Online Sentence Comprehension Research Articles

On the influence of discourse connectives on the predictions of humans and language models

Psycholinguistic literature has consistently shown that humans rely on a rich and organized understanding of event knowledge to predict the forthcoming linguistic input during online sentence comprehension. We, the authors, expect sentences to maintain coherence with the preceding context, making congruent sentence sequences easier to process than incongruent ones. It is widely known that discourse relations between sentences (e.g., temporal, contingency, comparison) are generally made explicit through specific particles, known as discourse connectives, (e.g., and, but, because, after). However, some relations that are easily accessible to the speakers, given their event knowledge, can also be left implicit. The goal of this paper is to investigate the importance of discourse connectives in the prediction of events in human language processing and pretrained language models, with a specific focus on concessives and contrastives, which signal to comprehenders that their event-related predictions have to be reversed. Inspired by previous work, we built a comprehensive set of story stimuli in Italian and Mandarin Chinese that differ in the plausibility and coherence of the situation being described and the presence or absence of a discourse connective. We collected plausibility judgments and reading times from native speakers for the stimuli. Moreover, we correlated the results of the experiments with the predictions given by computational modeling, using Surprisal scores obtained via Transformer-based language models. The human judgements were collected using a seven-point Likert scale and analyzed using cumulative link mixed modeling (CLMM), while the human reading times and language model surprisal scores were analyzed using linear mixed effects regression (LMER). We found that Chinese NLMs are sensitive to plausibility and connectives, although they struggle to reproduce expectation reversal effects due to a connective changing the plausibility of a given scenario; Italian results are even less aligned with human data, with no effects of either plausibility and connectives on Surprisal.

A Psycholinguistic Perspective on Sentence Understanding for Second Language Learners

This study delves into the intricacies of second-language sentence comprehension, investigating the impact of psycholinguistic variables. An online sentence comprehension task engaged 100 proficient English speakers with intermediate Spanish proficiency, integrating eye-tracking technology. Assessing sentence complexity, word order, lexical ambiguity, and working memory, the findings highlight higher accuracy in simple declarative sentences, while non-canonical word orders led to significantly prolonged response times. Lexical ambiguity correlated with increased error rates, yet superior working memory capacities were associated with enhanced processing efficiency. The research underscores the necessity for nuanced language instruction, emphasizing metacognitive strategies to address challenges posed by sentence complexity and ambiguity. Despite limitations in sample size and controlled settings, this study lays a foundational framework for future cognitive investigations into second-language sentence comprehension.

Is cats one word or two? L2 learners’ processing of number marking in English from the viewpoints of form–meaning mapping

This study examined number marking comprehension among Japanese learners of second language (L2) English, whose first language (L1) does not have an obligatory number marking system. The study conducted an online sentence comprehension experiment with 96 L1-Japanese learners and 32 native speakers of English, wherein participants engaged in a self-paced reading with Stroop-like number judgment tasks. Participants were required to determine the number of single words in stimuli (e.g. cat/ cats, one word; the cats/ the cat, two-word sets), and their judgment time was measured for singular and plural words. The results indicated that both groups took more time to judge single plural nouns, suggesting that Japanese L2 learners of English automatically activate plurality in online sentence comprehension as native speakers do. In contrast, neither group showed an interference effect of singularity in judging singular two-word noun sets ( the cat), unless the singularity is explicitly marked by indefinite article ( a cat). The lack of interference may be because of unmarkedness of singularity.

Lexically Independent Structural Priming in Second Language Online Sentence Comprehension

AbstractThis study examined structural priming during online second language (L2) comprehension. In two self‐paced reading experiments, 64 intermediate to advanced Chinese learners of English as a foreign language read coordinated noun phrases where the conjuncts had either the same structure or different structures. Experiment 1 showed that the second conjunct was read faster when it had the same structure as the first. This effect occurred for the structurally marked adjective phrases (e.g., a simple to grasp problem) but only showed a numerical trend for the less marked relative clauses (e.g., a problem that was simple to grasp). Experiment 2 compared unmarked adjective phrases and relative clauses (e.g., a simple problem vs. a problem that was simple) and found significant priming for both. Together, the two experiments showed that L2 comprehension priming could occur without repetition of the lexical head. Moreover, this priming was susceptible to inverse frequency effects, with the less frequent structure exhibiting greater priming.

Syntactic and semantic interference in sentence comprehension: Support from English and German eye-tracking data

A long-standing debate in the sentence processing literature concerns the time course of syntactic and semantic information processing in online sentence comprehension. The default assumption in cue-based models of parsing is that syntactic and semantic retrieval cues simultaneously guide dependency resolution. When retrieval cues match multiple items in memory, this leads to similarity-based interference. Both semantic and syntactic interference have been shown to occur in English. However, the relative timing of syntactic vs. semantic interference remains unclear. In this cross-linguistic investigation of the time course of syntactic vs. semantic interference, the data from two eye-tracking during reading experiments (English and German) suggest that the two types of interference can in principle arise simultaneously during retrieval. However, the data also indicate that semantic cues are evaluated with a small timing lag in German compared to English. This cross-linguistic difference between English and German may be due to German having richer morphosyntactic marking than English, resulting in syntactic cues dominating over semantic cues during dependency resolution. More broadly, our cross-linguistic results pose a challenge for the cue-based retrieval model’s default assumption that syntactic and semantic cues are used simultaneously during long-distance dependency formation. Our work also highlights the importance of collecting cross-linguistic data on psycholinguistic phenomena which can potentially advance theory development.

The acquisition of contrastive focus during online sentence-comprehension by children learning mandarin Chinese.

Contrastive focus, conveyed by prosodic cues, marks important information. Studies have shown that 6-year-olds learning English and Japanese can use contrastive focus during online sentence comprehension: focus used in a contrastive context facilitates the identification of a target referent (speeding up processing), whereas focus used inappropriately in a noncontrastive context misleads listeners to predict an incorrect referent, hindering the identification process (Ito et al., 2012, 2014). In Mandarin Chinese, the mapping between prosodic form and contrastive focus is less transparent, potentially delaying the acquisition of contrastive focus. This study assessed the online processing of contrastive focus by 196 Mandarin-speaking 4-10-year-olds and 34 adults in China, using the visual world paradigm. Stimuli contained a target NP in a mini discourse, with focus being used in contrastive (Experiment 1) versus Noncontrastive contexts (Experiment 2). Experiment 1 showed that the appropriate use of prosodic form for contrastive focus facilitated the identification of a target referent for 7-10-year-olds and adults, though not younger children. Experiment 2 showed that the inappropriate use of prosodic form for contrastive focus slowed the identification process only for 10-year-olds and adults. Thus, whereas 7-10-year-olds are sensitive to prosodic form for contrastive focus, only 10-year-olds use it as a primary cue to predict an upcoming referent like adults. The acquisition of contrastive focus in Mandarin is therefore a gradual process, with children showing sensitivity to contrastive focus during the early school years, and developing adult-like form-function mapping between prosody and focus until the end of primary school. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Effects of world knowledge on the prediction of upcoming verbs: an eye-tracking study

Some theories of sentence processing make a distinction between two kinds of meaning: a linguistic meaning encoded at the lexicon (i.e., selectional restrictions), and an extralinguistic knowledge derived from our everyday experiences (i.e., world knowledge). According to such theories, the former meaning is privileged over the latter in terms of the time-course of its access and influence during on-line language comprehension. The present study aims to examine whether world knowledge anomalies (that do not violate selectional restrictions) are rapidly detected during online sentence processing. In an eye-tracking experiment, we used materials in which the likelihood of a specific verb (entrevistar or secuestrar, the Spanish translations for to interview and to kidnap) depended on the agent of the event (periodista or terrorista, the Spanish translations for journalist and terrorist). The results showed an effect of typicality in regression path duration and total reading times at both the verb region and the spillover region, thus providing evidence that world knowledge is rapidly accessed and used during on-line sentence comprehension.

Towards a processing model for argument-verb computations in online sentence comprehension

The current study investigated the processing stages by which the parser incorporates different pieces of information, from clausehood to argument roles, to update predictions about the main verb. Using Mandarin to match word position across relevant conditions, we extend classic ERP findings on the impact of argument role reversals ([The millionaireSUBJECT the servantOBJECTfired] vs. #[The servantSUBJECT the millionaireOBJECTfired]), by investigating cases where one of the nouns is not an argument of the verb ([The millionaireSUBJECT the servantOBJECTfired] vs. #[The millionaire thought [the servantSUBJECTfired…]]). The pattern of N400 responses suggest a three-stage model of argument-verb computation: An initial stage demonstrates sensitivity at the verb to semantic association only. Soon after, responses show partial structure-sensitivity, differentiating whether the noun phrases are arguments of the upcoming verb or not. Only at the last stage do the arguments’ roles (e.g. agent/patient) become available to impact computations at the verb.

Semantic integration of multidimensional perceptual information in L1 sentence comprehension

AbstractMany studies have substantiated the perceptual symbol system, which assumes a routine generation of perceptual information during language comprehension, but little is known about the processing format in which the perceptual information of different dimensions is conveyed simultaneously during sentence comprehension. The current study provides the first experimental evidence of how multidimensional perceptual information (color and shape) was processed during online sentence comprehension in Mandarin. We designed three consecutive sentence–picture verification tasks that only differed in the delay of the display of pictures preceded by declarative sentences. The processing was analyzed in three stages based on time intervals (i.e., 0ms, +750ms, +1500ms). The response accuracy and response time data were reported. The initial stage (i.e., ISI=0ms) attested the match effect of color and shape, but the simulated representation of color and shape did not interact. In the intermediate stage (i.e., ISI=750ms), the routinely simulated color and shape interacted, but the match facilitation was found only in cases where one perceptual information was in mismatch while the other was not. In the final stage (i.e., ISI=1500ms), the match facilitation of one particular perceptual property was influenced by a mismatch with the other perceptual property. These results suggested that multiple perceptual information presented simultaneously was processed in an additive manner to a large extent before entering into the final stage, where the simulated perceptual information was integrated in a multiplicative manner. The results also suggested that color and shape were comparable to object recognition when conjointly conveyed. In relation to other evidence from behavioral and event-related potential studies on sentence reading in the discussion, we subscribed to the idea that the full semantic integration became available over time.

Access to verb bias and plausibility information during syntactic processing in adult Spanish-English bilinguals.

In two experiments, we examine how proficient second language speakers integrate verb bias and plausibility information during online sentence comprehension. Spanish-English speakers and native English speakers read sentences in English in which a post-verbal noun phrase (NP) could be interpreted as a direct object or a sentential subject. To examine the role of verb bias, the post-verbal NP was preceded by a verb that is preferentially followed by a direct object (DO-bias verbs) or a sentential complement (SC-bias verbs). To assess the role of plausibility, the semantic fit between the verb and the post-verbal NP was either congruent or incongruent with the direct object interpretation. The results show that both second language speakers and native speakers used verb bias information to assign a grammatical role to the post-verbal ambiguous NP with small differences. Syntactic revision of an initially incorrect DO interpretation was facilitated by the presence of an implausible NP.

Cognitive Predictors of Child Second Language Comprehension and Syntactic Learning

AbstractThis study examined the role of child cognitive abilities for procedural and declarative learning in the earliest stages of second language (L2) exposure. In the context of a computer game, 53 first language Italian monolingual children were aurally trained in a novel miniature language over 3 consecutive days. A mixed effects model analysis of the relationship between cognitive predictors and outcomes in morphosyntax measured via a grammaticality judgment test (GJT) was performed. Relative to adults trained in the same paradigm, children with higher procedural learning ability (measured via an alternate serial reaction time task) showed significantly better learning of word order, although the effect size was small. Modeling accuracy in online sentence comprehension during the game also evidenced that higher procedural learning ability was positively associated with significantly better outcomes as practice progressed. By contrast, a composite measure of verbal and visual declarative learning ability did not predict L2 outcomes in either the GJT or the online measure.

Subject Gaps Revisited: Complement Clauses and Complementizer-Trace Effects.

This study investigates how filler-gap dependencies associated with subject position are formed in online sentence comprehension. Since Crain and Fodor (1985), “filled-gap” studies have provided evidence that the parser actively seeks to associate a wh-filler with a gap in direct object position of a sentence wherever possible; the evidence that this same process applies for subject position, is, however, more limited (Stowe, 1986; Lee, 2004). We examine the processing of complement clauses, finding that wh dependency formation is actively attempted at embedded subject position (e.g., Kate in Who did Lucy think Kate could drive us home to?), unless, however, the embedded clause contains a complementizer (e.g., Who did Lucy think that Kate … .?). The absence of the dependency formation in the latter case demonstrates that the complementizer-trace effect (cf., ∗Who did Lucy think that could drive us home to mom?; Perlmutter, 1968) is, like syntactic island constraints (Ross, 1967; Keshev and Meltzer-Asscher, 2017), immediately operative in online structure building.

Türkçede Bürünsel Odak İşlemleme: Elektrofizyolojik Bir Araştırma

Focused elements are generally marked with syntactic canonicity and prosody. Being a scrambled language, Turkish uses both syntactic and prosodic information to mark the focus. However, it does not allow for focus marking in post-verbal position. In this study, the neurophysiological processes of the focus in Turkish are examined by using prosodic and syntactic information. Recent psycholinguistics studies assume that there is an interaction between prosody and syntax through the focus in the online sentence comprehension process. Thirty participants (16 female and 14 male between the ages of 19 and 33), whose native language was Turkish and who spoke monolingual Turkish, and who did not have any neurological, hearing, or linguistic impairments, took part in the experiments measured with Electroencephalogram (EEG). Using an event-related potentials (ERPs) design, this study provides evidence for an interaction between prosody and syntax in Turkish. The experimental design of the study consisted of prosodic, syntactic, and prosodic-syntactic violations. Participants were asked to listen 300 auditory stimuli (100 filler sentences) including sentences with both congruent and incongruent focus. The stimuli consisted of 50 sentences for each experimental condition. All critical words occurred in the sentence-final positions. For the prosodic violation critical words were focused via incongruent focusing on post-verbal position, and for the syntactic violation critical words were manipulated with case marking manipulation (i.e., accusative case versus dative case violations). In addition, for the interaction of prosodic and syntactic violations, critical words were incongruent focused and incongruent case was marked. The results revealed that prosodic incongruity elicited a broadly distributed positivity in posterior regions (400-1200 ms) lateralized to the left hemisphere and a right anterior negativity (RAN) (300-500 ms) effect. Syntactic violations also indicated a distributed anterior negativity (300-500 ms) effect. Supportive evidence for the late interaction of prosodic and syntactic processing in the neural integration of positive 600 (P600) and Closure Positive Shift (CPS) was observed. The findings provide support for recent neurocognitive approaches for late interaction between prosody and syntax in the sentence-final position in Turkish sentences.

Referential context and executive functioning influence children's resolution of syntactic ambiguity.

Classic studies reveal two striking differences between preschoolers and adults in online sentence comprehension. Adults (a) recruit referential context cues to guide syntactic parsing, interpreting an ambiguous phrase as a modifier if a modifier is needed to single out the intended referent among multiple options, and (b) use late-arriving information to recover from misinterpretation. Five-year-olds fail on both counts, appearing insensitive to the referential context and often failing to recover from parsing errors (Trueswell, Sekerina, Hill, & Logrip, 1999). But other findings suggest that 5-year-olds show delayed rather than absent sensitivity to the referential context, and that individual differences in executive functioning predict children's ability to recover from garden-path errors. In 2 experiments, we built on these findings, focusing on whether children recruit referential-context cues if given time to do so. Children heard temporarily ambiguous instructions (e.g., Put the frog on the pond into the tent), while we monitored their eye-gaze and actions. We used a slow speech rate, and manipulated referential context between rather than within subjects, to give children time to bring referential context cues into play. Across experiments, eye-movement and action analyses revealed emerging sensitivity to the referential context. Moreover, error rates and eye-movement patterns indicating failures to revise were predicted by individual differences in executive function (scores in Simon Says and Flanker tasks). These data suggest that children, like adults, use referential context information in syntactic processing under some circumstances; the findings are also consistent with a role for domain-general executive function in resolution of syntactic ambiguity. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

An Online Tool to Assess Sentence Comprehension in Teenagers at Risk for School Exclusion: Evidence From L2 Italian Students.

This study presents a web-based sentence comprehension test aimed at identifying high school students who are at risk for a language delay. By assessing linguistic skills on a sample of high school students with Italian as an L2 and their monolingual peers, attending a vocational school, we were able to identify a subgroup of L2 students with consistent difficulties in sentence comprehension, though their reading skills were within the average range. The same subgroup revealed to experience a lack of support within the school context, suggesting that poor L2 skills might be a critical variable to consider in order to identify students at risk for school exclusion. Regarding the test, accuracy to the on-line sentence comprehension task was significantly predicted by reading abilities and vocabulary skills, thus indicating that this test might represent a rapid but efficient way to assess linguistic abilities at school. We recommend that establishing a valid and practical procedure for the evaluation of linguistic skills in bilingual students who struggle with their L2 is the first step toward promoting social inclusion in the multilingual classroom, in order to increase their ability to actively participate in school and social activities.

Ahead of time: Early sentence slow cortical modulations associated to semantic prediction

According to prediction-based accounts of language comprehension, incoming contextual information is constantly used to guide the pre-activation of the most probable continuations to the unfolding sentences. However, there is still scarce evidence of the build-up of these predictions during sentence comprehension. Using event-related brain potentials, we investigated sustained processes associated to semantic prediction during online sentence comprehension. To address this, participants read sentences with varying levels of contextual constraint one word at a time. A 1000 ms interval preceded the final word, which could be congruent or incongruent. A slow sustained negativity developed gradually over the course of sentences, showing differences across conditions, with increasingly larger amplitudes for high than low levels of constraint. The effect was maximal in the interval preceding the closing word. This interval elicited a left-dominant slow negative potential with a graded amplitude modulation to contextual constraint, replicating previous results in speech comprehension. We argue that these slow potentials index the engagement of cognitive operations associated to semantic prediction. In addition, we replicated the finding of an earlier onset of the N400 effect (incongruent minus congruent) for high relative to low contextual constraint, suggesting facilitated processing for contextually-supported and highly expected words. Altogether, these results are consistent with prediction-based models of language comprehension and they also strengthen the value of investigating slow components as potential indices of mechanisms linked to language prediction.

Japanese EFL learners’ sentence processing of conceptual plurality: An analysis focusing on reciprocal verbs

ABSTRACTThis study aimed to investigate how Japanese learners of English as a foreign language, whose first language does not have obligatory morphological number marking, process conceptual plurality. The targeted structure was reciprocal verbs, which require conceptual plurality to interpret their meanings correctly. The results of a sentence completion task confirmed that participants could use reciprocal verbs reciprocally in English. In a self-paced reading experiment, participants read sentences with reciprocal verbs and those with optionally transitive verbs (e.g., while the king and the queen kissed/left the baby read the book in the bed). There was no reading time delay for reciprocal verbs but a delay for optionally transitive verbs. Therefore, the participants succeeded in processing second language conceptual plurality in the online sentence comprehension task.

Morphological variability in second language learners: An examination of electrophysiological and production data.

We examined sources of morphological variability in second language (L2) learners of Spanish whose native language (L1) is English, with a focus on L1-L2 similarity, morphological markedness, and knowledge type (receptive vs. expressive). Experiment 1 uses event-related potentials to examine noun-adjective number (present in L1) and gender agreement (absent in L1) in online sentence comprehension (receptive knowledge). For each feature, markedness was manipulated, such that half of the critical noun-adjective combinations were feminine (marked) and the other half were masculine; half were used in the plural (marked) and the other half were used in the singular. With this setup, we examined learners' potential overreliance on unmarked forms or "defaults" (singular/masculine). Experiment 2 examines similar dependencies in spoken sentence production (expressive knowledge). Learners (n = 22) performed better with number than gender overall, but their brain responses to both features were qualitatively native-like (i.e., P600), even though gender was probed with nouns that do not provide strong distributional cues to gender. In addition, variability with gender agreement was better accounted for by lexical (as opposed to syntactic) aspects. Learners showed no advantage for comprehension over production, and no systematic evidence of reliance on morphological defaults, although their online processing was sensitive to markedness in a native-like manner. Overall, these results suggest that there is facilitation for L2 properties that exist in the L1 and that markedness impacts L2 processing, but in a native-like manner. These results also speak against proposals arguing that adult L2ers have deficits at the level of the morphology or the syntax. (PsycINFO Database Record

Lexicalized structural priming in second language online sentence comprehension

Prior work has shown intriguing differences between first language (L1) and second language (L2) comprehension priming of relative clauses. We investigated English reduced relative clause priming in Chinese adult learners of English. Participants of different education levels read sentences in a self-paced, moving window paradigm. Critical sentences had a temporarily ambiguous reduced relative clause. Across lists, critical sentences were rotated, so that they occurred either as prime or as target, and had either the same or different verb as the critical sentence with which they were paired. Prime/target pairs were separated by several filler sentences, which never contained a relative clause. Mean reading times for the disambiguating region in the target sentences were faster than in the prime sentences, but only in the same-verb condition, not in the different-verb condition. This pattern of results is consistent with L1 comprehension priming research, suggesting that similar lexically specific mechanisms are involved in L1 and L2 comprehension priming of reduced relative clauses. These findings are in line with lexicalist accounts of sentence comprehension (e.g. MacDonald et al., 1994), according to which syntactic information is bound to specific words. In addition, these findings argue against theories that postulate fundamental differences in processing of L1 and L2 (e.g. Clahsen and Felser, 2006a, 2006b).

Memory retrieval and sentence processing: Differences between native and non-native speakers

Efficient comprehension of sentences requires rapidly and continuously accessing and integrating different sources of information in real time. Psycholinguists have developed detailed models and theories to account for the processes involved in on-line sentence comprehension as well as a number of sophisticated experimental designs for studying these processes. But how about real-time sentence processing in bilinguals? The study of bilingual sentence processing has received considerable attention and has led to a remarkable growth of experimental studies over the last 10 years. The focus of these studies has been on late bilinguals, i.e., on second-language (L2) learners who learned a non-native language after early childhood, as adolescents or adults. These studies have revealed both similarities and differences between native (L1) and non-native (L2) sentence processing. Several proposals have been made to account for the experimental findings, but the significance and nature of native vs. non-native differences in sentence processing has remained controversial. Some researchers have claimed that L1 and L2 sentence processing are essentially the same and that observed performance differences between native and non-native sentence comprehension are due to peripheral factors, e.g., decoding problems, working memory limitations, slower processing speed, difficulties with lexical access and retrieval, or a reduced ability to predict during L2 processing (e.g., McDonald, 2006; Hopp, 2016; Kaan, 2014). Others have posited more substantial differences between L1 and L2 processing. One prominent proposal is Clahsen and Felser's (2006a, b) Shallow-Structure Hypothesis (SSH). Assuming multi-stream models of language processing (e.g., Ferreira & Patson, 2007) with two routes from form to meaning, a heuristic one that employs surface-form information, lexical and semantic cues, and an algorithmic route that relies on a full grammatical parse, the SSH holds that L2 processing relies less on grammatical and more on non-grammatical information sources, in comparison to L1 processing of syntactic (and morphological) phenomena.