Effects Of Semantic Similarity Research Articles

Is predicting during language processing worth it? Effects of cloze probability and semantic similarity on failed predictions.

Prediction during language processing has been hypothesized to lead to processing benefits. These possible benefits have led to several prominent theories that center around prediction as an essential mechanism in language processing. Such theories typically assume predicting is better than not predicting at all, but do not always account for the potential processing costs from failed predictions. Predicting wrongly can be costly, but the cost may depend on how wrong the prediction was. Across three experiments, we manipulate cloze probability, semantic relatedness, and language modality (production vs. comprehension) to determine whether predicting almost correctly is better than predicting completely incorrectly, and if so, if predicting almost correctly is better than not predicting at all. Results showed that when a predicted ending is replaced with a related term, it is processed faster than when it is replaced with an unrelated term, but that related term is not named more quickly than when it appears after a low constraint sentence. This pattern held regardless of whether participants were asked to produce the sentence-final term by naming a picture (Experiments 1 and 2), or if they were asked to perform a semantic classification of the sentence-final word (Experiment 3). Thus, predicting almost correctly is better than predicting completely incorrectly, but it's not better than not predicting at all. This carries implications for current accounts that argue for processing benefits of prediction during language processing, and suggests that prediction may be used to fine-tune the language system rather than to speed language processing. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Strong Prediction: Language Model Surprisal Explains Multiple N400 Effects.

Theoretical accounts of the N400 are divided as to whether the amplitude of the N400 response to a stimulus reflects the extent to which the stimulus was predicted, the extent to which the stimulus is semantically similar to its preceding context, or both. We use state-of-the-art machine learning tools to investigate which of these three accounts is best supported by the evidence. GPT-3, a neural language model trained to compute the conditional probability of any word based on the words that precede it, was used to operationalize contextual predictability. In particular, we used an information-theoretic construct known as surprisal (the negative logarithm of the conditional probability). Contextual semantic similarity was operationalized by using two high-quality co-occurrence-derived vector-based meaning representations for words: GloVe and fastText. The cosine between the vector representation of the sentence frame and final word was used to derive contextual cosine similarity estimates. A series of regression models were constructed, where these variables, along with cloze probability and plausibility ratings, were used to predict single trial N400 amplitudes recorded from healthy adults as they read sentences whose final word varied in its predictability, plausibility, and semantic relationship to the likeliest sentence completion. Statistical model comparison indicated GPT-3 surprisal provided the best account of N400 amplitude and suggested that apparently disparate N400 effects of expectancy, plausibility, and contextual semantic similarity can be reduced to variation in the predictability of words. The results are argued to support predictive coding in the human language network.

The Semantic Similarity Effect on Short-Term Memory: Null Effects of Affectively Defined Semantic Similarity.

Studies on short-term memory have repeatedly demonstrated the beneficial effect of semantic similarity. Although the effect seems robust, the aspects of semantics targeted by these studies (e.g., categorical structure, associative relationship, or dimension of meaning) should be clarified. A recent meta-regression study inspired by Osgood's view, which highlights affective dimensions in semantics, introduced a novel index for quantifying semantic similarity using affective values. Building on the results of the meta-regression of past studies' data with that index, this study predicts that semantic similarity is deleterious to short-term memory if it is manipulated by affective dimensions, after controlling for other confounding factors. This prediction was directly tested. The experimental results of the immediate serial recall task (Study 1) and immediate serial reconstruction of order task (Study 2) indicated null effects of semantic similarity by affective dimensions and thus falsified the prediction. These results suggest that semantic similarity based on affective dimensions is negligible.

Semantic representations in inferior frontal and lateral temporal cortex during picture naming, reading, and repetition.

Reading, naming, and repetition are classical neuropsychological tasks widely used in the clinic and psycholinguistic research. While reading and repetition can be accomplished by following a direct or an indirect route, pictures can be named only by means of semantic mediation. By means of fMRI multivariate pattern analysis, we evaluated whether this well-established fundamental difference at the cognitive level is associated at the brain level with a difference in the degree to which semantic representations are activated during these tasks. Semantic similarity between words was estimated based on a word association model. Twenty subjects participated in an event-related fMRI study where the three tasks were presented in pseudo-random order. Linear discriminant analysis of fMRI patterns identified a set of regions that allow to discriminate between words at a high level of word-specificity across tasks. Representational similarity analysis was used to determine whether semantic similarity was represented in these regions and whether this depended on the task performed. The similarity between neural patterns of the left Brodmann area 45 (BA45) and of the superior portion of the left supramarginal gyrus correlated with the similarity in meaning between entities during picture naming. In both regions, no significant effects were seen for repetition or reading. The semantic similarity effect during picture naming was significantly larger than the similarity effect during the two other tasks. In contrast, several regions including left anterior superior temporal gyrus and left ventral BA44/frontal operculum, among others, coded for semantic similarity in a task-independent manner. These findings provide new evidence for the dynamic, task-dependent nature of semantic representations in the left BA45 and a more task-independent nature of the representational activation in the lateral temporal cortex and ventral BA44/frontal operculum.

The effect of semantic and physical similarity on vocabulary learning

ABSTRACT Choosing which words to teach is a key consideration for language teachers and materials writers. Some studies have shown that teaching words in semantically related clusters can make learning more difficult. However, others argue it is the physical similarity of the referents of words that causes confusion. Importantly, studies have employed different methodologies, which may have contributed to the contradictory results. The present study addressed this contradiction by examining the target words from with the methodology of to determine the extent to which semantic relatedness and physical relatedness contribute to the learning burden of L2 vocabulary. Results showed that (1) semantic and physical conditions increased burden relative to an unrelated condition, (2) the greatest burden was found when words were both semantically and physical related, and (3) there was no difference in burden between semantically and physically related words. Pedagogical implications are discussed.

A common neural code for meaning in discourse production and comprehension

How does the brain code the meanings conveyed by language? Neuroimaging studies have investigated this by linking neural activity patterns during discourse comprehension to semantic models of language content. Here, we applied this approach to the production of discourse for the first time. Participants underwent fMRI while producing and listening to discourse on a range of topics. We used a distributional semantic model to quantify the similarity between different speech passages and identified where similarity in neural activity was predicted by semantic similarity. When people produced discourse, speech on similar topics elicited similar activation patterns in a widely distributed and bilateral brain network. This network was overlapping with, but more extensive than, the regions that showed similarity effects during comprehension. Critically, cross-task neural similarities between comprehension and production were also predicted by similarities in semantic content. This result suggests that discourse semantics engages a common neural code that is shared between comprehension and production. Effects of semantic similarity were bilateral in all three RSA analyses, even while univariate activation contrasts in the same data indicated left-lateralised BOLD responses. This indicates that right-hemisphere regions encode semantic properties even when they are not activated above baseline. We suggest that right-hemisphere regions play a supporting role in processing the meaning of discourse during both comprehension and production.

Bidirectional cross-linguistic influence with different-script languages: Evidence from eye tracking

AbstractThis study compared patterns of nonselective cross-language activation in L1 and L2 visual word recognition with different-script bilinguals. The aim was to determine (1) whether lexical processing is nonselective in the L1 (as in L2), and (2) if the same cross-linguistic factors affected processing similarly in each language. To examine the time course of activation, eye movements were tracked during lexical decision. Thirty-two Japanese–English bilinguals responded to 250 target words in Japanese and in English. The same participants and items (i.e., cognate translation equivalents) were used to directly compare L1 and L2 processing. Response latencies as well as eye movements representing early and late processing were analyzed using mixed-effects regression modeling. Similar cross-linguistic effects, namely cognate word frequency, phonological similarity, and semantic similarity, were found in both languages. These factors affected processing to different degrees in each language, however. While cognate frequency was significant as early as the first fixation, effects of cross-linguistic phonological and semantic similarity arose later in time. Increased phonological similarity slowed responses in L2 but speeded them in L1, while greater semantic overlap was facilitatory in both languages. Results are discussed from the perspective of the BIA+ model of visual word recognition.

The effects of semantic similarity on Mandarin speakers' referential expressions.

Previous research has found apparently contradictory effects of a semantically similar competitor on how people refer to previously mentioned entities. To address this issue, we conducted two picture-description experiments in spoken Mandarin. In Experiment 1, participants saw pictures and heard sentences referring to both the target referent and a competitor, and then described actions involving only the target referent. They produced fewer omissions and more repeated noun phrases when the competitor was semantically similar to the target referent than otherwise. In Experiment 2, participants saw introductory pictures and heard sentences referring to only the target referent, and then described actions involving both the target referent and a competitor. They produced more omissions and fewer pronouns when the competitor was semantically similar to the target referent than otherwise. We interpret the results in terms of the representation of discourse entities and the stages of language production.

Verbal working memory encodes phonological and semantic information differently

Working memory (WM) is often tested through immediate serial recall of word lists. Performance in such tasks is negatively influenced by phonological similarity: People more often get the order of words wrong when they are phonologically similar to each other (e.g., cat, fat, mat). This phonological-similarity effect shows that phonology plays an important role for the representation of serial order in these tasks. By contrast, semantic similarity usually does not impact performance negatively. To resolve and understand this discrepancy, we tested the effects of phonological and semantic similarity for the retention of positional information in WM. Across six experiments (all Ns = 60 young adults), we manipulated between-item semantic and phonological similarity in tasks requiring participants to form and maintain new item-context bindings in WM. Participants were asked to retrieve items from their context, or the contexts from their item. For both retrieval directions, phonological similarity impaired WM for item-context bindings across all experiments. Semantic similarity did not. These results demonstrate that WM encodes phonological and semantic information differently. We propose a WM model accounting for semantic-similarity effects in WM, in which semantic knowledge supports WM through activated long-term memory.

A joint investigation of facilitation and interference effects of semantic and phonological similarity in a continuous naming task.

Semantic and phonological similarity effects provide critical constraints on the mechanisms underlying language production. In the present study, we jointly investigated effects of semantic and phonological similarity using the continuous naming task. In the semantic condition, Chinese Mandarin speakers named a list of pictures composed of 12 semantic category sets with 5 items from each semantic category, while in the phonological condition, they named a list of pictures from 12 phonological sets of 5 items sharing a spoken syllable. Related pictures occurred on adjacent trials, or were separated by 2, 4, or 6 unrelated pictures. Similar results were found across the semantic and phonological conditions: naming was facilitated by the directly preceding production of a related picture. For nonconsecutive related responses, naming latency increased linearly as a function of the number of preceding production instances of related pictures. Parallel patterns of facilitation and interference effects arising from semantic and phonological similarity suggest universal principles which govern language production. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Ultra-rapid and automatic interplay between L1 and L2 semantics in late bilinguals: EEG evidence

Converging behavioral and neuroimaging evidence suggests parallel activation of native (L1) and second (L2) language codes in bilinguals, with the modulation of the N400 as the most likely neural correlate of such L1-L2 interplay at lexico-semantic level. However, this relatively late effect may reflect secondary controlled processes, in contrast to earlier modulations found in monolinguals (<200 msec) indicative of fast and automatic lexico-semantic L1 access, which has so far not been documented for bilingualism. To address this, we investigated early neurophysiological crosslinguistic activation during bilingual word access. EEG signals were recorded from a group of late bilinguals during a masked-priming crosslinguistic task in which L1 (Russian) words were presented as subliminal primes for 50 msec before L2 (English) target words. Prime-target pairs matched either phonologically only, semantically only, both phonologically and semantically, or did not match. Cluster-based random permutation analyses revealed a main effect of semantic similarity at 40–60 msec over centro-posterior scalp sites, reflecting a negative-going shift of ERP amplitudes for semantically similar prime-target pairs. Importantly, neural source reconstruction showed activations within a left-hemispheric network comprising the middle and superior temporal cortex and the angular gyrus as the most likely neural substrate of this early semantic effect. Furthermore, analyses also revealed significant differences over frontocentral sites for the main effect of semantic and phonological similarity, ranging from 312 to 356 and 380–444 msec respectively, thus confirming previously described N400 crosslinguistic effects. Our findings confirm the existence of an integrated brain network for the bilingual lexicon and reveal the earliest (∼50 msec) crosslinguistic effect reported so far, suggesting fast and automatic L1–L2 interplay, followed by later (possibly top-down controlled) processing stages.

Concurrent semantic priming and lexical interference for close semantic relations in blocked-cyclic picture naming: Electrophysiological signatures.

In the present study, we employed event-related brain potentials to investigate the effects of semantic similarity on different planning stages during language production. We manipulated semantic similarity by controlling feature overlap within taxonomical hierarchies. In a blocked-cyclic naming task, participants named pictures in repeated cycles, blocked in semantically close, distant, or unrelated conditions. Only closely related items, but not distantly related items, induced semantic blocking effects. In the first presentation cycle, naming was facilitated, and amplitude modulations in the N1 component around 140-180ms post-stimulus onset predicted this behavioral facilitation. In contrast, in later cycles, naming was delayed, and a negative-going posterior amplitude modulation around 250-350ms post-stimulus onset predicted this interference. These findings indicate easier object recognition or identification underlying initial facilitation and increased difficulties during lexical selection. The N1modulation was reduced but persisted in later cycles in which interference dominated, and the posterior negativity was also present in cycle 1 in which facilitation dominated, demonstrating concurrent effects of conceptual priming and lexical interference in all naming cycles. Our assumptions about the functional role these two opposing forces play in producing semantic context effects are further supported by the finding that the joint modulation of these two ERPs on naming latency exclusively emerged when naming closely related, but not unrelated items. The current findings demonstrate that close relations, but not distant taxonomic relations, induce stronger semantic blocking effects, and that temporally overlapping electrophysiological signatures reflect a trade-off between facilitatory priming and interfering lexical competition.

The recognition of coordinative compound words by learners of Chinese as a foreign language: A mixed methods study

AbstractLearners of Chinese as a foreign language (CFL) tend to swap the two characters within a coordinative compound word in verbal identification and written production. This mixed methods study not only investigated how CFL learners identified intercharacter orthographic and semantic relationships within two‐morpheme coordinative compound words, but also explored how they used decoding strategies to resolve confusion related to constituent characters within compound words. Thirty‐two English‐speaking learners of Chinese participated in this study. The quantitative strand using a lexical decision task revealed significant character transposition effects and intercharacter semantic similarity effects but nonsignificant intercharacter orthographic similarity effects. Such effects were supported and further complemented by qualitative findings from a semistructured interview on learners' thought processes for compound word recognition. The nonsignificant intercharacter orthographic similarity effects were attributable to the meaning relationships between the constituent characters that shared the same radicals. Taken together, the results suggested that CFL learners tended to perceive compound words as whole units. For pedagogical recommendations, this study emphasized the importance of sublexical radical knowledge, morphological awareness for self‐sufficient vocabulary building, and character handwriting for CFL learners' efficient reading development.

Narratives bridge the divide between distant events in episodic memory

Many studies suggest that information about past experience, or episodic memory, is divided into discrete units called “events.” Yet we can often remember experiences that span multiple events. Events that occur in close succession might simply be linked because of their proximity to one another, but we can also build links between events that occur farther apart in time. Intuitively, some kind of organizing principle should enable temporally distant events to become bridged in memory. We tested the hypothesis that episodic memory exhibits a narrative-level organization, enabling temporally distant events to be better remembered if they form a coherent narrative. Furthermore, we tested whether post-encoding memory consolidation is necessary to integrate temporally distant events. In three experiments, participants learned and subsequently recalled events from fictional stories, in which pairs of temporally distant events involving side characters (“sideplots”) either formed one coherent narrative or two unrelated narratives. Across participants, we varied whether recall was assessed immediately after learning, or after a delay: 24 hours, 12 hours between morning and evening (“wake”), or 12 hours between evening and morning (“sleep”). Participants recalled more information about coherent than unrelated narrative events, in most delay conditions, including immediate recall and wake conditions, suggesting that post-encoding consolidation was not necessary to integrate temporally distant events into a larger narrative. Furthermore, post hoc modeling across experiments suggested that narrative coherence facilitated recall over and above any effects of sentence-level semantic similarity. This reliable memory benefit for coherent narrative events supports theoretical accounts which propose that narratives provide a high-level architecture for episodic memory.

An investigation of semantic similarity in PBOC’s communication on RMB volatility

This study investigates the effects of semantic similarity in the quarterly reports of the People’s Bank of China (PBOC) on renminbi (RMB) exchange rate volatility. Semantic similarity measures the similarity between two documents, and the more similar the two documents, the higher is the value. Our empirical findings show that semantic similarity effects are time-varying. Although high semantic similarity in PBOC’s communication raised the exchange rate volatility before the 2015 exchange rate reform, it reduced the volatility after that the reform. We also find that the semantic similarities between different topics in PBOC’s communication exert different impacts on RMB volatility. Domestic economy has larger impacts before the 2015 exchange rate reform, though domestic money and financial markets starts to play a significant important role since the 2015 reform. We contribute to the literature by showing the effects of PBOC’s communication are time-varying and depend on the content of communication.

Second Language Vocabulary Acquisition: The Effects of Semantic Relatedness, Form Similarity, and Translation Direction

AbstractCharacteristics of vocabulary lists as well as study and test translation direction may affect the ease of learning second language (L2) vocabulary. We examined immediate and delayed test performance of first language (L1) English speakers learning a fixed set of L2 vocabulary placed on lists formed by crossing semantic relatedness (unrelated vs. related) with L2 orthographic form similarity (not similar vs. similar). During the study phase, half the participants translated from L2 to L1 and half from L1 to L2; tests were then taken in both directions. Semantic relatedness in the absence of form similarity improved accuracy when the first test taken translated from L2 to L1, and tended to hurt accuracy when the first test taken translated from L1 to L2; it sometimes increased confusion errors. Form similarity usually hurt accuracy and always increased confusion errors. The combination of the L1‐to‐L2 study direction with the optimal semantic and form conditions yielded the best long‐term performance.

The detrimental effect of semantic similarity in short-term memory tasks: A meta-regression approach.

The literature suggests that semantic similarity has a weak or null effect for immediate serial reconstruction and a facilitative effect for immediate serial recall. These observed semantic similarity effects are inconsistent with the assumptions of short-term memory (STM) models on the detrimental effect of similarity (e.g., confusion) and with observations of a robust detrimental effect of phonological similarity. Our review indicates that the experimental results are likely dependent on the manipulation strength for semantic similarity and that manipulations used in previous studies might have affected semantic assvociation as well as semantic similarity. To address these possible issues, two indices are proposed: (a) strength of manipulation on semantic similarity, gained by quantifying semantic similarity based on Osgood and associates' dimensional view of semantics, and (b) inter-item associative strength, a possible confounding factor. Our review and the results of a meta-regression analysis using these two indices suggest that semantic similarity has a detrimental effect on both serial reconstruction and serial recall, while semantic association, which is correlated with semantic similarity, contributes to an apparent facilitative effect. An effect that is not attributable to similarity or association was also implied. Review on item and order memory further suggests the facilitative effect of semantic association on item memory and the detrimental effect of the semantic similarity on order memory. Based on our findings, we propose a unified explanation of observations of semantic similarity effects for both serial reconstruction and serial recall that is in good accord with STM models.

The Effect of Semantic Similarity on Learning Ambiguous Words in a Second Language: An Event-Related Potential Study.

Ambiguous words have multiple meanings. How these multiple meanings interact with each other during ambiguous word learning remains unclear. The current study adopted an event-related potentials (ERPs) technique to explore whether there is an interaction between two meanings when learning second language (L2) ambiguous words and how semantic similarity affects ambiguous word learning. In order to explore this issue, Chinese–English bilinguals were asked to learn pseudowords, which were paired with either two related new meanings (polysemes), two unrelated new meanings (homonyms), or one single new meaning (monosomies) over 2 consecutive days. ERP results revealed that learning the second meaning of a homonym induced a more negative N400 than the first meaning; learning the second meaning of a polyseme tended to produce a more positive late component (LPC) than the first meaning. These results indicate that the first meaning of homonyms may interfere with learning their second meaning. However, the first meaning of polysemous words may facilitate learning their second meaning. The current findings suggest that different mechanisms might be involved in learning L2 homonyms and polysemes.

Global semantic similarity effects in recognition memory: Insights from BEAGLE representations and the diffusion decision model

Abstract Recognition memory models posit that false alarm rates increase as the global similarity between the probe cue and the contents of memory is increased. Global similarity predictions have been commonly tested using category length designs where it has been found that false alarm rates increase as the number of studied items from a common category is increased. In this work, we explored global similarity predictions within unstructured lists of words using representations from the BEAGLE model (Jones & Mewhort, 2007). BEAGLE differs from traditional semantic space models in that it contains two types of representations: item vectors, which encode unordered co-occurrence, and order vectors, in which words are similar to the extent to which they are share neighboring words in the same relative positions. Global similarity among item and order vectors was regressed onto drift rates in the diffusion decision model (DDM: Ratcliff, 1978), which unifies both response times and accuracy. We implemented this model in a hierarchical Bayesian framework across seven datasets with lists composed of unrelated words. Results indicated clear deficits due to global similarity among item vectors, but only a minimal impact of global similarity among the order vectors. We also found evidence for a linear relationship between global similarity and drift rate and did not find any evidence that global similarity differentially affected performance in speed vs. accuracy emphasis conditions. In addition, we found that global semantic similarity could only partially account for the word frequency effect, suggesting that other factors besides semantic similarity may be responsible.

Inter-character Orthographic Similarity Effects on the Recognition of Chinese Coordinative Compound Words.

Chinese coordinative compound words are common and unique in inter-character semantic and orthographic relationships. This study explored the inter-character orthographic similarity effects on the recognition of transparent two-morpheme coordinative compound words. Seventy-two native Chinese readers participated in a lexical decision task. The findings demonstrated robust inhibitory inter-character orthographic similarity effects, intra-word character reversal effects, and inter-character semantic similarity effects. These results were compared to those of previous studies on coordinative compound word recognition and on the orthographic similarity phenomenon at both character and word levels. The findings were explained with the multi-level representational model of morphological processing of Chinese compound words (Zhou and Marslen-Wilson in Psychologia 43:47-66, 2000). The model was further extended by adding the activation of morpho-orthographic relationships and the mapping of morphemic orthographic information onto the semantic information of both morphemes and whole words.