Search

The use of statistical inference in linguistics and related areas like psychology typically involves a binary decision: either reject or accept some null hypothesis using statistical significance testing. When statistical power is low, this frequentist data-analytic approach breaks down: null results are uninformative, and effect size estimates associated with significant results are overestimated. Using an example from psycholinguistics, several alternative approaches are demonstrated for reporting inconsistencies between the data and a theoretical prediction. The key here is to focus on committing to a falsifiable prediction, on quantifying uncertainty statistically, and learning to accept the fact that – in almost all practical data analysis situations – we can only draw uncertain conclusions from data, regardless of whether we manage to obtain statistical significance or not. A focus on uncertainty quantification is likely to lead to fewer excessively bold claims that, on closer investigation, may turn out to be not supported by the data.

How do we decide what to say to ensure our meanings will be understood? The Rational Speech Act model (RSA, Frank & Goodman, 2012) asserts that speakers plan what to say by comparing the informativity of words in a particular context. We present the first example of an RSA model of sentence level (who-did-what-to-whom) meanings. In these contexts, the set of possible messages must be abstracted from entities in common ground (people and objects) to possible events (Jane eats the apple, Marco peels the banana), with each word contributing unique semantic content. How do speakers accomplish the transformation from context to compositional, informative messages? In a communication game, participants described transitive events (e.g. Jane pets the dog), with only two words, in contexts where two words either were or were not enough to uniquely identify an event. Adults chose utterances matching the predictions of the RSA even when there was no possible fully 'successful' utterance. Thus we show that adults’ communicative behavior can be described by a model that accommodates informativity in context, beyond the set of possible entities in common ground. This study provides the first evidence that adults' language production is affected, at the level of argument structure, by the graded informativity of possible utterances in context, and suggests that full- blown natural speech may result from speakers who model and adapt to the listener’s needs.

In a direction-giving task, we examine whether a high-level constraint—the task partners’ relative body alignment—influences spatial language use and task accuracy. In 32 pairs, task partners interacted in two conditions: for one route description, direction givers (DGs) and direction followers (DFs) sat side-by-side (aligned condition), and for another they sat opposite one another (counter-aligned condition). After each description, DFs drew the route on a map. When pairs were counter-aligned (vs. aligned), DGs increased their use of expressions from a survey perspective, using more frequently terms such as east-west. When counter-aligned, DFs also used more words per conversational turn, which was taken to reflect the increased difficulty of coordinating in that condition. Still, in terms of task performance, the accuracy of DFs’ drawings was unaffected by the partners’ body alignment or spatial language use it was only predicted by the DGs’ spatial ability. We argue that, because direction giving emphasizes accuracy, task partners invest in strategies that contribute to mutual understanding (e.g., recaps of the route by the DF at the end, evidenced by shifts in language use over time). Thus, body alignment in direction giving impacts coordination difficulty and spatial language use, but it does not singularly influence task performance. 57 2 99 121

Previous studies found that an automatic meaning-composition process affects the processing of morphologically complex words, and related this operation to conceptual combination. However, research on embodied cognition demonstrates that concepts are more than just lexical meanings, rather being also grounded in perceptual experience. Therefore, perception-based information should also be involved in mental operations on concepts, such as conceptual combination. Consequently, we should expect to find perceptual effects in the processing of morphologically complex words. In order to investigate this hypothesis, we present the first fully-implemented and data-driven model of perception-based (more specifically, vision-based) conceptual combination, and use the predictions of such a model to investigate processing times for compound words in four large-scale behavioral experiments employing three paradigms (naming, lexical decision, and timed sensibility judgments). We observe facilitatory effects of vision-based compositionality in all three paradigms, over and above a strong language-based (lexical and semantic) baseline, thus demonstrating for the first time perceptually grounded effects at the sub-lexical level. This suggests that perceptually-grounded information is not only utilized according to specific task demands but rather automatically activated when available.

In the present study, we provide a comprehensive analysis and a multi-dimensional dataset of semantic transparency measures for 1,810 German compound words. Compound words are considered semantically transparent when the contribution of the constituents’ meaning to the compound meaning is clear (as in airport), but the degree of semantic transparency varies between compounds (compare strawberry or sandman). Our dataset includes both compositional and relatedness-based semantic transparency measures, also differentiated by constituents. The measures are obtained from a computational and fully implemented semantic model based on distributional semantics. We validate the measures using data from four behavioral experiments: Explicit transparency ratings, two different lexical decision tasks using different nonwords, and an eye-tracking study. We demonstrate that different semantic effects emerge in different behavioral tasks, which can only be capturedusing a multi-dimensional approach to semantic transparency. We further provide the semantic transparency measures derived from the model for a dataset of 40,475 additional German compounds, as well as for 2,061 novel German compounds.

Conversation is generally characterized by smooth transitions between turns, with only very short gaps. This entails that responders often begin planning their response before the ongoing turn is finished. However, controversy exists about whether they start planning as early as they can, to make sure they respond on time, or as late as possible, to minimize the overlap between comprehension and production planning. Two earlier EEG studies have found neural correlates of response planning (positive ERP and alpha decrease) as soon as listeners could start planning their response, already midway through the current turn. However, in these studies, the questions asked were highly controlled with respect to the position where planning could start (e.g., very early) and required short and easy responses. The present study measured participants’ EEG while an experimenter interviewed them in a spontaneous interaction. Coding the questions in the interviews showed that, under these natural circumstances, listeners can, in principle, start planning a response relatively early, on average after only about one third of the question has passed. Furthermore, ERP results showed a large positivity, interpreted before as an early neural signature of response planning, starting about half a second after the start of the word that allowed listeners to start planning a response. A second neural signature of response planning, an alpha decrease, was not replicated as reliably. In conclusion, listeners appear to start planning their response early during the ongoing turn, also under natural circumstances, presumably in order to keep the gap between turns short and respond on time. These results have several important implications for turn-taking theories, which need to explain how interlocutors deal with the overlap between comprehension and production, how they manage to come in on time, and the sources that lead to variability between conversationalists in the start of planning.

Using truth-value judgment tasks, we investigated the on-line processing of counterfactual conditionals such as If kangaroos had no tails, they would topple over. Face-value plausibility of the counterfactual as well as the complexity of the antecedent were manipulated. Results show that readers’ judgments deviate from face-value plausibility more often when the antecedent is complex, and when the counterfactual is plausible rather than implausible. We interpret our results based on the modal horizon assumption of von Fintel (2001) and argue that they are compatible with a variably strict semantics for counterfactuals (Lewis, 1973). We make use of computational modeling techniques to account for reaction times and truth-value judgments simultaneously, showing that implementing detailed process models deepens our understanding of the cognitive mechanisms triggered by linguistic stimuli., Proceedings of Sinn und Bedeutung, Vol 24 No 2 (2020): Proceedings of Sinn und Bedeutung 24

Generic language (e.g., “Birds fly”) conveys generalizations about categories and is essential for learning beyond our direct experience. The meaning of generic language is notoriously hard to specify, however (e.g., penguins don’t fly). Tessler and Goodman (2019) proposed a model for generics that is mathematically equivalent to Bayesian belief-updating based on a single pedagogical example, suggesting a deep connection be- tween learning from experience and learning from language. Relatedly, Csibra and Shamsudheen (2015) argue that generics are inherently pedagogical, understood by infants as referring to a member of a kind. In two experiments with adults, we quantify the exchange-rate between generics and observations by relating their belief-updating capacity, varying both the number of observations and whether they are presented pedagogically or incidentally. We find generics convey stronger generalizations than single pedagogical observations (Expt. 1), even when the property is explicitly demarcated (Expt. 2). We suggest revisions to the vague quantifier model of generics that would allow it to accommodate this intriguing exchange-rate.

Syntactic ambiguities in isolated sentences can lead to increased difficulty in incremental sentence processing, a phenomenon known as a garden-path effect. This difficulty, however, can be alleviated for humans when they are presented with supporting discourse contexts. We tested whether recurrent neural network (RNN) language models (LMs) could learn linguistic representations that are similarly influenced by discourse context. RNN LMs have been claimed to learn a variety of syntactic constructions. However, recent work has suggested that pragmatically conditioned syntactic phenomena are not acquired by RNNs. In comparing model behavior to human behavior, we show that our models can, in fact, learn pragmatic constraints that alleviate garden-path effects given the correct training and testing conditions. This suggests that some aspects of linguistically relevant pragmatic knowledge can be learned from distributional information alone.

A growing body of evidence shows that theta-band (4-7Hz) activity in the auditory cortex phase-locks to the rhythm of overt speech. But does theta activity in the auditory cortex also phase-lock to inner speech? We explored this question with EEG while participants silently read direct (e.g., Mary said: “This dress is lovely!”) and indirect speech quotes (e.g., Mary said that the dress was lovely). Previous research established that direct (vs. indirect) quote reading elicits vivid inner speech which shares temporal properties with overt speech behaviourally (in reading patterns) and activates cortical areas for voice processing. In the present study, we found that direct (vs. indirect) quote reading was associated with increased theta phase-locking at 250-500 ms in EEG post-sentence presentation, with its sources estimated in the speech processing network. A separate eye-tracking experiment of the reading paradigm confirmed that the increased theta phase-locking in direct quote reading could not be attributed to eye movement patterns and may reflect the onset of inner speech. The findings provide novel evidence for a functional role of theta activity in reading-induced inner speech.

Do supraliminal and subliminal priming capture different facets of words’ semantic representations? We used metaphorical priming between space and time as a test bed for this question. While people conceptualize time along the lateral and sagittal axes, only the latter mapping comes up in language (the future is in front of you, not to your right). We assessed facilitation on temporal target words by lateral (left, right) and sagittal (back, front) primes, in masked and overt conditions. Supraliminally, we observe similar sagittal and lateral priming, while the masked effect is stronger on the sagittal axis, and weak to non–existent on the lateral one. These results are observed in an original and a replication studies; and are strongly confirmed by a Bayesian meta–analysis of the two. We conclude that unconscious word processing is limited to linguistically–encoded information, while consciousness may be needed to fully activate semantic representations.

This chapter provides an introduction to the phenomenon of prosodic focus, as well as to the theory of Alternative Semantics. Alternative Semantics provides an insightful account of what prosodic focus means, and gives us a notation that can help with better characterizing focus-related phenomena and the terminology used to describe them. We can also translate theoretical ideas about focus and givenness into this notation to facilitate a comparison between frameworks. The discussion will partly be structured by an evaluation of the theories of Givenness, the theory of Relative Givenness, and Unalternative Semantics, but we will cover a range of other ideas and proposals in the process. The chapter concludes with a discussion of phonological issues, and of association with focus.

Common ground can be mutually established between conversational partners in several ways. We examined whether the modality (visual or linguistic) with which speakers share information with their conversational partners is encoded in memory in a way that affects subsequent references addressed to a particular partner. In 32 triads, directors arranged a set of tangram cards with one matcher and then with another, but in different modalities, sharing some cards only linguistically (by describing cards the matcher couldn't see), some only visually (by silently showing them), some both linguistically and visually, and others not at all. Then directors arranged the cards again in separate rounds with each matcher. The modality with which they previously established common ground about a particular card with a particular matcher (e.g., linguistically with one partner and visually with the other) affected subsequent referring: References to cards previously shared only visually included more idea units, words, and reconceptualizations than those shared only linguistically, which in turn included more idea units, words, and reconceptualizations than those shared both linguistically and visually. Moreover, speakers were able to tailor references to the same card appropriately to the distinct modality shared with each addressee. Such gradient, partner-specific adaptation during re-referring suggests that memory encodes rich-enough representations of multimodal shared experiences to effectively cue relevant constraints about the perceptual conditions under which speakers and addressees establish common ground.

Chapter to be included in the volume Routledge Handbook of Skill and Expertise, Eds. Ellen Fridland and Carlotta Pavese, Springer 2020.

Chapter to be included in the volume Pragmatic Language Disorders: Complex and Underserved Populations, Ed. Louise Cummings, Springer, 2020. It reviews findings that have contributed to our understanding of pragmatic impairment in ALS.

Several evolutionary accounts of human social cognition posit that language has co-evolved with the sophisticated mindreading abilities of modern humans. It has also been argued that these mindreading abilities are the product of cultural, rather than biological, evolution. Taken together, these claims suggest that the evolution of language has played an important role in the cultural evolution of human social cognition. Here we present a new computational model which formalises the assumptions that underlie this hypothesis, in order to explore how language and mindreading interact through cultural evolution. This model treats communicative behaviour as an interplay between the context in which communication occurs, an agent’s individual perspective on the world, and the agent’s lexicon. However, each agent’s perspective and lexicon are private mental representations, not directly observable to other agents. Learners are therefore confronted with the task of jointly inferring the lexicon and perspective of their cultural parent, based on their utterances in context. Simulation results show that given these assumptions, an informative lexicon evolves not just under a pressure to be successful at communicating, but also under a pressure for accurate perspective-inference. When such a lexicon evolves, agents become better at inferring others’ perspectives; not because their innate ability to learn about perspectives changes, but because sharing a language (of the right type) with others helps them to do so.

Sentences like “The athlete jumped for 20 minutes / jogged for 2 years.” engender an iterative meaning that is well understood yet morpho-syntactically unsupported—a case manifesting “meaning underspecification.” We argue that such underspecified meaning results from a costly meaning search in the sentential context by evaluating (a) the conventional event duration denoted by the verb, (b) the interval length denoted by the for-adverbial, and (c) the discourse context if there is one. In a naturalness-rating questionnaire and a self- paced reading experiment, we crossed two verb types (Punctual: jump / Durative: jog) and the interval-lengths denoted by for-adverbials (Short: for 20 minutes / Long: for 2 years). In addition, we investigated the impact of individuals’ cognitive styles indexed by autistic tendency on real-time meaning computation. Results showed that, in the absence of discourse context, (i) the harder a specific meaning could be determined from sentential context, the less natural the sentence was perceived, (ii) all sentences with underspecified meaning engendered significantly longer reading times regardless of verb type. Moreover, (iii) individuals with higher autistic tendency lingered significantly longer at the sentence-final region, likely reflecting a difficulty in information integration or a tendency towards deliberative reasoning. These findings suggest that meaning comprehension is lexically-guided and context-constraining, while the online processing profile vary with individuals’ autistic tendency.

Logic tells us that two negatives make a positive, but in language, things are not so black and white: A person "not unhappy" may not be entirely happy. We hypothesize that innovative uses of double negatives like "not unhappy" stem from listeners entertaining flexible meanings for negation markers like "not" and "un-", which context can then help disambiguate. We formalize this hypothesis in a computational model of language understanding, which predicts that "not unhappy" means something different from "happy" but also makes the additional prediction that single negations ("unhappy" vs. "not happy") are interpreted identically except when a speaker uses both in the same context, which we confirm experimentally. Even double negations that flagrantly use the same negation marker twice (e.g., "not not happy") are interpreted in subtle ways. These findings suggest that even one of the most logical elements of language—negation—can mean many things at once.

This study examined the role of child cognitive abilities for procedural and declarative learning in the earliest stages of second language (L2) exposure. In the context of a computer game, 53 first language Italian monolingual children were aurally trained in a novel miniature language over 3 consecutive days. A mixed effects model analysis of the relationship between cognitive predictors and outcomes in morphosyntax measured via a grammaticality judgment test (GJT) was performed. Relative to adults trained in the same paradigm, children with higher procedural learning ability (measured via an alternate serial reaction time task) showed significantly better learning of word order, although the effect size was small. Modeling accuracy in online sentence comprehension during the game also evidenced that higher procedural learning ability was positively associated with significantly better outcomes as practice progressed. By contrast, a composite measure of verbal and visual declarative learning ability did not predict L2 outcomes in either the GJT or the online measure. © 2021 The Authors. Language Learning published by Wiley Periodicals LLC on behalf of Language Learning Research Club, University of Michigan

Recent research highlights the influence of (e.g., task) context on conceptual retrieval. In order to assess whether conceptual representations are context-dependent rather than static, we investigated the influence of spatial narrative context on accessibility for lexical-semantic information by exploring competition effects. In two visual world experiments, participants listened to narratives describing semantically related (piano-trumpet; Experiment 1) or visually similar (bat-cigarette; Experiment 2) objects in the same or separate narrative locations while viewing arrays displaying these (“target” and “competitor”) objects and other distractors. Upon re-mention of the target, we analysed eye movements to the competitor. In Experiment 1, we observed semantic competition only when targets and competitors were described in the same location; in Experiment 2, we observed visual competition regardless of context. We interpret these results as consistent with context-dependent approaches, such that spatial narrative context dampens accessibility for semantic but not visual information in the visual world.

Each culture has a distinct set of features that contribute to a unique communication style. For example, bilinguals often balance multiple social contexts and may undergo cognitive changes that consequently support different communication styles. The present work examines how individual differences in bilingual experience affect one form of communication style: sarcastic and indirect language. A diverse sample of largely bilingual adults (first language English) rated their likelihood of using sarcastic and indirect language across different daily settings. They also rated their second language experience. There were two key findings: Bilinguals use sarcasm for similar social functions as do monolinguals (general sarcasm, frustration diffusion, and embarrassment diffusion) and greater global second language proficiency linked to greater usage of general sarcasm in daily life. These results suggest that bilinguals may use sarcasm to achieve various communicative goals and bilingual experience may affect general cognitive capacities that support sarcasm use across real-world contexts.

Previous research has shown that early-acquired words are produced faster than late-acquired words (see Juhasz, 2005). Juhasz and colleagues (Juhasz, Lai & Woodcock, 2015; Juhasz, 2018) argue that the Age-of-Acquisition (AoA) loci for complex words, specifically compound words, are found at the lexical/semantic level. In the current study, two experiments were conducted to evaluate this claim and investigate the influence of AoA in reading compound words aloud. In Experiment 1, 48 participants completed a word naming task. Using general linear mixed modelling, we found that the age at which the compound word was learned significantly affected the naming latencies beyond the other psycholinguistic properties measured. The second experiment required 48 participants to name the compound word when the two morphemes were presented with a space in-between (combinatorial naming, e.g. air plane). We found that the age at which the compound word was learned, as well as the AoA of the individual morphemes that formed the compound word, significantly influenced combinatorial naming latency. These findings are discussed in relation to theories of the AoA in language processing.

The way speakers adapt their language during interaction may be influenced by their perception of their partner’s linguistic knowledge. We examined the degree to which speakers adapt to the syntactic structures used by their interlocutor, comparing interaction between native speakers of English and a partner who was a native or a nonnative speaker. In three lab-based experiments, naive participants took turns to describe and match pictures with a confederate who was a native or nonnative speaker of English. We found that participants tended to produce the same syntactic form (prepositional object or double object) that the confederate had just produced on the turn before. We observed different levels of adaptation to native and nonnative confederates, with participants adapting more to nonnative confederates than native confederates, but only when confederates exhibited linguistic inflexibility through a rigid production of double object constructions, some of which were ungrammatical (e.g. “The wizard donates the golfer the cake”). A further two online experiments were conducted to replicate the design of the lab-based experiments. We found differential adaptation to natives and nonnatives again only when confederates’ behaviour exhibited the same linguistic inflexibility; however, the effect of confederate nativeness was the opposite to the lab experiments, with participants adapting more to the inflexible native speaker. Together, our results provide evidence that adaptation is at least partly driven by listener-oriented mechanisms; speakers take into account their partner’s linguistic knowledge and communicative needs when these needs are foregrounded by the communicative context.

We investigated whether the phonological co-activation of alternative names in picture naming (e.g. “fish” for target “shark”) is reduced by contextual constraints which render them inappropriate. In the constraining context, the target naming response was preceded by a naming response to an object from the same category (e.g. an eel) which remained visible during target naming. Therefore, use of the alternative target name “fish” would result (a) in an ambiguous response because of the visual context and (b) in a pragmatically odd response because of the previous naming response. In Experiment 1 the context pictures were named by the participants themselves and in Experiment 2 by a communication partner. In both experiments, interference from distractor words phonologically related (“finger”) versus unrelated (“book”) to the alternative name was observed regardless of context. This finding indicates limited flexibility in lexical activation during speech planning. Our research was supported by the German Research Foundation (DFG JE229/11-1). Andreas Mädebach was supported by a Beatriu de Pinós postdoctoral grant (2017 BP 00180) of the Catalan Government during the writing of this manuscript. We thank Tobias Struck and Ismail Ayoub for assistance in data collection.

Sentences such as “The pop singer began the album” are ambiguous between an agentive reading (The pop singer began recording/listening to/etc. the album) and a constitutive reading (The pop singer’s song was contained in the first track of the album). The ambiguity is rooted in the meaning specification of a semantic class, the “aspectual verb class.” This semantic verb class demands that its complement be construed as a structured individual along a dimension (e.g., spatial, informational, eventive...). In the case of “begin the album,” the complement “the album” can be construed as a set of singing eventualities (a structured individual construed along the eventive dimension) or as a set of songs (a structured individual construed along the informational dimension). Previous work has shown that real-time sentence composition profile of these verbs is consistent with the implementation of two processes (a) exhaustive lexical-function retrieval and (b) construal of multiple dimension-specific structured individuals, leading to agentive and constitutive readings. Which dimension-specific structured individual is chosen ultimately depends on which dimension biases are present in the context. Understanding the real-time behavior of aspectual-verb composition under different contexts (agentive vs. constitutive) provides a glimpse into the time-course of context modulation during real-time sentence comprehension. Results from an eye-tracking study comparing agentive vs. constitutive readings of these sentences show not only the standard effect of the aspectual-verb composition, previously reported for the agentive readings, but also a comparable processing profile for the constitutive readings, a novel finding supporting the unified linguistic analysis and corresponding processing implementation of the two readings. Results also show that regardless of reading, the aspectual-verb composition effect is observable even after the complement has been retrieved, and interpretation has been disambiguated, thus indicating that the above two-step process is a nonnegotiable component of the interpretation of the verb+complement composition and therefore must take place before preceding context can serve as a sentence-external constraining force. Moreover, the sustained nature of the cost suggests that all possible dimension-specific structured individuals are being construed at least immediately after the complement has been retrieved, consistent with comprehenders’ intuitions. This suggests that although not directly relevant for the interpretation at issue those meaning construals are not incompatible with the complement’s conceptual representation, and therefore need not be “pruned.” Overall, the pattern observed suggests a model of comprehension whereby context is allowed to modulate sentence interpretation once the fundamental lexico-semantic compositional processes of the sentence have taken place.

[This article appeared as a chapter in 'Vagueness and Rationality in Language Use and Cognition' published by Springer International Publishing in 2019] One of the traditional pragmatic approaches to vagueness suggests that there needs to be a significant gap between individuals or objects that can be described using a vague adjective like 'tall' and those that cannot. In contrast, intuitively, an explicit comparative like 'taller' does not require fulfillment of the gap requirement. Our starting point for this paper is the consideration that people cannot make precise measures under time pressure and their ability to discriminate approximate heights (or other values) obeys Weber’s law. We formulate and experimentally test three hypotheses relating to the difference between positive and comparative forms of the vague adjectives, gap requirement, and Weber’s law. In two experiments, participants judged appropriateness of usage of positive and comparative forms of vague adjectives in a sentence-picture verification task. Consequently, we review formal analysis of vagueness using weak orders and semi-orders and suggest adjustments based on the experimental results and properties of Weber’s law.

Despite wide variation among natural languages, there are linguistic properties universal to all (or nearly all) languages. An important challenge is to explain why these linguistic universals hold. One explanation employs a learnability argument: semantic universals hold because expressions that satisfy them are easier to learn than those that do not. In an exploratory study we investigate the relation between learnability and complexity and whether the presence of semantic universals for quantifiers can also be explained by differences in complexity. We develop a novel application of (approximate) Kolmogorov complexity to measure fine-grained distinctions in complexity between different quantifiers. Our results indicate that the monotonicity universal can be explained by complexity while the conservativity universal cannot. For quantity we did not find a robust result. We also found that learnability and complexity pattern together in the monotonicity and conservativity cases that we consider, while that pattern is less robust in the quantity cases.

Natural languages exhibit many \emph{semantic universals}: properties of meaning shared across all languages. In this paper, we develop an explanation of one very prominent semantic universal: that all simple determiners denote monotone quantifiers. While existing work has shown that monotone quantifiers are easier to learn, we provide a complete explanation by considering the emergence of quantifiers from the perspective of cultural evolution. In particular, in an iterated learning paradigm, with neural networks as agents, monotone quantifiers regularly evolve.

Generic statements convey generalizations about categories, but how generic predications combine is unclear. “Elephants live in Africa and Asia” does not mean that individual elephants live on both continents. In addition, such conjunctive generics pose interesting questions for theories of incremental processing because the meaning of the sentence can change part-way through: “Elephants live in Africa” would imply most or all do, but “Africa and Asia” implies some live in each. We extend a recently proposed computational model of generic language understanding with an incremental processing mechanism that can begin to interpret an utterance before a speaker has finished their sentence. This model makes novel predictions about partial interpretations of conjunctive generic sentences, which we test in two behavioral experiments. The results support a strong view of incrementality, wherein listeners continuously update their beliefs based on expectations about where a speaker will go next with their utterance.

Generic language (e.g., "Birds have hollow bones") conveys generalizations about categories and is a fundamental and ubiquitous mechanism of learning about the world. How exactly this learning occurs, however, is unclear: Generics exhibit so much flexibility in how they are interpreted that a quantitative theory of how generics update beliefs has not only not been developed or tested, but is often explicitly dismissed. Here, we explore a hypothesis introduced by Tessler and Goodman (2019) that generics update beliefs via an uncertain threshold like a vague quantifier: A category–property generic (Ks F) means that the property F is expected to be relatively widespread in the category K, where what counts as being relatively widespread is a priori uncertain and resolved by consulting one’s prior beliefs about the property. We compare this model to a family of alternatives with the same background knowledge but where generics have a fixed, precise meaning; in addition, we formalize a quantitative model of a strictly conceptual-based approach to generics, where generics are a direct connection to conceptual knowledge. Across three experiments in which we both measure and manipulate prior beliefs, we find the uncertain meaning approach to generics to be the best explanation of participants’ highly heterogeneous interpretations of novel generic statements. This result taken together with the result that the same model of generic meaning can explain the variability in generic endorsements as shown by Tessler & Goodman (2019) suggests that generics are an effective medium for faithful transmission of knowledge between interlocutors. This work adds to the growing enterprise of formal, quantitative studies of human understanding of generic language.

Much has been learned about how individual concepts and semantic dimensions are represented in the human brain using methods from the field of cognitive neuroscience; however, the process of conceptual combination, in which a new concept is created from pre-existing concepts, has received far less attention. We discuss theories and findings from cognitive science and cognitive neuroscience that shed light on the processing stages and neural systems that allow humans to form new conceptual combinations. We review systematic and creative applications of cognitive neuroscience methods, including neuroimaging, neuropsychological patients, neurostimulation and behavioral studies that have yielded fascinating insights into the cognitive nature and neural underpinnings of conceptual combination. Studies have revealed important features of the cognitive processes that are central to successful conceptual combination. Furthermore, we are beginning to understand how regions of the semantic system, such as the anterior temporal lobe and angular gyrus, integrate features and concepts, and evaluate the plausibility of potential resulting combinations, bridging work in linguistics and semantic memory. Despite the relative newness of these questions for cognitive neuroscience, the investigations we review give a very strong foundation for ongoing and future work that seeks to fully understand how the human brain can flexibly integrate existing concepts to form new and never-before experienced combinations at will.

To what extent do people attribute meanings to "nonsense" words? How general is such attribution of meaning? We used a set of words lacking conventional meanings to elicit drawings of made-up creatures. Separate groups of participants rated the nonsense words and the drawings on several semantic dimensions and selected what name best corresponded to each creature. Despite lacking conventional meanings, "nonsense" words elicited a high level of consistency in the produced drawings. Meaning attributions made to nonsense words corresponded with meaning attributions made by separate people to drawings that were inspired by the name. Naive participants were able to recover the name that inspired the drawing with greater-than-chance accuracy. These results suggest that people make liberal and consistent use of non-arbitrary relationships between forms and meanings. OPEN RESEARCH BADGES: This article has been awarded Open Materials and Open Data badges. All materials and data are publicly accessible via the Open Science Framework at https://osf.io/8juyc/. Learn more about the Open Practices badges from the Center for Open Science: https://osf.io/tvyxz/wiki.

This paper presents a framework for how the multifaceted nature of “gender” (human and linguistic) interacts with grammatical operations such as coreference dependency formation. It frames the question through the lens of English, in which it focuses on how personal names and referents who identify as nonbinary can provide insight into the conceptual representations of gender. Additional data from a variety of modern languages supports a model of how gender might be cognitively represented such that the observed linguistic patterns are available. I propose a three-tiered model of gender that unites grammatical, cognitive, social, and biological aspects and describes how implications of this model might be tested in future work.

The paper assesses free verbal reports describing subjective sensations and emotional feelings eliciting by tactile stimuli provided using manual androbotic stimulation. Tickle (knismesis) and pleasure sensations dominated subjective responses, and tickle-related lexical units were significantly more common for stimuli of lower forces. Frequency of pleasure related words did not depend in the force of stimuly. The data prove that, at least for certain perceptual contexts, knismesis may be assosiated with positive rather than with negative emotions.

Research has shown that adults are highly skilled at learning new words and meanings. Here, we examined whether learning new meanings for familiar words affects the processing of their existing meanings. In Experiments 1 and 2, adult participants learnt new, fictitious meanings for previously unambiguous words (e.g., “sip” denoting a small amount of computer data) through four 30-minute training sessions completed over four consecutive days. We tested participants’ comprehension of existing meanings before and after training using a semantic relatedness decision task in which the probe word was related to the existing but not the new meaning of the trained word (e.g., “sip-juice”). Following the training, responses were slower to the trained, but not to the untrained, words, indicating competition between newly-acquired and well-established meanings. Furthermore, consistent with studies of semantic ambiguity, the effect was smaller for meanings that were semantically related to existing meanings than for the unrelated counterparts, demonstrating that meaning relatedness modulates the degree of competition. Overall, the key findings confirm that new word meanings can be integrated into the mental lexicon after just a few days’ exposure, and provide support for current models of ambiguity that predict semantic competition in word comprehension.

Age of acquisition (AoA) is an important psycholinguistic variable that affects the performance of healthy individuals and patients in a large variety of cognitive tasks. For this reason, it becomes more and more compelling to collect new AoA norms for a large set of stimuli in order to allow better control and manipulation of AoA in future research. An important motivation of the present study is to extend previous Italian norms by collecting AoA ratings for a much larger range of Italian words for which concreteness and semantic-affective norms are now available thus ensuring greater coverage of words varying along these dimensions. In the present study, we collected AoA ratings for 1,957 Italian content words (adjectives, nouns and verbs), by asking healthy adult participants to estimate the age at which they thought they had learnt the word in a Web survey procedure. First, we found high split-half correlation within our sample, suggesting strong internal reliability. Second, our data indicate that the ratings collected in this study are as valid and reliable as those collected in previous studies for Italian across different age populations (adult and children) and other languages. Finally, we analyzed the relation between AoA ratings and other lexical-semantic variables (e.g. word frequency, imageability, valence, arousal) and showed that these correlations were generally consistent with the correlations reported in other normative studies for Italian and other languages. Therefore, our new AoA norms are a valuable source of information for future research in the Italian language. The datasets for this study can be found at the Open Science Framework repository (osf.io/3trg2).

We review the range of embodied semantic effects that have been found in visual word recognition paradigms (lexical decision, naming). Many different embodied effects have been elicited by distinct measures of sensorimotor information, and are associated with different theoretical accounts of why semantic content affects how quickly a word can be recognised. We discuss effects due to imageability, body-object interaction, relative embodiment, sensory experience, and modality-specific perceptual strength. Finally, we discuss the impact of embodied semantic effects on current models of visual word recognition.

This chapter reviews recent research on speakers’ referential choices in discourse. It focuses specifically on the choice to mention a certain referent first on the one hand, and on the choice to produce a pronoun or a more elaborate noun phrase on the other. Whereas traditional theories of reference have relied mainly on the influence of the linguistic context to explain these choices, recent psycholinguistic studies have started to investigate the effect of non-linguistic factors as well. Taken together, the results of these studies suggest that the two referential choices (choice of referent and choice of referring expression) are largely driven by different sets of factors, contrary to the claim that they are both related to the accessibility of mental representations in memory. It is concluded that these findings are best explained by assuming that referents can be accessible on different levels: a local and a more global discourse level. Referential choices may be influenced by both levels of accessibility, but with different probabilities for different referential choices.

Until now, there has been little experimental work investigating the processing and formal properties of the singular they suite of pronouns. As scientific and popular attention to singular they increases, it will be critical for research to acknowledge theoretical and ethical issues regarding discussion of this phenomenon. This commentary uses the recent paper by Doherty & Conklin (2017) as a starting point to discuss issues surrounding work on the various forms of singular they. It concludes that there is sufficient theoretical and empirical evidence to claim they has a grammatically singular form (at least in colloquial English). It also recommends care be taken in academic discussions of the grammaticality and acceptability of terms which are associated with marginalised communities.

Neuropsychological studies suggest a distinction between (i) semantic knowledge and (ii) control processes that shape the retrieval of conceptual information to suit the task or context. These aspects of semantic cognition are specifically impaired in patients with semantic dementia and semantic aphasia respectively. However, interactions between the structure of knowledge and control processes that are expected during semantic retrieval have not been fully characterised. In particular, domain-general executive resources may not have equal relevance for the capacity to promote weak yet relevant features (i.e., “controlled retrieval), and to ignore or suppress distracting information (i.e., “selection”). Here, using a feature selection task, we tested the contribution of featural relevance to semantic performance in healthy participants under conditions of divided attention. Healthy participants showed greater dual-task disruption as the relevance value of the distracter feature linearly increased, supporting the emerging view that semantic relevance is one of the organizing principles of the structure of semantic representation. Moreover, target relevance, word frequency and inter-correlational strength affected overall performance, but they did not show an interaction with dual-task conditions. These results suggest that domain-general control processes, disrupted by divided attention, are more relevant to the capacity to efficiently avoid distracting information during semantic decision-making than to the promotion of weak target features. The present study therefore provides novel information about the nature of the interaction between structured conceptual knowledge and control processes that support the retrieval of appropriate information, and relates these results to a new theoretical framework, termed controlled semantic cognition.

One of the strategies that researchers have used to investigate the role of sensorimotor information in lexical-semantic processing is to examine effects of words’ rated body-object interaction (BOI; the ease with which the human body can interact with a word’s referent). Processing tends to be facilitated for words with high BOI compared to words with low BOI, across a wide variety of tasks. Such effects have been referenced in debates over the nature of semantic representations, but their theoretical import has been limited by the fact that BOI is a fairly coarse measure of sensorimotor experience with words’ referents. In the present study we collected ratings for 621 words on seven semantic dimensions (graspability, ease of pantomime, number of actions, animacy, size, danger, and usefulness) in order to investigate which attributes are most strongly related to BOI ratings, and to lexical-semantic processing. BOI ratings were obtained from previous norming studies (Bennett, Burnett, Siakaluk, & Pexman, 2011; Tillotson, Siakaluk, & Pexman, 2008) and measures of lexical-semantic processing were obtained from previous behavioural megastudies involving the semantic categorization task (concrete/abstract decision; Pexman, Heard, Lloyd, & Yap, 2017) and the lexical decision task (Balota et al., 2007). Results showed that the motor dimension of graspability, ease of pantomime, and number of actions were all related to BOI and that these dimensions together explained more variance in semantic processing than did BOI ratings alone. These ratings will be useful for researchers who wish to study how different kinds of bodily interactions influence lexical-semantic processing and cognition.

How does listeners’ perceptual bias influence their interpretation of an ambiguous multiword utterance? We address this question by investigating the relationship between word order in a binomial (an expression of type "A and B") and visual properties of image pairs serving as its potential referents. We found that listeners’ choices were strongly influenced by iconicity and relative salience of images within the pair: participants preferred referents where the first mentioned image was located on the left, as well as pairs where the first image was larger than the second image. The effect of image order tended to be stronger than the influence of image size, and both were modulated by participants’ general visual field preferences (determined in a separate experimental condition). We further show that binomial phrase interpretation can be simulated by a Rational Speech Act model that includes both word order effects and utterance-independent preferences of the participants.

we test the hypothesis that naming an object depicted in a picture, and reading aloud an object’s name, are affected by the object’s speed. We contend that the mental representations of everyday objects and situations include their speed, and that the latter influences behavior in instantaneous and systematic ways. An important corollary is that high-speed objects are named faster than low-speed objects despite the fact that object speed is irrelevant to the naming task at hand. The results of a series of 7 studies with pictures and words support these predictions.

Perspective taking-the ability to see things from someone else's point of view-can boost success in communication. A signaler might take perspective when designing an utterance that is informative from the receiver's point of view, or the receiver might take perspective when inferring the signaler's communicative intentions. Perspective taking is supposed to play a particularly vital role when people try to communicate in the absence of a conventional signaling system. However, the task demands in such cases are extremely different from those in typical experimental approaches to perspective taking. Thus, current evidence for perspective taking does not establish whether humans can take perspective in those cases where perspective taking is arguably most helpful. We describe experimental tests of perspective taking that are suitable for settling the matter. Our task focuses on the use of shared world knowledge rather than shared visual scenes, and it is suitable for both open-ended and contextually constrained responses. We show that people generally fail at perspective taking in a novel signaling task, but that perspective taking can be boosted by contextual constraint. In that case, however, it is context, rather than perspective taking or shared world knowledge, that explains communicative success. (PsycINFO Database Record (c) 2018 APA, all rights reserved).

Many researchers have tried to predict semantic priming effects using a myriad of variables (e.g., prime–target associative strength or co-occurrence frequency). The idea is that relatedness varies across prime–target pairs, which should be reflected in the size of the priming effect (e.g., cat should prime dog more than animal does). However, it is only insightful to predict item-level priming effects if they can be measured reliably. Thus, in the present study we examined the split-half and test–retest reliabilities of item-level priming effects under conditions that should discourage the use of strategies. The resulting priming effects proved extremely unreliable, and reanalyses of three published priming datasets revealed similar cases of low reliability. These results imply that previous attempts to predict semantic priming were unlikely to be successful. However, one study with an unusually large sample size yielded more favorable reliability estimates, suggesting that big data, in terms of items and participants, should be the future for semantic priming research. ispartof: Behavior Research Methods vol:50 issue:6 pages:2173-2183 ispartof: location:United States status: published

There is a lively philosophical debate concerning which epistemic norm regulates assertion. Some scholars claim that such norm is factive (and only true propositions are assertable), while others deny this thesis (they contend that assertability does not entail truth). There is widespread agreement in the literature that these two opposite hypotheses are empirically testable, and John Turri (2013a, 2013b, 2016a, 2016b) has collected compelling evidence in favour of a factive norm of assertion.This paper makes a negative and positive claim. The negative claim concerns previous evidence found in support of factive norms. We argue that the empirical data collected by Turri does not support his conclusions. Turri drew conclusions about assertability from judgments about whether a given agent 'should' assert something. We argue that there are two possible interpretation of ‘should’, a teleological and a normative one, and that only a normative interpretation of the ‘should’-judgments would support Turri’s conclusion. In experiment 1, we show that the majority of participants in a follow-up study reported that they interpret ‘should’ in a teleological, non-normative way, showing that Turri’s results do not seem to support his conclusions.The positive claim of the paper is that assertion is not regulated by a factive norm. We conducted three test to support this claim. In experiment 2, we modified the temporal structure of Turri’s original vignettes so as to avoid a teleological interpretation of the ‘should’ question. In experiment 3, we introduced three alternative probes, and asked participants whether the speaker ‘ought’ to assert p, and whether it was ‘permissible’ or ‘appropriate’ to do it. The result of both experiments strongly support the non-factive hypothesis against the factive one. However, we also identified some reasons to suspect that our alternative vignettes could also have been interpreted in a way that does not track judgments of assertability. To collect judgments that are unmistakably about assertability, in experiment 4 we explicitly instructed participants that the study was about an epistemic norm regulating assertion, and asked them to judge whether given assertions violated the norm. Consistently with the previous experiments, the results show that participants do not take assertability to require truth.

A key component of most models of pragmatics is that speakers consider more than one way of conveying a message, and how informative each version is in context. Theories of pragmatics, and particularly pragmatic development, are hampered by the fact that while we can often observe what a participant does (either as a speaker or a listener), we can rarely observe the choices they consider. Here, we focus on the problem that remains once two alternative utterances are available: selecting the most informative one. By providing 3-6 year old children with a choice between transitive sentences that drop either an uninformative argument (i.e. one that can be directly recovered from the visual context) or an informative one, we can directly test their ability to select between alternate ways of conveying the same message, abstracting away from the many other processes required to produce a multi-word expression. While 3 year olds select between these alternatives at chance (a pattern which, based on control version, may be partially due to task demands), we find clear evidence that children aged 4 and older preferentially select the more informative option. This result shows that (1) children's awareness of informativity contrasts extends to sentence-level messages, not just object referent selection and (2) directly testing the development of specific cognitive abilities theorized to be involved in pragmatic communication can both clarify and constrain our theories of language development.

How abstract is language? We show that abstractness pervades every corner of language, going far beyond the usual examples of freedom and justice. In light of the ubiquity of abstract words, the need to understand where abstract meanings come from becomes ever more acute. We argue that the best source of knowledge about abstract meanings may be language itself. We then consider a seemingly unrelated question: Why isn’t language more iconic? Iconicity—a resemblance between the form of words and their meanings—can be immensely useful in language learning and communication. Languages could be much more iconic than they currently are. So why aren’t they? We suggest that one reason is that iconicity is inimical to abstraction become iconic forms are too connected to specific contexts and sensory depictions. Form-meaning arbitrariness may allow language to better convey abstract meanings.

Asking questions is one of our most efficient and reliable means of learning about the world. Yet we do not often pose these questions to an impartial oracle; we ask cooperative social partners, in dialogue. In this paper, we aim to reconcile formal models of optimal question asking and answering with classic effects of social context. We begin from the observation that question-answer dialogue is motivated by a two-sided asymmetry in beliefs: questioners have a private goal but lack goal-relevant information about the world, and answerers have private information but lack knowledge about the questioner's goal. We formalize this problem in a computational framework and derive pragmatic questioner and answerer behavior from recursive social reasoning. Critically, we predict that pragmatic answerers go beyond the literal meaning of the question to be informative with respect to inferred goals, and that pragmatic questioners may therefore select questions to more unambiguously signal their goals. We evaluate our pragmatic models against asocial models in two ways. First, we present computational simulations accounting for three classic answerer effects in psycholinguistics. We then introduce the Hidden Goal paradigm for experimentally eliciting questioner and answerer behavior in scenarios where there is uncertainty about the questioner's goal. We report data from three experiments in this paradigm and show how our core computational framework can be composed with more sophisticated question semantics, hierarchical goal spaces, and a persistent state over which extended dialogue can unfold. We find that social inference is needed to account for critical aspects of the data.

Current models of word-meaning access typically assume that lexical-semantic representations of ambiguous words (e.g., 'bark of the dog/tree') reach a relatively stable state in adulthood, with only the relative frequencies of meanings and immediate sentence context determining meaning preference. However, recent experience also affects interpretation: recently encountered word-meanings become more readily available (Rodd et al., 2016, 2013). Here, 3 experiments investigated how multiple encounters with word-meanings influence the subsequent interpretation of these ambiguous words. Participants heard ambiguous words contextually-disambiguated towards a particular meaning and, after a 20- to 30-min delay, interpretations of the words were tested in isolation. We replicate the finding that 1 encounter with an ambiguous word biased the later interpretation of this word towards the primed meaning for both subordinate (Experiments 1, 2, 3) and dominant meanings (Experiment 1). In addition, for the first time, we show cumulative effects of multiple repetitions of both the same and different meanings. The effect of a single subordinate exposure persisted after a subsequent encounter with the dominant meaning, compared to a dominant exposure alone (Experiment 1). Furthermore, 3 subordinate word-meaning repetitions provided an additional boost to priming compared to 1, although only when their presentation was spaced (Experiments 2, 3); massed repetitions provided no such boost (Experiments 1, 3). These findings indicate that comprehension is guided by the collective effect of multiple recently activated meanings and that the spacing of these activations is key to producing lasting updates to the lexical-semantic network. (PsycINFO Database Record