Offline dominance and zeugmatic similarity normings of variably ambiguous words assessed against a neural language model (BERT).

For any research program examining how ambiguous words are processed in broader linguistic contexts, a first step is to establish factors relating to the frequency balance or dominance of those words' multiple meanings, as well as the similarity of those meanings to one other. Homonyms-words with divergent meanings-are one ambiguous word type commonly utilized in psycholinguistic research. In contrast, although polysemes-words with multiple related senses-are far more common in English, they have been less frequently used as tools for understanding one-to-many word-to-meaning mappings. The current paper details two norming studies of a relatively large number of ambiguous English words. In the first, offline dominance norming is detailed for 547 homonyms and polysemes via a free association task suitable for words across the ambiguity continuum, with a goal of identifying words with more equibiased meanings. The second norming assesses offline meaning similarity for a partial subset of 318 ambiguous words (including homonyms, unambiguous words, and polysemes divided into regular and irregular types) using a novel, continuous rating method reliant on the linguistic phenomenon of zeugma. In addition, we conduct computational analyses on the human similarity norming data using the BERT pretrained neural language model (Devlin et al., 2018, BERT: Pre-training of deep bidirectional transformers for language understanding. ArXiv Preprint. arXiv:1810.04805) to evaluate factors that may explain variance beyond that accounted for by dictionary-criteria ambiguity categories. Finally, we make available the summarized item dominance values and similarity ratings in resultant appendices (see supplementary material), as well as individual item and participant norming data, which can be accessed online ( https://osf.io/g7fmv/ ).

Dpr10 and Nocte are required for Drosophila motor axon pathfinding.

The paths axons travel to reach their targets and the subsequent synaptic connections they form are highly stereotyped. How cell surface proteins (CSPs) mediate these processes is not completely understood. The Drosophila neuromuscular junction (NMJ) is an ideal system to study how pathfinding and target specificity are accomplished, as the axon trajectories and innervation patterns are known and easily visualized. Dpr10 is a CSP required for synaptic partner choice in the neuromuscular and visual circuits and for axon pathfinding in olfactory neuron organization. In this study, we show that Dpr10 is also required for motor axon pathfinding. To uncover how Dpr10 mediates this process, we used immunoprecipitation followed by mass spectrometry to identify Dpr10 associated proteins. One of these, Nocte, is an unstructured, intracellular protein implicated in circadian rhythm entrainment. We mapped nocte expression in larvae and found it widely expressed in neurons, muscles, and glia. Cell-specific knockdown suggests nocte is required presynaptically to mediate motor axon pathfinding. Additionally, we found that nocte and dpr10 genetically interact to control NMJ assembly, suggesting that they function in the same molecular pathway. Overall, these data reveal novel roles for Dpr10 and its newly identified interactor, Nocte, in motor axon pathfinding and provide insight into how CSPs regulate circuit assembly.

Testing limits: ERP evidence for word form preactivation during speeded sentence reading.

While it is commonly agreed upon that language comprehenders preactivate information at multiple levels, there is less consensus regarding what and when information is predicted, under which circumstances, and via which mechanism(s). Regarding when, Ito, Corley, Pickering, Martin, & Nieuwland (2016) concluded that during sentence processing, word form-unlike semantic-preactivation crucially relies on the time available to generate late-stage predictions via language production mechanisms, setting this limit between 500 and 700 ms/word. The current event-related brain potential (ERP) study tests this proposal at a substantially faster serial visual presentation (SVP) rate of four words/s, on par with normal reading. We utilize the experimental design and replicate the general ERP findings of the two words/s SVP study of DeLong, Chan, & Kutas (2019), with results showing similar N400 reductions to unpredictable sentence continuations whether semantically or orthographically related to contextually predictable words, as well as an increased late posterior positivity to orthographic neighbors. These findings indicate that processing of written word information can be rapidly facilitated through context-based expectancies, establishing that if there is a time constraint for word form preactivation, it must be far less than limits specified by the prediction-by-production comprehension model championed by Ito et al.

An exploratory data analysis of word form prediction during word-by-word reading.

In 2005, we reported evidence indicating that upcoming phonological word forms-e.g., kite vs. airplane-were predicted during reading. We recorded brainwaves (electroencephalograms [EEGs]) as people read word-by-word and then correlated the predictability in context of indefinite articles that preceded nouns ( a kite vs. an airplane) with the average event-related brain potentials (ERPs) they elicited [K. A. DeLong, T. P. Urbach, M. Kutas, Nat. Neurosci. 8, 1117-1121 (2005)]. Amid a broader controversy about the role of word-form prediction in comprehension, those findings were recently challenged by a failed putative direct replication attempt [M. S. Nieuwland et al., eLife 7, e33468 (2018); nine labs, one experiment, and 2.6e4 observations]. To better understand the empirical justification for positing an association between prenominal article predictability and scalp potentials, we conducted a wide-ranging exploratory data analysis (EDA), pooling our original data with extant data from two followup studies (one lab, three experiments, and 1.2e4 observations). We modeled the time course of article predictability in the single-trial data by fitting linear mixed-effects regression (LMER) models at each time point and scalp location spanning a 3-s interval before, during, and after the article. Model comparisons based on Akaike information criteria (AIC) and slope-regression ERPs [rERPs; N. J. Smith, M. Kutas, Psychophysiology 52, 157-168 (2015)] provide substantial empirical support for a small positive association between article predictability and scalp potentials approximately 300 to 500 ms after article onset, predominantly over bilateral posterior scalp. We think this effect may reasonably be attributed to prediction of upcoming word forms.

Comprehending surprising sentences: sensitivity of post-N400 positivities to contextual congruity and semantic relatedness.

Any proposal for predictive language comprehension must address receipt of less expected information. While a relationship between the N400 and sentence predictability is well established, a clear picture is still emerging of the link between post-N400 positivities (PNPs) and processing of semantically unexpected words, as well as any relation to other not-specifically-linguistic and/or syntactic late positivities. The current study employs event-related brain potentials (ERPs) to congruent and anomalous words to assess the impacts of semantic relatedness and contextual plausibility on processing unpredictable sentences. We observe PNPs with different scalp topographies to plausible unexpected words unrelated to predictable continuations (anterior PNP) and to anomalous words, regardless of, but delayed by, relatedness (posterior PNP). We offer functional explanations that reconcile inconsistencies with reported PNP findings and place added constraints on the anterior PNP's proposed link to inhibitory processing. We also suggest a testable general cognitive account for the posterior PNP.

Similar time courses for word form and meaning preactivation during sentence comprehension.

Current psycholinguistic research generally acknowledges that aspects of sentence comprehension benefit from neural preactivation of different types of information. However, despite strong support from a number of studies, routine specific word form preactivation has been challenged by Ito, Corley, Pickering, Martin, and Nieuwland (2016). They suggest that word form prediction is contingent upon having enough processing time and resources (afforded by slower input rates) to progress through unidirectional, productionlike stages of comprehension to arrive at word forms via semantic feature preactivation. This conclusion is based on findings from their ERP study, which used a related anomaly paradigm and reported form preactivation at a slow (700 ms) word presentation rate but not a faster one (500 ms). The present experimental design is a conceptual replication of Ito et al. (2016), testing young adults by measuring ERP amplitudes to unpredictable words related either semantically/associatively or orthographically to predictable sentence continuations, relative to unrelated continuations. Results showed that, at a visual presentation rate of two words per second, both types of related words show similarly reduced N400s, as well as varying degrees of increased posterior post-N400 positivity. These findings indicate that word form preactivation during sentence comprehension is detectable along a similar time course as semantic feature preactivation, and such processing does not necessarily require additional time beyond that afforded by near-normal reading rates.

Hemispheric differences and similarities in comprehending more and less predictable sentences.

With a growing literature demonstrating the predictive nature of language processing, the current study examines contributions of the brain's two hemispheres in processing more and less probable sentence continuations. Specifically, we use the ERP method in conjunction with the visual half-field paradigm to test for hemispheric utilization of sentential constraint to (pre-)activate lexical information and resolve meaning. Taking advantage of the N400's semantic sensitivities, we find support for both hemispheres exhibiting remarkably similar involvement, across a range of message level constraint, in meaning construction. In contrast, hemispheric ERP patterns at a later processing stage differed, as reflected in an anterior post-N400 positivity (PNP) to constraint violations for words presented to the right but not left visual field (indicating a left hemisphere processing bias). We show here that hemispheric involvement in predictive sentence comprehension varies at different stages of word processing, and we examine these patterns' (in)consistencies with findings from the hemi-field and central visual presentation literature.

Quantifiers are incrementally interpreted in context, more than less.

Language interpretation is often assumed to be incremental. However, our studies of quantifier expressions in isolated sentences found N400 event-related brain potential (ERP) evidence for partial but not full immediate quantifier interpretation (Urbach & Kutas, 2010). Here we tested similar quantifier expressions in pragmatically supporting discourse contexts (Alex was an unusual toddler. Most/Few kids prefer sweets/vegetables…) while participants made plausibility judgments (Experiment 1) or read for comprehension (Experiment 2). Control Experiments 3A (plausibility) and 3B (comprehension) removed the discourse contexts. Quantifiers always modulated typical and/or atypical word N400 amplitudes. However, only the real-time N400 effects only in Experiment 2 mirrored offline quantifier and typicality crossover interaction effects for plausibility ratings and cloze probabilities. We conclude that quantifier expressions can be interpreted fully and immediately, though pragmatic and task variables appear to impact the speed and/or depth of quantifier interpretation.

Close, but no garlic: Perceptuomotor and event knowledge activation during language comprehension.

Recent research has shown that language comprehension is guided by knowledge about the organization of objects and events in long-term memory. We use event-related brain potentials (ERPs) to determine the extent to which perceptuomotor object knowledge and event knowledge are immediately activated during incremental language processing. Event-related but anomalous sentence continuations preceded by single-sentence event descriptions elicited reduced N400s, despite their poor fit within local sentence contexts. Anomalous words sharing particular sensory or motor attributes with contextually expected words also elicited reduced N400s, despite being inconsistent with global context (i.e., event information). We rule out plausibility as an explanation for both relatedness effects. We show that perceptuomotor-related facilitation is not due to lexical priming between words in the local context and the target or to associative or categorical relationships between expected and unexpected targets. Overall our results are consistent with the immediate and incremental activation of perceptual and motor object knowledge and generalized event knowledge during sentence processing.

Metaphors are physical and abstract: ERPs to metaphorically modified nouns resemble ERPs to abstract language.

Metaphorical expressions very often involve words referring to physical entities and experiences. Yet, figures of speech such as metaphors are not intended to be understood literally, word-by-word. We used event-related brain potentials (ERPs) to determine whether metaphorical expressions are processed more like physical or more like abstract expressions. To this end, novel adjective-noun word pairs were presented visually in three conditions: (1) Physical, easy to experience with the senses (e.g., "printed schedule"); (2) Abstract, difficult to experience with the senses (e.g., "conditional schedule"); and (3) novel Metaphorical, expressions with a physical adjective, but a figurative meaning (e.g., "thin schedule"). We replicated the N400 lexical concreteness effect for concrete vs. abstract adjectives. In order to increase the sensitivity of the concreteness manipulation on the expressions, we divided each condition into high and low groups according to rated concreteness. Mirroring the adjective result, we observed a N400 concreteness effect at the noun for physical expressions with high concreteness ratings vs. abstract expressions with low concreteness ratings, even though the nouns per se did not differ in lexical concreteness. Paradoxically, the N400 to nouns in the metaphorical expressions was indistinguishable from that to nouns in the literal abstract expressions, but only for the more concrete subgroup of metaphors; the N400 to the less concrete subgroup of metaphors patterned with that to nouns in the literal concrete expressions. In sum, we not only find evidence for conceptual concreteness separable from lexical concreteness but also that the processing of metaphorical expressions is not driven strictly by either lexical or conceptual concreteness.

Predictability, plausibility, and two late ERP positivities during written sentence comprehension.

Van Petten and Luka's (2012, International Journal of Psychophysiology, 83(2), 176-190) literature survey of late positive ERP components elicited by more or less predictable words during sentence processing led them to propose two topographically and functionally distinct positivities: a parietal one associated with semantically incongruent words related to semantic reanalysis and a frontal one with unknown significance associated with congruent but lexically unpredicted words. With the goal of testing this hypothesis within a single set of experimental materials and participants, we report results from two ERP studies: Experiment 1, a post-hoc analysis of a dataset that varied on dimensions of both cloze probability (predictability) and plausibility, and Experiment 2, a follow-up study in which these factors were manipulated in a controlled fashion. In both studies, we observed distinct post-N400 positivities: a more anterior one to plausible, but not anomalous, low cloze probability sentence medial words, and a more posterior one to semantically anomalous sentence continuations. Taken together with an observed canonical cloze-modulated N400, these dual positivities indicate a dissociation between brain processes relating to written words׳ sentential predictability versus plausibility, clearly an important distinction for any viable neural or psycholinguistic model of written sentence processing.

Pre-processing in sentence comprehension: Sensitivity to likely upcoming meaning and structure.

For more than a decade, views of sentence comprehension have been shifting toward wider acceptance of a role for linguistic pre-processing-that is, anticipation, expectancy, (neural) pre-activation, or prediction-of upcoming semantic content and syntactic structure. In this survey, we begin by examining the implications of each of these "brands" of predictive comprehension, including the issue of potential costs and consequences to not encountering highly constrained sentence input. We then describe a number of studies (many using online methodologies) that provide results consistent with prospective sensitivity to various grains and levels of semantic and syntactic information, acknowledging that such pre-processing is likely to occur in other linguistic and extralinguistic domains, as well. This review of anticipatory findings also includes some discussion on the relationship of priming to prediction. We conclude with a brief examination of some possible limits to prediction, and with a suggestion for future work to probe whether and how various strands of prediction may integrate during real-time comprehension.

Thinking ahead or not? Natural aging and anticipation during reading.

Despite growing evidence of young adults neurally pre-activating word features during sentence comprehension, less clear is the degree to which this generalizes to older adults. Using ERPs, we tested for linguistic prediction in younger and older readers by means of indefinite articles (a's and an's) preceding more and less probable noun continuations. Although both groups exhibited cloze probability-graded noun N400s, only the young showed significant article effects, indicating probabilistic sensitivity to the phonology of anticipated upcoming nouns. Additionally, both age groups exhibited prolonged increased frontal positivities to less probable nouns, although in older adults this effect was prominent only in a subset with high verbal fluency (VF). This ERP positivity to contextual constraint violations offers additional support for prediction in the young. For high VF older adults, the positivity may indicate they, too, engage in some form of linguistic pre-processing when implicitly cued, as may have occurred via the articles.

Overlapping dual ERP responses to low cloze probability sentence continuations.

In 2005, DeLong, Urbach, and Kutas took advantage of the a/an English indefinite article phonological alternation and the sensitivities of the N400 ERP component to show that readers can neurally preactivate individual words of a sentence (including nouns and their prenominal indefinite articles) in a graded fashion with a likelihood estimated from the words' offline probabilities as sentence continuations. Here we report an additional finding from that study: a prolonged ERP frontal positivity to less probable noun continuations. We suggest that this positivity is consistent with hypotheses that additional neural processing may be invoked when highly expected continuations are not encountered in the input and speculate briefly on possible functional correlates.

Probabilistic word pre-activation during language comprehension inferred from electrical brain activity.

Despite the numerous examples of anticipatory cognitive processes at micro and macro levels in many animal species, the idea that anticipation of specific words plays an integral role in real-time language processing has been contentious. Here we exploited a phonological regularity of English indefinite articles ('an' precedes nouns beginning with vowel sounds, whereas 'a' precedes nouns beginning with consonant sounds) in combination with event-related brain potential recordings from the human scalp to show that readers' brains can pre-activate individual words in a graded fashion to a degree that can be estimated from the probability that each word is given as a continuation for a sentence fragment offline. These findings are evidence that readers use the words in a sentence (as cues to their world knowledge) to estimate relative likelihoods for upcoming words.