Center-embedded Clauses Research Articles

Beta and gamma binaural beats enhance auditory sentence comprehension.

Binaural beats-an auditory illusion produced when two pure tones of slightly different frequencies are dichotically presented-have been shown to modulate various cognitive and psychological states. Here, we investigated the effects of binaural beat stimulation on auditory sentence processing that required interpretation of syntactic relations (Experiment 1) or an evaluation of syntactic well formedness (Experiment 2) with a large cohort of healthy young adults (N = 200). In both experiments, participants performed a language task after listening to one of four sounds (i.e., between-subject design): theta (7Hz), beta (18Hz), and gamma (40Hz) binaural beats embedded in music, or the music only (baseline). In Experiment 1, 100 participants indicated the gender of a noun linked to a transitive action verb in spoken sentences containing either a subject or object-relative center-embedded clause. We found that both beta and gamma binaural beats yielded better performance, compared to the baseline, especially for syntactically more complex object-relative sentences. To determine if the binaural beat effect can be generalized to another type of syntactic analysis, we conducted Experiment 2 in which another 100 participants indicated whether or not there was a grammatical error in spoken sentences. However, none of the binaural beats yielded better performance for this task indicating that the benefit of beta and gamma binaural beats may be specific to the interpretation of syntactic relations. Together, we demonstrate, for the first time, the positive impact of binaural beats on auditory language comprehension. Both theoretical and practical implications are discussed.

Poor writing, not specialized concepts, drives processing difficulty in legal language

Despite their ever-increasing presence in everyday life, contracts remain notoriously inaccessible to laypeople. Why? Here, a corpus analysis (n ≈10 million words) revealed that contracts contain startlingly high proportions of certain difficult-to-process features–including low-frequency jargon, center-embedded clauses (leading to long-distance syntactic dependencies), passive voice structures, and non-standard capitalization–relative to nine other baseline genres of written and spoken English. Two experiments (N=184) further revealed that excerpts containing these features were recalled and comprehended at lower rates than excerpts without these features, even for experienced readers, and that center-embedded clauses inhibited recall more-so than other features. These findings (a) undermine the specialized concepts account of legal theory, according to which law is a system built upon expert knowledge of technical concepts; (b) suggest such processing difficulties result largely from working-memory limitations imposed by long-distance syntactic dependencies (i.e., poor writing) as opposed to a mere lack of specialized legal knowledge; and (c) suggest editing out problematic features of legal texts would be tractable and beneficial for society at-large.

Hemispheric dissociations in regions supporting auditory sentence comprehension in older adults.

We investigated how the aging brain copes with acoustic and syntactic challenges during spoken language comprehension. Thirty-eight healthy adults aged 54 - 80 years (M = 66 years) participated in an fMRI experiment wherein listeners indicated the gender of an agent in short spoken sentences that varied in syntactic complexity (object-relative vs subject-relative center-embedded clause structures) and acoustic richness (high vs low spectral detail, but all intelligible). We found widespread activity throughout a bilateral frontotemporal network during successful sentence comprehension. Consistent with prior reports, bilateral inferior frontal gyrus and left posterior superior temporal gyrus were more active in response to object-relative sentences than to subject-relative sentences. Moreover, several regions were significantly correlated with individual differences in task performance: Activity in right frontoparietal cortex and left cerebellum (Crus I & II) showed a negative correlation with overall comprehension. By contrast, left frontotemporal areas and right cerebellum (Lobule VII) showed a negative correlation with accuracy specifically for syntactically complex sentences. In addition, laterality analyses confirmed a lack of hemispheric lateralization in activity evoked by sentence stimuli in older adults. Importantly, we found different hemispheric roles, with a left-lateralized core language network supporting syntactic operations, and right-hemisphere regions coming into play to aid in general cognitive demands during spoken sentence processing. Together our findings support the view that high levels of language comprehension in older adults are maintained by a close interplay between a core left hemisphere language network and additional neural resources in the contralateral hemisphere.

The missing-VP effect in readers of English as a second language

English sentences with double center-embedded clauses are read faster when they are made ungrammatical by removing one of the required verb phrases. This phenomenon is known as the missing-VP effect. German and Dutch speakers do not experience the missing-VP effect when reading their native language, but they do when reading English as a second language (L2). We investigate whether the missing-VP effect when reading L2 English occurs in native Dutch speakers because their knowledge of English is similar to that of native English speakers (the high exposure account), or because of the difficulty of L2 reading (the low proficiency account). In an eye-tracking study, we compare the size of the missing-VP effect between native Dutch and native English participants, and across native Dutch participants with varying L2 English proficiency and exposure. Results provide evidence for both accounts, suggesting that both native-like knowledge of English and L2 reading difficulty play a role.

What Limits Our Capacity to Process Nested Long-Range Dependencies in Sentence Comprehension?

Sentence comprehension requires inferring, from a sequence of words, the structure of syntactic relationships that bind these words into a semantic representation. Our limited ability to build some specific syntactic structures, such as nested center-embedded clauses (e.g., “The dog that the cat that the mouse bit chased ran away”), suggests a striking capacity limitation of sentence processing, and thus offers a window to understand how the human brain processes sentences. Here, we review the main hypotheses proposed in psycholinguistics to explain such capacity limitation. We then introduce an alternative approach, derived from our recent work on artificial neural networks optimized for language modeling, and predict that capacity limitation derives from the emergence of sparse and feature-specific syntactic units. Unlike psycholinguistic theories, our neural network-based framework provides precise capacity-limit predictions without making any a priori assumptions about the form of the grammar or parser. Finally, we discuss how our framework may clarify the mechanistic underpinning of language processing and its limitations in the human brain.

Differences in Hearing Acuity among "Normal-Hearing" Young Adults Modulate the Neural Basis for Speech Comprehension.

In this paper, we investigate how subtle differences in hearing acuity affect the neural systems supporting speech processing in young adults. Auditory sentence comprehension requires perceiving a complex acoustic signal and performing linguistic operations to extract the correct meaning. We used functional MRI to monitor human brain activity while adults aged 18–41 years listened to spoken sentences. The sentences varied in their level of syntactic processing demands, containing either a subject-relative or object-relative center-embedded clause. All participants self-reported normal hearing, confirmed by audiometric testing, with some variation within a clinically normal range. We found that participants showed activity related to sentence processing in a left-lateralized frontotemporal network. Although accuracy was generally high, participants still made some errors, which were associated with increased activity in bilateral cingulo-opercular and frontoparietal attention networks. A whole-brain regression analysis revealed that activity in a right anterior middle frontal gyrus (aMFG) component of the frontoparietal attention network was related to individual differences in hearing acuity, such that listeners with poorer hearing showed greater recruitment of this region when successfully understanding a sentence. The activity in right aMFGs for listeners with poor hearing did not differ as a function of sentence type, suggesting a general mechanism that is independent of linguistic processing demands. Our results suggest that even modest variations in hearing ability impact the systems supporting auditory speech comprehension, and that auditory sentence comprehension entails the coordination of a left perisylvian network that is sensitive to linguistic variation with an executive attention network that responds to acoustic challenge.

English Speakers’ Comprehension of Embedded Relative Clauses in L2 Japanese

This study investigated native English speakers’ comprehension of Japanese sentences in which relative clauses are embedded. Specifically, this study contrasted between (a) short-before-long sentences with center-embedded relative clauses and (b) long-before-short sentences with non-center-embedded relative clauses. Sentence-type (a) indicates a sentence that includes a short phrase before a long phrase and includes a relative clause that is embedded in the middle of the sentence, e.g., Onna-ga Ken-ga kiratteiru giin-o hometa ‘The woman praised the senator who Ken hated’. Sentence-type (b) indicates a sentence with a long phrase before a short phrase and includes a relative clause that is embedded peripherally, e.g., Ken-ga kiratteiru onna-ga giin-o hometa ‘The woman who Ken hated praised the senator’. Experiment 1 revealed that native English speakers, who are learners of Japanese, comprehended the type (b) sentences with long-before-short phrases and with non-center-embedded relative clauses more accurately than the type (a) sentences with short-before-long phrases with center-embedded relative clauses. The results indicate that the preference for the non-center-embedded clauses to center-embedded clauses is universal across languages, while the preference for short-before-long phrases is language-specific. However, Experiment 2 indicated that the different accuracy rates in comprehensions of (a) and (b) disappeared when the matrix subjects are marked by the topic-morpheme wa. The outcome indicated that the topic phrases are immediately interpreted as a part of main clauses.

Acoustic richness modulates the neural networks supporting intelligible speech processing

The information contained in a sensory signal plays a critical role in determining what neural processes are engaged. Here we used interleaved silent steady-state (ISSS) functional magnetic resonance imaging (fMRI) to explore how human listeners cope with different degrees of acoustic richness during auditory sentence comprehension. Twenty-six healthy young adults underwent scanning while hearing sentences that varied in acoustic richness (high vs. low spectral detail) and syntactic complexity (subject-relative vs. object-relative center-embedded clause structures). We manipulated acoustic richness by presenting the stimuli as unprocessed full-spectrum speech, or noise-vocoded with 24 channels. Importantly, although the vocoded sentences were spectrally impoverished, all sentences were highly intelligible. These manipulations allowed us to test how intelligible speech processing was affected by orthogonal linguistic and acoustic demands. Acoustically rich speech showed stronger activation than acoustically less-detailed speech in a bilateral temporoparietal network with more pronounced activity in the right hemisphere. By contrast, listening to sentences with greater syntactic complexity resulted in increased activation of a left-lateralized network including left posterior lateral temporal cortex, left inferior frontal gyrus, and left dorsolateral prefrontal cortex. Significant interactions between acoustic richness and syntactic complexity occurred in left supramarginal gyrus, right superior temporal gyrus, and right inferior frontal gyrus, indicating that the regions recruited for syntactic challenge differed as a function of acoustic properties of the speech. Our findings suggest that the neural systems involved in speech perception are finely tuned to the type of information available, and that reducing the richness of the acoustic signal dramatically alters the brain's response to spoken language, even when intelligibility is high.

An interference account of the missing-VP effect.

Sentences with doubly center-embedded relative clauses in which a verb phrase (VP) is missing are sometimes perceived as grammatical, thus giving rise to an illusion of grammaticality. In this paper, we provide a new account of why missing-VP sentences, which are both complex and ungrammatical, lead to an illusion of grammaticality, the so-called missing-VP effect. We propose that the missing-VP effect in particular, and processing difficulties with multiply center-embedded clauses more generally, are best understood as resulting from interference during cue-based retrieval. When processing a sentence with double center-embedding, a retrieval error due to interference can cause the verb of an embedded clause to be erroneously attached into a higher clause. This can lead to an illusion of grammaticality in the case of missing-VP sentences and to processing complexity in the case of complete sentences with double center-embedding. Evidence for an interference account of the missing-VP effect comes from experiments that have investigated the missing-VP effect in German using a speeded grammaticality judgments procedure. We review this evidence and then present two new experiments that show that the missing-VP effect can be found in German also with less restricting procedures. One experiment was a questionnaire study which required grammaticality judgments from participants without imposing any time constraints. The second experiment used a self-paced reading procedure and did not require any judgments. Both experiments confirm the prior findings of missing-VP effects in German and also show that the missing-VP effect is subject to a primacy effect as known from the memory literature. Based on this evidence, we argue that an account of missing-VP effects in terms of interference during cue-based retrieval is superior to accounts in terms of limited memory resources or in terms of experience with embedded structures.

Strategies for Processing Structural Ambiguity in English-Korean Translation: Focusing on Center-embedded Clauses

Strategies for Processing Structural Ambiguity in English-Korean Translation: Focusing on Center-embedded Clauses

Sentence processing in Lewy body spectrum disorder: The role of working memory

Prior work has related sentence processing to executive deficits in non-demented patients with Parkinson’s disease (PD). We extended this investigation to patients with dementia with Lewy bodies (DLB) and PD dementia (PDD) by examining grammatical and working memory components of sentence processing in the full range of patients with Lewy body spectrum disorder (LBSD). Thirty-three patients with LBSD were given a two-alternative, forced-choice sentence-picture matching task. Sentence type, working memory, and grammatical structure were systematically manipulated in the sentences. We found that patients with PDD and DLB were significantly impaired relative to non-demented PD patients and healthy controls. The deficit in PDD/DLB was most pronounced for sentences lengthened by the strategic placement of an additional prepositional phrase and for sentences with an additional proposition due to a center-embedded clause. However, there was no effect for subject-relative versus object-relative grammatical structure. An MRI voxel-based morphometry analysis in a subset of patients showed significant gray matter thinning in the frontal lobe bilaterally, and this extended to temporal, parietal and occipital regions. A regression analysis related sentence processing difficulty in LBSD to frontal neocortex, including inferior prefrontal, premotor, and dorsolateral prefrontal regions, as well as right superior temporal cortex. These findings are consistent with the hypothesis that patients with PDD and DLB have difficulty processing sentences with increased working memory demands and that this deficit is related in part to their frontal disease.

Hierarchical prosodic structures in the intonation of center-embedded relative clauses

There is no doubt that recursion exists in syntax, but whether this is reflected in prosody is still an open question. In this paper, the prosody of sentences with syntactically recursive center-embedded clauses is examined empirically in two languages. On the basis of the results, we argue for recursion of higher prosodic domains in German. Evidence for this proposal comes from the F0 range of the matrix clause, which is larger if embedded material is present. It is both higher in the high regions and lower in the low regions than in matrix clauses without embedded material. A second piece of evidence relates to pitch scaling of high tones at the boundaries between clauses, which is sensitive to syntactic embedding. In Hindi, by contrast, no evidence for embedding of prosodic phrases could be found. Prosodic phrases are concatenated and are in a downstep relationship to each other, regardless of the syntactic structure of the sentence. The difference between the prosodic structure of German and Hindi is tentatively explained by their different intonational properties: German is an intonation language with lexical stress and Hindi is a phrase language without lexical stress, in which tonal structures are phrase-based.

Dissociable patterns of brain activity during comprehension of rapid and syntactically complex speech: Evidence from fMRI

Sentence comprehension is a complex task that involves both language-specific processing components and general cognitive resources. Comprehension can be made more difficult by increasing the syntactic complexity or the presentation rate of a sentence, but it is unclear whether the same neural mechanism underlies both of these effects. In the current study, we used event-related functional magnetic resonance imaging (fMRI) to monitor neural activity while participants heard sentences containing a subject-relative or object-relative center-embedded clause presented at three different speech rates. Syntactically complex object-relative sentences activated left inferior frontal cortex across presentation rates, whereas sentences presented at a rapid rate recruited frontal brain regions such as anterior cingulate and premotor cortex, regardless of syntactic complexity. These results suggest that dissociable components of a large-scale neural network support the processing of syntactic complexity and speech presented at a rapid rate during auditory sentence processing.

Attentional resource and processing speed limitations during sentence processing in Parkinson’s disease

Several studies have suggested that patients with Parkinson’s disease (PD) have sentence comprehension difficulty in part because of their limited executive resources. However, these assessments confound the executive resources contributing to sentence comprehension with the resources needed for task performance. In the present study, we used a word detection technique that minimizes task demands in order to evaluate attentional and processing speed resources during the comprehension of simple sentences without subordinate clauses and sentences containing subject-relative and object-relative center-embedded subordinate clauses. We found that PD patients have poor sensitivity to phonetic errors embedded in unbound grammatical morphemes, regardless of the clausal structure of the sentence, suggesting difficulty attending to grammatical morphemes. We also found that PD patients are significantly slowed in their sensitivity to phonetic errors in content words embedded in object-relative center-embedded sentences. Slowed sensitivity to content words in object-relative sentences was correlated with timed executive measures of planning. On a traditional measure of comprehension, these PD patients were impaired for sentences containing object-relative center-embedded clauses compared to sentences with subject-relative center-embedded clauses, and comprehension of object-relative sentences was correlated with executive measures. Our findings are consistent with the claim that limited executive resources for strategic attention and processing speed contribute to the sentence comprehension difficulties of PD patients.

Neural basis for sentence comprehension deficits in frontotemporal dementia

Many patients with frontotemporal dementia (FTD) have impaired sentence comprehension. However, the pattern of comprehension difficulty appears to vary depending on the clinical subgroup. The purpose of this study was to elucidate the neural basis for these deficits in FTD. We studied patients with two different presentations: Three patients with Progressive Non-Fluent Ahasia (PNFA), and five non-aphasic patients with a dysexecutive and social impairment (EXEC). The FTD patient subgroups were compared to a cohort of 11 healthy seniors with intact sentence comprehension. We monitored regional cerebral activity with blood oxygen level dependent (BOLD) functional magnetic resonance imaging (fMRI) while subjects read sentences featuring both a grammatically complex object-relative center-embedded clause and a long linkage between the head noun phrase (NP) and the gap where the NP is interpreted in the center-embedded clause. Subjects decided whether the agent of the action is a male or a female. Healthy seniors activated both ventral portions of inferior frontal cortex (vIFC) and dorsal portions of IFC (dIFC) in the left hemisphere, often associated with grammatical and working memory components of these sentences, respectively. PNFA patients differed from healthy controls since they have reduced activation of left vIFC, while EXEC patients have less recruitment of left dIFC. We conclude that FTD subgroups have distinct patterns of sentence comprehension difficulty in part because of selective interruptions of a large-scale neural network for sentence processing.

Neural basis for sentence comprehension: grammatical and short-term memory components.

We monitored regional cerebral activity with BOLD fMRI while subjects were presented written sentences differing in their grammatical structure (subject-relative or object-relative center-embedded clauses) and their short-term memory demands (short or long antecedent-gap linkages). A core region of left posterior superior temporal cortex was recruited during all sentence conditions in comparison to a pseudofont baseline, suggesting that this area plays a central role in sustaining comprehension that is common to all sentences. Right posterior superior temporal cortex was recruited during sentences with long compared to short antecedent-gap linkages regardless of grammatical structure, suggesting that this brain region supports passive short-term memory during sentence comprehension. Recruitment of left inferior frontal cortex was most clearly associated with sentences that featured both an object-relative clause and a long antecedent-gap linkage, suggesting that this region supports the cognitive resources required to maintain long-distance syntactic dependencies during the comprehension of grammatically complex sentences.

Cognitive Resource Limitations during Sentence Comprehension in Parkinson's Disease

Patients with idiopathic Parkinson's disease (PD) were asked to identify the agent of the action in orally presented sentences with subject-relative or object-relative center-embedded clauses while simultaneously performing a secondary task that was less resource-demanding (finger tapping) or more resource-demanding (recognition span). We found that a subgroup of PD patients with impaired sentence comprehension at baseline (no secondary task) did not differ from random in their accuracy understanding all types of sentences during the more demanding (recognition span) condition and also had difficulty understanding the most complex sentences during the less demanding (finger tapping) condition. Control subjects and PD patients without baseline sentence comprehension difficulty were random only in their comprehension of the most complex sentences under the more demanding (recognition span) secondary task condition. Examination of response latencies for accurately understood sentences revealed only an effect for the type of sentence, and this was equally evident across all groups of subjects and regardless of the condition under which the sentences were administered. The sensitivity of PD patients' sentence comprehension accuracy to secondary task resource demands is most consistent with the hypothesis that limited cognitive resources contribute to sentence comprehension difficulty in PD.

Parsing Complex Sentences with Structured Connectionist Networks

A modular, recurrent connectionist network is taught to incrementally parse complex sentences. From input presented one word at a time, the network learns to do semantic role assignment, noun phrase attachment, and clause structure recognition, for sentences with both active and passive constructions and center-embedded clauses. The network makes syntactic and semantic predictions at every step. Previous predictions are revised as expectations are confirmed or violated with the arrival of new information. The network induces its own "grammar rules" for dynamically transforming an input sequence of words into a syntactic/semantic interpretation. The network generalizes well and is tolerant of ill-formed inputs.