This paper develops a novel psycholinguistic parser and tests it against experimental and corpus reading data. The parser builds on the recent research into memory structures, which argues that memory retrieval is content‐addressable and cue‐based. It is shown that the theory of cue‐based memory systems can be combined with transition‐based parsing to produce a parser that, when combined with the cognitive architecture ACT‐R, can model reading and predict online behavioral measures (reading times and regressions). The parser's modeling capacities are tested (...) against self‐paced reading experimental data (Grodner & Gibson, 2005), eye‐tracking experimental data (Staub, 2011), and a self‐paced reading corpus (Futrell et al., 2018). (shrink)

Source code is a form of human communication, albeit one where the information shared between the programmers reading and writing the code is constrained by the requirement that the code executes correctly. Programming languages are more syntactically constrained than natural languages, but they are also very expressive, allowing a great many different ways to express even very simple computations. Still, code written by developers is highly predictable, and many programming tools have taken advantage of this phenomenon, relying on language model (...) surprisal as a guiding mechanism. While surprisal has been validated as a measure of cognitive load in natural language, its relation to human cognitive processes in code is still poorly understood. In this paper, we explore the relationship between surprisal and programmer preference at a small granularity—do programmers prefer more predictable expressions in code? Using meaning‐preserving transformations, we produce equivalent alternatives to developer‐written code expressions and run a corpus study on Java and Python projects. In general, language models rate the code expressions developers choose to write as more predictable than these transformed alternatives. Then, we perform two human subject studies asking participants to choose between two equivalent snippets of Java code with different surprisal scores (one original and transformed). We find that programmers do prefer more predictable variants, and that stronger language models like the transformer align more often and more consistently with these preferences. (shrink)

A key component of research on human sentence processing is to characterize the processing difficulty associated with the comprehension of words in context. Models that explain and predict this difficulty can be broadly divided into two kinds, expectation‐based and memory‐based. In this work, we present a new model of incremental sentence processing difficulty that unifies and extends key features of both kinds of models. Our model, lossy‐context surprisal, holds that the processing difficulty at a word in context is proportional to (...) the surprisal of the word given a lossy memory representation of the context—that is, a memory representation that does not contain complete information about previous words. We show that this model provides an intuitive explanation for an outstanding puzzle involving interactions of memory and expectations: language‐dependent structural forgetting, where the effects of memory on sentence processing appear to be moderated by language statistics. Furthermore, we demonstrate that dependency locality effects, a signature prediction of memory‐based theories, can be derived from lossy‐context surprisal as a special case of a novel, more general principle called information locality. (shrink)

We present a self-organizing approach to sentence processing that sheds new light on notional plurality effects in agreement attraction, using pseudopartitive subject noun phrases. We first show that notional plurality ratings predict verb agreement choices in pseudopartitives, in line with the “Marking” component of the Marking and Morphing theory of agreement processing. However, no account to date has derived notional plurality values from independently needed principles of language processing. We argue on the basis of new experimental evidence and a dynamical (...) systems model that the theoretical black box of notional plurality can be unpacked into objectively measurable semantic features. With these semantic features driving structure formation, our model reproduces the human verb production patterns as a byproduct of normal processing. Finally, we discuss how the self-organizing approach might be extended to other agreement attraction phenomena. (shrink)

How do cognitive pressures shape the lexicons of natural languages? Here, we reframe George Kingsley Zipf's proposed “law of abbreviation” within a more general framework that relates it to cognitive pressures that affect speakers and listeners. In this new framework, speakers' drive to reduce effort (Zipf's proposal) is counteracted by the need for low‐frequency words to have word forms that are sufficiently distinctive to allow for accurate recognition by listeners. To support this framework, we replicate and extend recent work using (...) the prevalence of subword phonemic sequences (phonotactic probability) to measure speakers' production effort in place of Zipf's measure of length. Across languages and corpora, phonotactic probability is more strongly correlated with word frequency than word length. We also show this measure of ease of speech production (phonotactic probability) is strongly correlated with a measure of perceptual difficulty that indexes the degree of competition from alternative interpretations in word recognition. This is consistent with the claim that there must be trade‐offs between these two factors, and is inconsistent with a recent proposal that phonotactic probability facilitates both perception and production. To our knowledge, this is the first work to offer an explanation why long, phonotactically improbable word forms remain in the lexicons of natural languages. (shrink)

Cognitive Science, Volume 46, Issue 10, October 2022.

In this study, we investigate the processing of object-denoting emoji in sentences using eye tracking. We hypothesize that (a) such emoji are more difficult to process when used as word replacement; and (b) their processing is subject to ambiguity constraints similarly to what happens with words. We conduct two experiments in which participants have to read sentences in which an emoji either follows or replaces a word. Control stimuli not containing emoji are also tested. In the second experiment, the emoji (...) are presented in two different disambiguating contexts. First fixation duration, total visit duration, and total revisit duration are modeled in the various conditions using linear mixed models. Both our hypotheses are supported. We observe longer total visit time for non-redundant emoji, and higher values for all three measures for ambiguous emoji. We conclude that lexical access may be more difficult for emoji especially when they are used as word replacements and their meaning is not immediately clear. Furthermore, we also conclude that non-redundant emoji are more difficult to integrate in the processing of the sentence than the equivalent words, or emoji used in a redundant way. In turn, our results indicate that emoji may not always be as immediate and easy to process for readers in spite of their popularity. (shrink)

This paper investigates the effects of 10 factors on the choice between alternative ba sentences and SVO sentences in Mandarin Chinese. These factors are givenness, definiteness, animacy and pronominality of NP2s, NP2 length, VP length, verb sense, syntactic parallelism, dependency distance, and surprisal. Using corpus data and mixed-effects logistic regression modeling, we find that on the one hand, givenness, syntactic parallelism, and the log-transformed ratio of NP2 length and VP length are significant predictors of the choice between ba sentences and (...) SVO sentences. A new NP2, a large length ratio and a parallel construction predict an SVO sentence rather than a ba sentence. On the other hand, dependency distance and surprisal estimated by the trigram model are effective in predicting the choice between naturally occurring ba/svo sentences and their alternatives. Naturally occurring sentences are more likely to have shorter dependency distances and smaller surprisal values than the converted sentences. The effects of these five factors on syntactic choice are congruent with results of previous studies, which suggests that some determinants of syntactic choice are shared among languages. (shrink)

The information theoretic principle of rational adaptation predicts that individuals with aphasia adapt to their language impairments by relying more heavily on comparatively unimpaired non-linguistic knowledge to communicate. This prediction was examined by assessing the extent to which adults with chronic aphasia due to left-hemisphere stroke rely more on conceptual rather than lexical information during verb retrieval, as compared to age-matched neurotypical controls. A primed verb naming task examined the degree of facilitation each participant group received from either conceptual event-related (...) or lexical collocate cues, compared to unrelated baseline cues. The results provide evidence that adults with aphasia received amplified facilitation from conceptual cues compared to controls, whereas healthy controls received greater facilitation from lexical cues. This indicates that adaptation to alternative and relatively unimpaired information may facilitate successful word retrieval in aphasia. Implications for models of rational adaptation and clinical neurorehabilitation are discussed. (shrink)

Spoken communication occurs in a “noisy channel” characterized by high levels of environmental noise, variability within and between speakers, and lexical and syntactic ambiguity. Given these properties of the received linguistic input, robust spoken word recognition—and language processing more generally—relies heavily on listeners' prior knowledge to evaluate whether candidate interpretations of that input are more or less likely. Here we compare several broad-coverage probabilistic generative language models in their ability to capture human linguistic expectations. Serial reproduction, an experimental paradigm where (...) spoken utterances are reproduced by successive participants similar to the children's game of “Telephone,” is used to elicit a sample that reflects the linguistic expectations of English-speaking adults. When we evaluate a suite of probabilistic generative language models against the yielded chains of utterances, we find that those models that make use of abstract representations of preceding linguistic context (i.e., phrase structure) best predict the changes made by people in the course of serial reproduction. A logistic regression model predicting which words in an utterance are most likely to be lost or changed in the course of spoken transmission corroborates this result. We interpret these findings in light of research highlighting the interaction of memory-based constraints and representations in language processing. (shrink)

English speakers use vertical language to talk about power, such as when speaking of people being “at the bottom of the social hierarchy” or “rising to the top.” Experimental research has shown tha...

There is substantial support for the general idea that a formalization of comprehenders' expectations about the likely next word in a sentence helps explaining data related to online sentence processing. While much research has focused on syntactic, semantic, and discourse expectations, the present event‐related potentials (ERPs) study investigates neurolinguistic correlates of pragmatic expectations, which arise when comprehenders expect a sentence to conform to Gricean Maxims of Conversation. For predicting brain responses associated with pragmatic processing, we introduce a formal model of (...) such Gricean pragmatic expectations, using an idealized incremental interpreter. We examine whether pragmatic expectancies derived from this model modulate the amplitude of the N400, a component that has been associated with predictive processing. As part of its parameterization, the model distinguishes genuine pragmatic interpreters, who expect maximally informative true utterances, from literal interpreters, who only expect truthfulness. We explore the model's non‐trivial predictions for an experimental setup which uses picture‐sentence verification with ERPs recorded at several critical positions in sentences containing the scalar implicature trigger some. We find that Gricean expectations indeed affect the N400, largely in line with the predictions of our model, but also discuss discrepancies between model predictions and observations critically. (shrink)

Determining optimal units of representing morphologically complex words in the mental lexicon is a central question in psycholinguistics. Here, we utilize advances in computational sciences to study human morphological processing using statistical models of morphology, particularly the unsupervised Morfessor model that works on the principle of optimization. The aim was to see what kind of model structure corresponds best to human word recognition costs for multimorphemic Finnish nouns: a model incorporating units resembling linguistically defined morphemes, a whole-word model, or a (...) model that seeks for an optimal balance between these two extremes. Our results showed that human word recognition was predicted best by a combination of two models: a model that decomposes words at some morpheme boundaries while keeping others unsegmented and a whole-word model. The results support dual-route models that assume that both decomposed and full-form representations are utilized to optimally process complex words within the mental lexicon. (shrink)

There is now considerable evidence that human sentence processing is expectation based: As people read a sentence, they use their statistical experience with their language to generate predictions about upcoming syntactic structure. This study examines how sentence processing is affected by readers' uncertainty about those expectations. In a self-paced reading study, we use lexical subcategorization distributions to factorially manipulate both the strength of expectations and the uncertainty about them. We compare two types of uncertainty: uncertainty about the verb's complement, reflecting (...) the next prediction step; and uncertainty about the full sentence, reflecting an unbounded number of prediction steps. We find that uncertainty about the full structure, but not about the next step, was a significant predictor of processing difficulty: Greater reduction in uncertainty was correlated with increased reading times. We additionally replicated previously observed effects of expectation violation, orthogonal to the effect of uncertainty. This suggests that both surprisal and uncertainty affect human RTs. We discuss the consequences for theories of sentence comprehension. (shrink)

Word co‐occurrence patterns in language corpora contain a surprising amount of conceptual knowledge. Large language models (LLMs), trained to predict words in context, leverage these patterns to achieve impressive performance on diverse semantic tasks requiring world knowledge. An important but understudied question about LLMs’ semantic abilities is whether they acquire generalized knowledge of common events. Here, we test whether five pretrained LLMs (from 2018's BERT to 2023's MPT) assign a higher likelihood to plausible descriptions of agent−patient interactions than to minimally (...) different implausible versions of the same event. Using three curated sets of minimal sentence pairs (total n = 1215), we found that pretrained LLMs possess substantial event knowledge, outperforming other distributional language models. In particular, they almost always assign a higher likelihood to possible versus impossible events (The teacher bought the laptop vs. The laptop bought the teacher). However, LLMs show less consistent preferences for likely versus unlikely events (The nanny tutored the boy vs. The boy tutored the nanny). In follow‐up analyses, we show that (i) LLM scores are driven by both plausibility and surface‐level sentence features, (ii) LLM scores generalize well across syntactic variants (active vs. passive constructions) but less well across semantic variants (synonymous sentences), (iii) some LLM errors mirror human judgment ambiguity, and (iv) sentence plausibility serves as an organizing dimension in internal LLM representations. Overall, our results show that important aspects of event knowledge naturally emerge from distributional linguistic patterns, but also highlight a gap between representations of possible/impossible and likely/unlikely events. (shrink)

The present “visual world” eye-tracking study examined the time-course of how native and non-native speakers keep track of implied object-state representations during real-time language processing. Fifty-two native speakers of English and 46 non-native speakers with advanced English proficiency joined this study. They heard short stories describing a target object either having undergone a substantial change-of-state or a minimal change-of-state while their eye movements toward competing object-states and two unrelated distractors were tracked. We found that both groups successfully directed their visual (...) attention toward the end-state of the target object that was implied in the linguistic context. However, neither group showed anticipatory eye movements toward the implied object-state when hearing the critical verb. Only native English speakers but not non-native speakers showed a bias in visual attention during the determiner before the noun. Our results suggested that although native and non-native speakers of English largely overlapped in their time-courses of keeping track of object-state representations during real-time language comprehension, non-native speakers showed a short delay in updating the implied object-state representations. (shrink)

This paper synthesizes a wide range of literature from sociolinguistics and cognitive psychology, to argue for a central role for the “word” as a vehicle of language variation and change. Three crucially interlinked strands of research are reviewed—the role of context in associative learning, the word-level storage of phonetic and contextual detail, and the phonetic consequences of skewed distributions of words across different contexts. I argue that the human capacity for associative learning, combined with attention to fine-phonetic detail at the (...) level of the word, plays a significant role in predicting a range of subtle but systematically robust observed socioindexical patterns in speech production and perception. (shrink)

The disambiguation of a syntactically ambiguous sentence in favor of a less preferred parse can lead to slower reading at the disambiguation point. This phenomenon, referred to as a garden‐path effect, has motivated models in which readers initially maintain only a subset of the possible parses of the sentence, and subsequently require time‐consuming reanalysis to reconstruct a discarded parse. A more recent proposal argues that the garden‐path effect can be reduced to surprisal arising in a fully parallel parser: words consistent (...) with the initially dispreferred but ultimately correct parse are simply less predictable than those consistent with the incorrect parse. Since predictability has pervasive effects in reading far beyond garden‐path sentences, this account, which dispenses with reanalysis mechanisms, is more parsimonious. Crucially, it predicts a linear effect of surprisal: the garden‐path effect is expected to be proportional to the difference in word surprisal between the ultimately correct and ultimately incorrect interpretations. To test this prediction, we used recurrent neural network language models to estimate word‐by‐word surprisal for three temporarily ambiguous constructions. We then estimated the slowdown attributed to each bit of surprisal from human self‐paced reading times, and used that quantity to predict syntactic disambiguation difficulty. Surprisal successfully predicted the existence of garden‐path effects, but drastically underpredicted their magnitude, and failed to predict their relative severity across constructions. We conclude that a full explanation of syntactic disambiguation difficulty may require recovery mechanisms beyond predictability. (shrink)

In referential communication, Grice's Maxim of Quantity is thought to imply that utterances conveying unnecessary information should incur comprehension difficulties. There is, however, considerable evidence that speakers frequently encode redundant information in their referring expressions, raising the question as to whether such overspecifications hinder listeners’ processing. Evidence from previous work is inconclusive, and mostly comes from offline studies. In this article, we present two event‐related potential (ERP) experiments, investigating the real‐time comprehension of referring expressions that contain redundant adjectives in complex (...) visual contexts. Our findings provide support for both Gricean and bounded‐rational accounts. We argue that these seemingly incompatible results can be reconciled if common ground is taken into account. We propose a bounded‐rational account of overspecification, according to which even redundant words can be beneficial to comprehension to the extent that they facilitate the reduction of listeners’ uncertainty regarding the target referent. (shrink)

Recent work has shown that linguistic and visual contexts jointly modulate linguistic expectancy and, thus, the processing effort for a expected critical word. According to these findings, uncertainty about the upcoming referent in a visually-situated sentence can be reduced by exploiting the selectional restrictions of a preceding word, which then reduces processing effort on the critical word. Interestingly, however, no such modulation was observed in these studies on the expectation-generating word itself. The goal of the current study is to investigate (...) whether the reduction of uncertainty simply does not modulate processing effort-or whether the particular subject-verb-object sentence structure used in these studies accounts for the observed pattern. To test these questions, the current design reverses the functional roles of nouns and verbs by using sentence constructions in which the noun reduces uncertainty about upcoming verbs, and the verb provides the disambiguating and reference-resolving piece of information. Experiment 1 and Experiment 2 both replicate the effect found in previous work, albeit on the verb in the current study. Results on the noun, where uncertainty is reduced and expectations are generated in the current design, were mixed and were most likely influenced by design decisions specific to each experiment. These results show that processing of the reference-resolving word—whether it be a noun or a verb—reliably benefits from the prior linguistic and visual information that lead to the generation of concrete expectations. (shrink)

What are the effects of word-by-word predictability on sentence processing times during the natural reading of a text? Although information complexity metrics such as surprisal and entropy reduction have been useful in addressing this question, these metrics tend to be estimated using computational language models, which require some degree of commitment to a particular theory of language processing. Taking a different approach, this study implemented a large-scale cumulative cloze task to collect word-by-word predictability data for 40 passages and compute surprisal (...) and entropy reduction values in a theory-neutral manner. A separate group of participants read the same texts while their eye movements were recorded. Results showed that increases in surprisal and entropy reduction were both associated with increases in reading times. Furthermore, these effects did not depend on the global difficulty of the text. The findings suggest that surprisal and entropy reduction independently contribute to variation in reading times, as these metrics seem to capture different aspects of lexical predictability. (shrink)

Prior research has shown that various types of conventional multiword chunks are processed faster than matched novel strings, but it is unclear whether this processing advantage extends to variant multiword chunks that are less formulaic. To determine whether the processing advantage of multiword chunks accommodates variations in the canonical phrasal template, we examined the robustness of the processing advantage (i.e., predictability) of binomial phrases with non‐canonical conjunctions (e.g.,salt and also pepper; salt as well as pepper). Results from the cloze study (...) (Experiment 1) showed that there was a high tendency of producing the canonical conjunct (pepper), even in the binomials that contained non‐formulaic conjunctions. Consistent with these findings, results from two eye tracking studies (Experiments 2a and 2b) showed that canonical conjuncts were read faster than novel conjuncts that were matched on word length (e.g.,paprika), even in the binomials with variant conjunctions. This robust online processing advantage was replicated in a self‐paced reading study that compared all three conjunction types (Experiment 3). Taken together, these findings show that binomials with variant function words also receive facilitated processing relative to matched novel strings, even though both types of strings are neither conventional nor relatively frequent. Exploratory analyses revealed that this processing speed advantage was driven by the lexical–semantic association between the canonical conjuncts (salt–pepper), rather than lexical and phrasal frequency. Overall, these results highlight flexibility in the processing of multiword chunks that current models of multiword storage and processing must take into account. (shrink)

Expectation-based theories of language comprehension, in particular Surprisal Theory, go a long way in accounting for the behavioral correlates of word-by-word processing difficulty, such as reading times. An open question, however, is in which component of the Event-Related brain Potential signal Surprisal is reflected, and how these electrophysiological correlates relate to behavioral processing indices. Here, we address this question by instantiating an explicit neurocomputational model of incremental, word-by-word language comprehension that produces estimates of the N400 and the P600—the two most (...) salient ERP components for language processing—as well as estimates of “comprehension-centric” Surprisal for each word in a sentence. We derive model predictions for a recent experimental design that directly investigates “world-knowledge”-induced Surprisal. By relating these predictions to both empirical electrophysiological and behavioral results, we establish a close link between Surprisal, as indexed by reading times, and the P600 component of the ERP signal. The resultant model thus offers an integrated neurobehavioral account of processing difficulty in language comprehension. (shrink)

This paper argues that neurolinguistics has the potential to yield insights that can feed back into corpus-based Cognitive Linguistics. It starts by discussing how far the cognitive realism of probabilistic statements derived from corpus data currently goes. Against this background, it argues that the cognitive realism of usage-based models could be further enhanced through deeper engagement with neurolinguistics, but also highlights a number of common misconceptions about what neurolinguistics can and cannot do for linguistic theorizing.