Memory mechanisms supporting syntactic comprehension

Efforts to characterize the memory system that supports sentence comprehension have historically drawn extensively on short-term memory as a source of mechanisms that might apply to sentences. The focus of these efforts has changed significantly in the past decade. As a result of changes in models of short-term working memory (ST-WM) and developments in models of sentence comprehension, the effort to relate entire components of an ST-WM system, such as those in the model developed by Baddeley (Nature Reviews Neuroscience 4: 829–839, 2003) to sentence comprehension has largely been replaced by an effort to relate more specific mechanisms found in modern models of ST-WM to memory processes that support one aspect of sentence comprehension—the assignment of syntactic structure (parsing) and its use in determining sentence meaning (interpretation) during sentence comprehension. In this article, we present the historical background to recent studies of the memory mechanisms that support parsing and interpretation and review recent research into this relation. We argue that the results of this research do not converge on a set of mechanisms derived from ST-WM that apply to parsing and interpretation. We argue that the memory mechanisms supporting parsing and interpretation have features that characterize another memory system that has been postulated to account for skilled performance—long-term working memory. We propose a model of the relation of different aspects of parsing and interpretation to ST-WM and long-term working memory.     

句子加工和语义记忆结构的实验研究

本文报告的实验,以句子验证范式为基础,采用常规试验和信号试验相混合的程序,对汉语句子加工的有关问题作了考察,实验结果表明,句子理解的时间,依赖于人对句中所含概念之间语义关系的认知度;人对句子的理解是一个部分信息逐渐积累的过程。根据实验材料,作者对语义记忆的结构问题作了讨论。     

Lossy-Context Surprisal: An Information-Theoretic Model of Memory Effects in Sentence Processing

A key component of research on human sentence processing is to characterize the processing difficulty associated with the comprehension of words in context. Models that explain and predict this difficulty can be broadly divided into two kinds, expectation-based and memory-based. In this work, we present a new model of incremental sentence processing difficulty that unifies and extends key features of both kinds of models. Our model, lossy-context surprisal, holds that the processing difficulty at a word in context is proportional to the surprisal of the word given a lossy memory representation of the context—that is, a memory representation that does not contain complete information about previous words. We show that this model provides an intuitive explanation for an outstanding puzzle involving interactions of memory and expectations: language-dependent structural forgetting, where the effects of memory on sentence processing appear to be moderated by language statistics. Furthermore, we demonstrate that dependency locality effects, a signature prediction of memory-based theories, can be derived from lossy-context surprisal as a special case of a novel, more general principle called information locality.     

Interpretation-based processing: a unified theory of semantic sentence comprehension

We present interpretation-based processing—a theory of sentence processing that builds a syntactic and a semantic representation for a sentence and assigns an interpretation to the sentence as soon as possible. That interpretation can further participate in comprehension and in lexical processing and is vital for relating the sentence to the prior discourse. Our theory offers a unified account of the processing of literal sentences, metaphoric sentences, and sentences containing semantic illusions. It also explains how text can prime lexical access. We show that word literality is a matter of degree and that the speed and quality of comprehension depend both on how similar words are to their antecedents in the preceding text and how salient the sentence is with respect to the preceding text. Interpretation-based processing also reconciles superficially contradictory findings about the difference in processing times for metaphors and literals. The theory has been implemented in ACT-R [Anderson and Lebiere, The Atomic Components of Thought, Lawrence Erlbaum Associates Publishers, Mahwah, NJ, 1998].     

Separating comprehension from the verification process

The present experiments aimed at separating comprehension of a sentence from its verification. Presentation of a first sentence was terminated by the subject’s keypress after his comprehension. This gave a measure of sentence comprehension. A second sentence appeared immediately for a fixed interval. Then, a picture was presented, and the subject verified the first sentence against the picture. Finally, a second picture appeared, and the subject verified the second sentence against the second picture. The purpose of presenting a second sentence was to stop the processing of the first sentence after the subject’s keypress (so that comprehension did not continue beyond what was measured) and also to insure the subject’s full comprehension (comprehending the sentence to form a unitary representation) that would resist forgetting in the face of processing a second sentence. The purpose of presenting a second picture was to insure the subject’s processing of a second sentence. Using this experimental technique, an atypical pattern of verification latencies ITA < FA < TN < FN) was obtained (Experiment 1). When a sentence could be falsified in more than one way (Experiment 2), TN was found to be about equal to FN. A theoretical model was proposed to account for the present and the previous findings.     

Remembering words not presented in sentences: How study context changes patterns of false memories

People falsely endorse semantic associates and morpheme rearrangements of studied words at high rates in recognition testing. The coexistence of these results is paradoxical: Models of reading that presume automatic extraction of meaning cannot account for elevated false memory for foils that are related to studied stimuli only by their visual form; models without such a process cannot account for false memory for semantic foils. Here we show how sentence and list study contexts encourage different encoding modes and consequently lead to different patterns of memory errors. Participants studied compound words, such as tailspin and floodgate, as single words or embedded in sentences. We show that sentence contexts led subjects to be better able to discriminate conjunction lures (e.g., tailgate) from old words than did list contexts. Conversely, list contexts led to superior discrimination of semantic lures (e.g., nosedive) from old words than did sentence contexts.     

Emotion simulation during language comprehension

We report a novel finding on the relation of emotion and language. Covert manipulation of emotional facial posture interacts with sentence valence when measuring the amount of time to judge valence (Experiment 1) and sensibility (Experiment 2) of the sentence. In each case, an emotion-sentence compatibility effect is found: Judgment times are faster when facial posture and sentence valence match than when they mismatch. We interpret the finding using a simulation account; that is, emotional systems contribute to language comprehension much as they do in social interaction. Because the effect was not observed on a lexical decision task using emotion-laden words (Experiment 3), we suggest that the emotion simulation affects comprehension processes beyond initial lexical access.     

Not Propositions

Current computational accounts of meaning in the cognitive sciences are based on abstract, amodal symbols (e.g., nodes, links, propositions) that are arbitrarily related to their referents. We argue that such accounts lack convincing empirical support and that they do not provide a satisfactory account for linguistic meaning. One historic set of results supporting the abstract symbol view has come from investigation into comprehension of negated sentences, such as “The buttons are not black.” These sentences are presumed to be understood as two propositions composed of abstract symbols. One proposition corresponds to “the buttons are black,” and it is embedded in another proposition corresponding to “it is not true.” Thus, the propositional account predicts (a) that comprehension of negated sentences should take longer than comprehension of the corresponding positive sentence (because of the time needed to construct the embedding), but (b) that the resulting embedded propositions are informationally equivalent (but of opposite valence) to the simple proposition underlying the positive sentence. Contrary to these predictions, Experiment 1 demonstrates that negated sentences out of context are interpreted as situationally ambiguous, that is, as conveying less specific information than positive sentences. Furthermore, Experiment 2 demonstrates that when negated sentences are used in an appropriate context, readers do not take longer to understand them. Thus, difficulty with negation is demonstrated to be an artifact of presentation out of context. After discussing other serious problems with the use of abstract symbols, we describe the Indexical Hypothesis. This embodied account of meaning does not depend on abstract symbols, and hence it provides a more satisfactory account of meaning.     

The temporal structure of spoken sentence comprehension in Chinese

Three experiments with 60 native Chinese speakers were conducted to examine the temporal structure of spoken sentence comprehension in Chinese. In the two sentence-gating experiments, listeners heard sentence fragments of various sizes, either successively or individually, and decided on the agent role of the sentence. In the agent-naming experiment, listeners repeated aloud the sentence agent upon hearing the sentence. Converging evidence was derived from these experiments indicating the usefulness of sentence gating in tapping into the temporal structure of comprehension and in evaluating listeners’ comprehension on the basis of partial information. Results from these experiments show that partial sentence information plays a very important role in Chinese sentence comprehension, and that Chinese listeners integrate incomplete and ungrammatical sentence fragments by reference to complete and grammatical models in the language. It is argued that these processing characteristics reflect properties of the Chinese language in which few morphological, grammatical, and syntactic constraints are placed on the relationships between sentence constituents.     

Neural network processing of natural language: II. Towards a unified model of corticostriatal function in learning sentence comprehension and non-linguistic sequencing

A central issue in cognitive neuroscience today concerns how distributed neural networks in the brain that are used in language learning and processing can be involved in non-linguistic cognitive sequence learning. This issue is informed by a wealth of functional neurophysiology studies of sentence comprehension, along with a number of recent studies that examined the brain processes involved in learning non-linguistic sequences, or artificial grammar learning (AGL). The current research attempts to reconcile these data with several current neurophysiologically based models of sentence processing, through the specification of a neural network model whose architecture is constrained by the known cortico-striato-thalamo-cortical (CSTC) neuroanatomy of the human language system. The challenge is to develop simulation models that take into account constraints both from neuranatomical connectivity, and from functional imaging data, and that can actually learn and perform the same kind of language and artificial syntax tasks. In our proposed model, structural cues encoded in a recurrent cortical network in BA47 activate a CSTC circuit to modulate the flow of lexical semantic information from BA45 to an integrated representation of meaning at the sentence level in BA44/6. During language acquisition, corticostriatal plasticity is employed to allow closed class structure to drive thematic role assignment. From the AGL perspective, repetitive internal structure in the AGL strings is encoded in BA47, and activates the CSTC circuit to predict the next element in the sequence. Simulation results from Caplan's [Caplan, D., Baker, C., & Dehaut, F. (1985). Syntactic determinants of sentence comprehension in aphasia. Cognition, 21, 117-175] test of syntactic comprehension, and from Gomez and Schvaneveldts' [Gomez, R. L., & Schvaneveldt, R. W. (1994). What is learned from artificial grammars?. Transfer tests of simple association. Journal of Experimental Psychology: Learning, Memory and Cognition, 20, 396-410] artificial grammar learning experiments are presented. These results are discussed in the context of a brain architecture for learning grammatical structure for multiple natural languages, and non-linguistic sequences.     

Aphasic's comprehension of sentences expressing temporal order of events

Fifty-seven aphasic patients' comprehension of five sentence forms which express the temporal order of events has been investigated. Three major patterns of errors of temporal relations have emerged. The dominant pattern is exhibited by patients making errors on Before's but not on After's. The second and third patterns comprise a fewer number of patients making errors on After's but not on Before's and those making errors on both Before's and After's, respectively. It has been shown that different syndromes of aphasia are equally distributed among these three patterns. In the follow-up study of several patients, fluctuating courses of recovery in which comprehension of before and after alternately increases and decreases have been observed. These results are discussed in connection with a possible account of the acquisition of before and after in Japanese.     

Expectation-based syntactic comprehension

This paper investigates the role of resource allocation as a source of processing difficulty in human sentence comprehension. The paper proposes a simple information-theoretic characterization of processing difficulty as the work incurred by resource reallocation during parallel, incremental, probabilistic disambiguation in sentence comprehension, and demonstrates its equivalence to the theory of Hale [Hale, J. (2001). A probabilistic Earley parser as a psycholinguistic model. In Proceedings of NAACL (Vol. 2, pp. 159-166)], in which the difficulty of a word is proportional to its surprisal (its negative log-probability) in the context within which it appears. This proposal subsumes and clarifies findings that high-constraint contexts can facilitate lexical processing, and connects these findings to well-known models of parallel constraint-based comprehension. In addition, the theory leads to a number of specific predictions about the role of expectation in syntactic comprehension, including the reversal of locality-based difficulty patterns in syntactically constrained contexts, and conditions under which increased ambiguity facilitates processing. The paper examines a range of established results bearing on these predictions, and shows that they are largely consistent with the surprisal theory.     

句子加工中核心名词动物性效应及其认知机制

动物性核心名词在语言认知中有着重要的作用.基于句子产生和句子理解两方面的研究表明,句式选择、语态运用、词序排列、论元建构、格标记使用及从句加工等诸多方面都显著的受到句中核心名词动物性的影响.就该效应的来源及其作用机制,生命性名词表征特异性理论、词汇概念通达可及性层级理论、动物性名词句法突出性特征理论、基于频率分布的语言经验理论从不同的方面进行了分析.最后,就名词动物性效应的存在本质、作用机理以及文化和语境的影响等方面展开了评价和展望.     

Connectionist semantic systematicity

Fodor and Pylyshyn [Fodor, J. A., & Pylyshyn, Z. W. (1988). Connectionism and cognitive architecture: A critical analysis. Cognition, 28, 3–71] argue that connectionist models are not able to display systematicity other than by implementing a classical symbol system. This claim entails that connectionism cannot compete with the classical approach as an alternative architectural framework for human cognition. We present a connectionist model of sentence comprehension that does not implement a symbol system yet behaves systematically. It consists in a recurrent neural network that maps sentences describing situations in a microworld, onto representations of these situations. After being trained on particular sentence–situation pairs, the model can comprehend new sentences, even if these describe new situations. We argue that this systematicity arises robustly and in a psychologically plausible manner because it depends on structure inherent in the world.     

Word order in sentence processing: an experimental study of verb placement in German

We examine the question of whether the human comprehension device exhibits word-order preferences during on-line sentence comprehension. The focus is on the positioning of finite verbs and auxiliaries relative to subjects and objects in German. Results from three experiments (using self-paced reading and event-related brain potentials) show that native speakers of German prefer to process finite verbs in second position (i.e., immediately after the subject and before the object). We will account for this order preference in terms of the relative processing costs associated with SV_fO and SOV_f. Our finding that word-order preferences play an important role in the on-line comprehension of German sentences is compatible with results from previous studies on English and other languages.     

Syntactic Creativity Errors in Children's Wh-Questions

Previous work has reported that children creatively make syntactic errors that are ungrammatical in their target language, but are grammatical in another language. One of the most well-known examples is medial wh-question errors in English-speaking children's wh-questions (e.g., What do you think who the cat chased? from Thornton, 1990). The evidence for this non-target-like structure in both production and comprehension has been taken to support the existence of innate, syntactic parameters that define all possible grammatical variation, which serve as a top-down constraint guiding children's language acquisition process. The present study reports new story-based production and comprehension experiments that challenge this interpretation. While we replicated previous observations of medial wh-question errors in children's sentence production (Experiment 1), we saw a reduction in evidence indicating that English-speaking children assign interpretations that conform to the medial wh-question pattern (Experiment 2). Crucially, we found no correlation between production and comprehension errors (Experiment 3). We suggest that these errors are the result of children's immature sentence production mechanisms rather than immature grammatical knowledge.     

Meaning construction and integration in children with hydrocephalus

Text comprehension processes were investigated in children with hydrocephalus, a neurodevelopmental disorder associated with good word decoding, but deficient reading comprehension. In Experiment 1, hydrocephalus and control groups were similar in processes related to activating word meanings and using context to enhance meaning. The hydrocephalus group was poorer at suppressing contextually irrelevant meanings. In Experiment 2, the hydrocephalus group had difficulty integrating information from an earlier read sentence to understand a new sentence as textual distance between the two propositions increased, suggesting difficulty in reactivation processes related to comprehension. Results are discussed in relation to cognitive and neurocognitive models of comprehension.     

Uncertainty and Expectation in Sentence Processing: Evidence From Subcategorization Distributions

There is now considerable evidence that human sentence processing is expectation based: As people read a sentence, they use their statistical experience with their language to generate predictions about upcoming syntactic structure. This study examines how sentence processing is affected by readers' uncertainty about those expectations. In a self‐paced reading study, we use lexical subcategorization distributions to factorially manipulate both the strength of expectations and the uncertainty about them. We compare two types of uncertainty: uncertainty about the verb's complement, reflecting the next prediction step; and uncertainty about the full sentence, reflecting an unbounded number of prediction steps. We find that uncertainty about the full structure, but not about the next step, was a significant predictor of processing difficulty: Greater reduction in uncertainty was correlated with increased reading times (RTs). We additionally replicated previously observed effects of expectation violation (surprisal), orthogonal to the effect of uncertainty. This suggests that both surprisal and uncertainty affect human RTs. We discuss the consequences for theories of sentence comprehension.     

A probabilistic model of semantic plausibility in sentence processing

Experimental research shows that human sentence processing uses information from different levels of linguistic analysis, for example, lexical and syntactic preferences as well as semantic plausibility. Existing computational models of human sentence processing, however, have focused primarily on lexico-syntactic factors. Those models that do account for semantic plausibility effects lack a general model of human plausibility intuitions at the sentence level. Within a probabilistic framework, we propose a wide-coverage model that both assigns thematic roles to verb–argument pairs and determines a preferred interpretation by evaluating the plausibility of the resulting ( verb , role , argument ) triples. The model is trained on a corpus of role-annotated language data. We also present a transparent integration of the semantic model with an incremental probabilistic parser. We demonstrate that both the semantic plausibility model and the combined syntax/semantics model predict judgment and reading time data from the experimental literature.     

Sentence comprehension and memory for embedded structure

An experimental simulation technique is introduced to study memory constraints affecting comprehension of embedded sentences. Memory for embedded structure is measured in a pairing task that simulates requirements imposed by sentence comprehension. With increasing memory load (number of embeddings), the rate of performance decline is consistent across all (recall and comprehension) conditions, and reflects a loss of order (as opposed to item) information. Short-term memory overload, rather than linguistic complexity, appears sufficient to account for comprehension difficulty. Performance was imperfect even with singly embedded constructions. Recall showed no abrupt failure with multiple embeddings, indicating that memory constraints are not as discontinuous as English syntax might suggest.