatives increase. We outline a novel game theoretic model of pragmatic reasoning based on probabilistic back-and-forth reasoning about interlocutors’ rational choices and beliefs. The model provides a good fit to the data and raises interesting issues for future research. } ,
author = { Degen, Judith and Franke, Michael and Jäger, Gerhard } ,
booktitle = { Proceedings of the 35th Annual Meeting of the Cognitive Science Society } ,
title = { Cost-Based Pragmatic Inference about Referential Expressions } ,
year = { 2013 }
}

People’s representations of most and arguably all linguistic and non-linguistic categories are probabilistic. However, in lin- guistic theory, quantifier meanings have traditionally been de- fined set-theoretically in terms of categorical evaluation func- tions. In 4 “adaptation” experiments, we provide evidence for the alternative hypothesis that quantifiers are represented as probability distributions over scales (e.g., Zadeh, 1965). We manipulate exposure to different distributions of “some” and “many” and find that listeners adapt to those distributions, as predicted. Our results suggest that the interpretation of quanti- fiers is best modeled as a process involving rich, probabilistic representations.

abstract = { People’s representations of most and arguably all linguistic and non-linguistic categories are probabilistic. However, in lin- guistic theory, quantifier meanings have traditionally been de- fined set-theoretically in terms of categorical evaluation func- tions. In 4 “adaptation” experiments, we provide evidence for the alternative hypothesis that quantifiers are represented as probability distributions over scales (e.g., Zadeh, 1965). We manipulate exposure to different distributions of “some” and “many” and find that listeners adapt to those distributions, as predicted. Our results suggest that the interpretation of quanti- fiers is best modeled as a process involving rich, probabilistic representations. } ,
author = { Yildrim, Ilker and Degen, Judith and Tanenhaus, Michael J., and Jaeger, T. Florian } ,
booktitle = { Proceedings of the Annual Meeting of the Cognitive Science Society } ,
title = { Linguistic Variability and Adaptation in Quantifier Meanings } ,
volume = { 35 } ,
year = { 2013 }
}

author = { Degen, Judith and Franke, Michael } ,
booktitle = { Proceedings of the 16th Workshop on the Semantics and Pragmatics of Dialogue } ,
editor = { Brown-Schmidt, Sarah and Ginzburg, Jonathan and Larsson, S. } ,
pages = { 2 -- 11 } ,
title = { Optimal Reasoning About Referential Expressions } ,
year = { 2012 }
}

Scalar implicature has served as a test case for investigating the nature of inference processes in language comprehension. Specifically, the question of whether or not scalar implicatures are computed by default has been extensively investigated in recent years. We argue that the question of default is overly simplistic and propose instead to think of scalar implicature computation as a problem of optimal cue combination within a constraint-based framework. We provide evidence from three experiments supporting the view that multiple constraints of differing strength operate in parallel to provide probabilistic support for or against an implicature.

abstract = { Scalar implicature has served as a test case for investigating the nature of inference processes in language comprehension. Specifically, the question of whether or not scalar implicatures are computed by default has been extensively investigated in recent years. We argue that the question of default is overly simplistic and propose instead to think of scalar implicature computation as a problem of optimal cue combination within a constraint-based framework. We provide evidence from three experiments supporting the view that multiple constraints of differing strength operate in parallel to provide probabilistic support for or against an implicature. } ,
author = { Degen, Judith and Tanenhaus, Michael K. } ,
booktitle = { Proceedings of the 33rd Annual Conference of the Cognitive Science Society } ,
editor = { Carlson, L. and Hölscher, C. and Shipley, T. } ,
keywords = { experimental pragmatics,eye-tracking,scalar implicature,subitizing } ,
pages = { 3299--3304 } ,
title = { Making inferences: the case of scalar implicature processing } ,
year = { 2011 }
}