When speakers choose how to refer to a previously mentioned character in the presence of several potential antecedents, they experience multiple competing pressures. Predictability is an important factor in this process, and its role in shaping speakers’ choices has been the topic of ongoing debate. Here, we examine how event predictability, viewed as a specific form of information predictability, shapes both the production and interpretation of anaphoric references. In Experiments 1a and 1b (free production), participants described events involving interacting characters. They used overt forms of reference, such as definite noun phrases, more often when describing surprising events compared to familiar ones. Speakers further produced fewer phonologically non-overt forms, such as zero anaphors, when events were surprising. This pattern is consistent with sensitivity to possible listener misinterpretation of non-overt forms despite them being grammatically unambiguous. In Experiments 2a and 2b (speech perception), listeners heard descriptions of the same kinds of events under varying levels of auditory noise. Listeners were more likely to interpret utterances containing unambiguous zero anaphors as if they contained ambiguous pronouns when this interpretation allowed them to construe a more plausible event, particularly under noisy conditions. Taken together, the results show that event-level expectations shape both referential production and interpretation: speakers favor more overt forms of reference for less predictable events, whereas listeners use event plausibility when interpreting potentially degraded input. This convergence between production and comprehension is consistent with audience-design accounts of reference and extends noisy-channel accounts of communication to the domain of anaphoric interpretation.
The phonological structure of a word can influence the ease with which it is learned. While there is evidence that words with identical consonants are better recalled than words with dissimilar consonants, it remains unclear whether similarity (i.e., partial identity) between consonants facilitates or hinders acquisition. Both possibilities are supported by different memory models: some posit that shared features within a sequence should lead to overlap and interference, while others suggest that repeated features can reinforce each other's representation. We tested these competing predictions in two experiments where English-speaking adults learned object labels containing repeated consonants (e.g., gegu), place-agreeing consonants (e.g., geku), or place-disagreeing consonants (e.g., getu). Experiment 1 used artificial lexicons in which all three word types were equally represented, whereas in Experiment 2, lexicons were systematically skewed towards one of the word types. In both experiments, repeated-consonant words were recalled more accurately than place-disagreeing words; words with place-agreeing consonants showed intermediate performance and did not differ reliably from either of their counterparts. Furthermore, analyses of erroneous trials revealed a tendency for participants to overproduce place-agreeing forms across all lexicons. These findings suggest that consonantal similarity is not detrimental to novel word learning and point towards a reconstruction bias favouring place agreement when recall is uncertain.
Contact between speakers and languages may play a key role in driving language change. During interaction, speakers tend to adapt to their interlocutors' speech, which can cumulatively lead to longer-term changes. Yet it remains unclear how innovations emerge, and who drives their adoption. These questions are addressed through two experiments on Italian motion-events framing. Study 1 examines German-Italian bilinguals with varying cumulative German contact, exploring whether exposure to innovative satellite-framed constructions triggers adaptation within Italian and across languages. Study 2 tests Italian monolinguals to assess adaptation in the absence of cross-linguistic influence. Results show that both bilinguals and monolinguals adapt to innovative structures at an abstract level, but only bilinguals with a high degree of German contact maintain these changes in a post-test, highlighting the role of bilingual experience. Findings suggest language contact may enhance the uptake of innovations, providing insight into the cognitive mechanisms underlying language change.
Understanding how comprehenders interpret linguistic input in real time has long been a focus of sentence processing research. The recently-proposed Resource-Rational Lossy-context Surprisal (RR-LCS) model can be applied for making predictions at the level of individual readers: it predicts that individuals with low working memory capacity (WMC) have a lossy representation of the previous context and may therefore experience high surprisal at a long-distance dependency, while individuals with high WMC will experience a lower surprisal effect as this dependent to them may still be highly predictable from previous context. To date, these predictions have not been empirically tested. Here, we evaluate them using a controlled A-Maze task (Experiment 1) and an analysis of reading more naturalistic text based on the InDiCo corpus (Experiment 2). In Experiment 1, we investigated how WMC influences sentence processing across short, long, and very long dependency conditions. The results revealed that low-WMC participants initially showed a strong locality effect between the short and long conditions, which decreased over time due to adaptation. High-WMC participants demonstrated locality effects between both the short-long and long-very long conditions from the start, with these effects gradually diminishing as their memory allocation became more efficient. These findings align with the RR-LCS model's predictions, showing that high-WMC individuals maintained larger context sizes, such that they could more easily process long-distance dependencies, whereas low-WMC individuals, constrained by smaller context sizes, experienced pronounced locality effects. Experiment 2 confirmed that WMC modulates sensitivity to lossy-context surprisal (LCS), with LCS improving model fit particularly strongly for low-WMC participants. These findings underscore the significance of individual differences in WMC for processing complex syntactic structures. By combining experimental data with computational modeling, we provide a deeper understanding of how WMC shapes language comprehension.
Comprehenders often retain incorrect interpretations of garden-path sentences, suggesting that sentence representations are not always fully revised after misanalysis. A common assumption is that the additional processing time observed at disambiguation reflects effortful syntactic repair and should therefore improve final interpretations. We tested this prediction by examining whether reading times predict successful recovery from misanalysis. Participants read ambiguous and unambiguous sentences in a self-paced reading task and answered comprehension questions probing four components of interpretation: inhibition of the initial misanalysis, recovery of the ambiguous region, and correct and incorrect interpretations of the disambiguating region. Reading times were analyzed at the disambiguating, spillover, and sentence-final regions. Garden-path sentences produced robust processing slowdowns and reduced comprehension accuracy. However, longer reading times did not increase the likelihood of obtaining the correct interpretation. Slower reading improved accuracy equally for ambiguous and unambiguous sentences and even in sentence regions unrelated to reanalysis, indicating a general benefit of careful processing. Critically, for the correct interpretation of the ambiguous region, processing time had no effect in garden-path sentences: readers either recovered the intended analysis or retained a misinterpretation regardless of how long they processed the input. These findings challenge the assumption that processing slowdowns index successful syntactic revision and instead suggest that reanalysis outcomes are largely time-independent, reflecting whether a viable alternative interpretation is found rather than how long readers process the disambiguating input.
