Do people routinely pre-activate the meaning and even the phonological form of upcoming words? The most acclaimed evidence for phonological prediction comes from a 2005 Nature Neuroscience publication by DeLong, Urbach and Kutas, who observed a graded modulation of electrical brain potentials (N400) to nouns and preceding articles by the probability that people use a word to continue the sentence fragment (‘cloze’). In our direct replication study spanning 9 laboratories (N=334), pre-registered replication-analyses and exploratory Bayes factor analyses successfully replicated the noun-results but, crucially, not the article-results. Pre-registered single-trial analyses also yielded a statistically significant effect for the nouns but not the articles. Exploratory Bayesian single-trial analyses showed that the article-effect may be non-zero but is likely far smaller than originally reported and too small to observe without very large sample sizes. Our results do not support the view that readers routinely pre-activate the phonological form of predictable words.

Composing sentence meaning is easier for predictable words than for unpredictable words. Are predictable words genuinely predicted, or simply more plausible and therefore easier to integrate with sentence context? We addressed this persistent and fundamental question using data from a recent, large-scale ( n = 334) replication study, by investigating the effects of word predictability and sentence plausibility on the N400, the brain's electrophysiological index of semantic processing. A spatio-temporally fine-grained mixed-effect multiple regression analysis revealed overlapping effects of predictability and plausibility on the N400, albeit with distinct spatio-temporal profiles. Our results challenge the view that the predictability-dependent N400 reflects the effects of either prediction or integration, and suggest that semantic facilitation of predictable words arises from a cascade of processes that activate and integrate word meaning with context into a sentence-level meaning. This article is part of the theme issue ‘Towards mechanistic models of meaning composition’.

The present study examines the brain-level representation and composition of meaning in scalar quantifiers (e.g., some), which have both a semantic meaning (at least one) and a pragmatic meaning (not all). We adopted a picture-sentence verification design to examine event-related potential (ERP) effects of reading infelicitous quantifiers for which the semantic meaning was correct with respect to the context but the pragmatic meaning was not, compared to quantifiers for which the semantic meaning was inconsistent with the context and no additional pragmatic meaning is available. In the first experiment, only pragmatically inconsistent quantifiers, not semantically inconsistent quantifiers, elicited a sustained posterior negative component. This late negativity contrasts with the N400 effect typically elicited by nouns that are incongruent with their context, suggesting that the recognition of scalar implicature errors elicits a qualitatively different ERP signature than the recognition of lexico-semantic errors. We hypothesize that the sustained negativity reflects cancellation of the pragmatic inference and retrieval of the semantic meaning. In our second experiment, we found that the process of re-interpreting the quantifier was independent from lexico-semantic processing: the N400 elicited by lexico-semantic violations was not modulated by the presence of a pragmatic inconsistency. These findings suggest that inferential pragmatic aspects of meaning are processed using different mechanisms than lexical or combinatorial semantic aspects of meaning, that inferential pragmatic meaning can be realized rapidly, and that the computation of meaning involves continuous negotiation between different aspects of meaning.

No abstract

The present study examines the online realization of pragmatic meaning using event-related potentials (ERPs). Participants read sentences including the English quantifier some, which has both a semantic meaning (at least one) and a pragmatic meaning (not all). Unlike previous ERP studies of this phenomenon, sentences in the current study were evaluated not in terms of their truth with respect to the real world, but in terms of their consistency with a picture presented before the sentence. Sentences (such as “The boy cut some of the steaks in this story”) were constructed such that either 1) both the semantic and pragmatic interpretations were true with respect to the preceding picture (when the boy in fact cut some but not all of the steaks); 2) neither interpretation was true (when the boy in fact cut none of the steaks); or 3) the semantic interpretation was true but the pragmatic interpretation false (when the boy in fact cut all of the steaks). ERPs at the object word, which determined whether the sentence was consistent with the story, showed the largest N400 effect for objects that made the sentence false, whereas they showed an intermediate effect for objects that made the sentence false under the pragmatic interpretation but true under the semantic interpretation. The results suggest that this pragmatic aspect of meaning is computed online and integrated into the sentence model rapidly enough to influence comprehension of later words.

In current theories of language comprehension, people routinely and implicitly predict upcoming words by pre-activating their meaning, morpho-syntactic features and even their specific phonological form. To date the strongest evidence for this latter form of linguistic prediction comes from a 2005 Nature Neuroscience landmark publication by DeLong, Urbach and Kutas, who observed a graded modulation of article- and noun-elicited electrical brain potentials (N400) by the pre-determined probability that people continue a sentence fragment with that word ('cloze'). In a direct replication study spanning 9 laboratories (N=334), we failed to replicate the crucial article-elicited N400 modulation by cloze, while we successfully replicated the commonly-reported noun-elicited N400 modulation. This pattern of failure and success was observed in a pre-registered replication analysis, a pre-registered single-trial analysis, and in exploratory Bayesian analyses. Our findings do not support a strong prediction view in which people routinely pre-activate the phonological form of upcoming words, and suggest a more limited role for prediction during language comprehension.

Composing sentence meaning is easier for predictable words than for unpredictable words. Are predictable words genuinely predicted, or simply more plausible and therefore easier to integrate with sentence context? We addressed this persistent and fundamental question using data from a recent, large-scale (N = 334) replication study, by investigating the effects of word predictability and sentence plausibility on the N400, the brain's electrophysiological index of semantic processing. A spatiotemporally fine-grained mixedeffects multiple regression analysis revealed overlapping effects of predictability and plausibility on the N400, albeit with distinct spatiotemporal profiles. Our results challenge the view that the predictability-dependent N400 reflects the effects of either prediction or integration, and suggest that semantic facilitation of predictable words arises from a cascade of processes that activate and integrate word meaning with context into a sentence-level meaning. Keywords Predictability; Plausibility; Semantic Similarity; N400Composing sentence meaning is easier with predictable words than with unpredictable words: for example, 'bicycle' is easier to process than 'elephant' in "You never forget how to ride a bicycle/an elephant once you've learned." This effect of predictability can be observed in behavioural measures of comprehension such as reading times [1] and on amplitude modulation of the N400 ERP component [2]. The N400 is a negative-going and centroparietally distributed component of the ERP, which occurs approximately 200-500 ms after word onset and that is strongly associated with lexico-semantic processing [2-3]. Predictable words elicit smaller N400 amplitude than unpredictable words, suggesting facilitation during semantic processing. However, to what extent is such an effect of predictability driven by other relationships between a word and its context, such as the plausibility of the described event? We addressed this issue by re-analysing data from a large-scale (N=334) replication study [4]. In a temporally fine-grained analysis, we explored whether predictability, plausibility, and semantic similarity have dissociable effects on word-elicited ERP activity and how these effects unfold over time.Word predictability is commonly operationalized as 'cloze probability', the probability of being used in a non-speeded, offline sentence completion test. The correlation between word predictability and N400 amplitude is well-established [5], with some studies reporting correlations as high as or even higher than r = .8 [2,[6][7]. Such results are often considered to demonstrate effects of prediction: people pre-activate a word fully before it appears (as when a specific lexical form can be strongly anticipated in a highly constraining context) or partially (as when some semantic features are activated due to passive spreading of information). Prediction facilitates the semantic activation processes that are initiated when the word appears and are reflected in N400 activity [5].

Most investigations of the representation and processing of speech sounds focus on their segmental representations, and considerably less is known about the representation of suprasegmental phenomena (e.g., Mandarin tones). Here we examine the mismatch negativity (MMN) response to the contrast between Mandarin Tone 3 (T3) and other tones using a passive oddball paradigm. Because the MMN response has been shown to be sensitive to the featural contents of speech sounds in a way that is compatible with theories of phonological representations, here, we test the predictions of such theories regarding suprasegmental phenomena. Assuming T3 to be underspecified in Mandarin (because it has variable surface representations and low pitch), we predicted that an asymmetric MMN response would be elicited when T3 is contrasted with another tone. In 2 of our 3 experiments, this was observed, but in non-Mandarin-speaking participants as well as native speakers, suggesting that the locus of the effect was perceptual (acoustic or phonetic) rather than phonological. In a third experiment, the predicted asymmetry was limited to native speakers. These results highlight the importance of distinguishing phonological and perceptual contributions to MMN asymmetries, but also demonstrate a role of abstract phonological representations in which certain information is underspecified in long-term memory.