Hemispheric processing of inferences: The effects of textual constraint and working memory capacity

In this study, we investigated hemispheric differences in the generation of bridging and predictive inferences. Participants read texts that provided either strong or weak causal constraints for a particular bridging (Experiment 1) or predictive (Experiment 2) inference and performed a lexical decision task to inference-related targets presented to the left or the right hemisphere. Facilitation for strongly constrained bridging and predictive inferences was found in both hemispheres. In contrast, facilitation for weakly constrained inferences was stronger in the right than in the left hemisphere for both bridging and predictive inferences, although for the latter there was some facilitation in the left hemisphere as well. We also considered whether these effects differ as a function of the working memory capacity of the reader. High working memory capacity readers showed greater facilitation for strongly constrained inferences than for weakly constrained inferences in both hemispheres, whereas low working memory capacity readers showed this same pattern in the left hemisphere but showed equal facilitation for strongly and weakly constrained inferences in the right hemisphere. These results suggest that hemispheric processing, textual constraint, and working memory capacity interact to affect how readers generate causal inferences.     

Working memory and inferences: evidence from eye fixations during reading

Eye fixations during reading were monitored to examine the relationship between individual differences in working memory capacity-as assessed by the reading span task-and inferences about predictable events. Context sentences predicting likely events, or non-predicting control sentences, were presented. They were followed by continuation sentences in which a target word represented an event to be inferred (inferential word) or an unlikely event (non-predictable word). A main effect of reading span showed that high working memory capacity was related to shorter gaze durations across sentence regions. More specific findings involved an interaction between context, target, and reading span on late processing measures and regions. Thus, for high- but not for low-span readers, the predicting condition, relative to the control condition, facilitated reanalysis of the continuation sentence that represented the inference concept. This effect was revealed by a reduction in regression-path reading time in the last region of the sentence, involving less time reading that region and fewer regressions from it. These results indicate that working memory facilitates elaborative inferences during reading, but that this occurs at late text-integration processes, rather than at early lexical-access processes.     

关键词对精加工推理的影响作用初探

本实验研究在文本中突出关键词对精加工推理的产生水平和产生阶段的影响。研究采用了让被试阅读后做测试题的方法,通过分析测试题的成绩得出的结果表明：在文章中突出关键词有助于产生近推理和内涵推理,且使这两类推理更有可能在阅读过程中产生。本研究补充了McDaniel和Donnelly 1996年的研究结果。     

SEMANTIC AND PHONOLOGICAL CODING PROCESSES IN SKILLED AND LESS SKILLED READERS' READING OF TEXT

Abstract

Two experiments examine the memory coding processes of skilled and less skilled readers during the reading of connected text. In experiment 1, students read several paragraphs which required a lexical decision about an underlined letter string within a sentence. Underlined letter strings were either synonyms, repeated words, or control words in reference to items in the sentence. Students were later asked to recall words related to their lexical decision, as well as verify the occurrence of sentences from the text. Skilled readers recalled more synonyms than poor readers, whereas no differences emerged between groups in their recall of other types of words related to the lexical task or for the verification of sentences. Experiment 2 procedures were similar to Experiment 1, except that synonyms were replaced with homophones and the sentence verification task included phrases related to the homophones. When compared to less skilled readers, skilled readers recalled more homophones and repeated words, but were more likely to be disrupted in correct verification of sentences with homophones. Taken together, the experiments suggest that along with phonological coding, semantic processing contributes an important amount of variance to deficiencies in the reading of connected text.     

Testing the shared resource assumption in theories of text processing

Eight experiments evaluated a core assumption of several theories of text processing, the shared resource assumption, which states that component text processes share limited processing resources. Short texts each contained two critical sentences that together warranted a causal inference. The syntactic structure of the second sentence was either more or less difficult to parse. Results from a lexical decision task suggested that readers formed the causal inferences when the syntactic structure was less difficult to parse but that inferencing was constrained when syntactic structure was more difficult. Follow-up experiments suggested that this interference was not due to inferior output of the syntactic parser nor to the increased demands of difficult syntax interfering with maintenance of information needed to form the inference. The results suggest insufficient resources were available for the operation of inference processes due to the increased demands of syntactic parsing, consistent with the shared resource assumption.     

文本阅读中预期推理生成的容量制约性

本研究通过操纵读者信息加工的时间量考察了工作记忆容量对预期推理生成的影响性质。结果表明,如果没有信息加工时间的限制,则工作记忆容量低的读者也能表现出在阅读中生成预期推理的证据,而如果读者的阅读加工时间受一定限制,则工作记忆容量低者建构预期推理的证据随着被允许的加工时间的缩短而逐渐消失。结果支持“工作记忆容量只影响阅读中预期推理建构速度”的观点。     

The effect of foregrounding on readers’ use of predictive inferences

This research extends previous attempts to determine whether subjects make predictive inferences during comprehension. For example, when subjects read a passage about someone falling out of a 14th-story window, do they infer that the person is dead? Previous research in which lexical decision, word naming, and recognition tasks have been used for detecting predictive inferences has had mixed results. In experiment 1, a word-stem completion task was used to test for predictive inferences. The word stems were formed from target inferences that followed either priming or control passages. The data revealed that predictive inferences are generated only about concepts that are foregrounded in the passages. In Experiments 2 and 3, lexical decision and naming were used to test for predictive inferences. The lexical decision data replicated the word-stem completion data. A control experiment ruled out a simple context-checking explanation for the lexical decision results. The naming data indicated that this tasks was not sensitive to elaborative inference generation. Theresults show that readers make predictive inferences, but do so selectively.     

The measurement of reading comprehension: How not to trade construct validity for predictive power

This paper proposes a way to measure and understand individual differences in reading comprehension. The measure is based on the theory that individual differences in reading skill reflect differences in the processing capacity of working memory. The measure of processing capacity, called the reading span test, requires subjects to read aloud increasingly longer sets of sentences and to recall the final word of each sentence in the set. Reading span, the number of final words recalled, varies from two to five for university students. According to the proposed theory, poor readers devote so much capacity to processing the sentences, that they have less residual capacity for storing and producing the final words. The ability to keep information accessible in working memory is particularly crucial for the processes that integrate each ne word with previously read information. And, indeed, reading span is highly correlated with integration processes such as computing pronominal reference, monitoring and revising comprehension errors, and making inferences. The theoretical and predictive value of the reading span measure can be contrasted with the relatively limited value of existing atheoretical and theoretical measures of reading comprehension.     

不同阅读能力聋人语篇理解中连接推理的眼动比较

采用记录被试阅读完成后回答问题的正确率、反应时与眼动指标,试图探讨不同阅读能力聋人语篇理解中连接推理的加工特点与效率。结果表明:在语篇局部连贯中断情况下,高低阅读能力聋人在语篇理解过程中的连接推理加工是主动的即时性加工,而非语篇阅读完成后被动的延时性加工;但高阅读能力聋人能有效激活背景知识,所发生的连接推理加工是一个自动化的加工;低阅读能力聋人主动尝试即时激活背景知识、构建句子间的连接推理,但连接推理的加工还未达到自动化程度,推理加工的效率较低。     

Time course of priming for associate and inference words in a discourse context

The construction of word meanings in a discourse context was conceptualized as a process of sense activation, sense selection, and sense elaboration. In three experiments, subjects read texts presented by a rapid serial visual procedure and performed a lexical decision on visually presented targets that followed ambiguous prime words. When the target was a word, it was either an associate of the prime word, a probable inference suggested by the discourse, or an unrelated word. For associates, lexical decisions that related to either the appropriate or the inappropriate sense of the ambiguous word were generally facilitated at short (200-400 msec) prime-target stimulus onset asynchronies (SOAs). At longer SOAs, responses were faster to appropriate than to inappropriate associates. For the thematic inferences, there was no difference between these (appropriate) inferences and (inappropriate) control words at short SOAs. At long SOAs (1,000 and 1,500 msec), however, inference words were facilitated. The results are interpreted as consistent with a model of lexical processing in which sense activation functions independently of context. Discourse context effects, whether on sense selection (suppression of inappropriate associates) or on sense elaboration (creation of inferences), are seen as postlexical.     

WORKING MEMORY CAPACITY DISSOCIATES LEXICAL AND SENTENTIAL CONTEXT EFFECTS

Abstract— Semantically associated and unassociated word pairs were embedded in normal meaningful sentences and in sentences that were semantically anomalous throughout The influence of lexical context was isolated via comparison of responses to the second words of the associated and unassociated pairs. The influence of sentence-level context was isolated by comparing responses to the same words in the two sentence types Subjects of high, medium and low working memory capacity (as evaluated by the reading span test) showed modulations of event-related brain potentials in response so lexical context. In contrast, only the high- and medium-capacity groups were responsive to purely sentence-level semantic context. The results demonstrate that sentential context influences the processing of words in intermediate sentence positions at normal reading speeds but that the on-line utilization of this context is more demanding of working memory than single-word contexts.     

An on-line assessment of causal reasoning during comprehension

Fletcher and Bloom (1988) have argued that as readers read narratives, clause by clause, they repeatedly focus their attention on the last preceding clause that contains antecedents but no consequences in the text. This strategy allows them to discover a causal path linking the text's opening to its final outcome while minimizing the number of times long-term memory must be searched for missing antecedents or consequences. In order to test this hypothesis, we examined the reading times of 25 subjects for each clause of eight simple narrative texts. The results show that: (1) causal links between clauses that co-occur in short-term memory (as predicted by the strategy) increase the time required to read the second clause; (2) potential causal links between clauses that never co-occur in short-term memory (again as predicted by the strategy) have no effect on reading time; and (3) reinstatement searches are initiated at the end of sentences that are causally unrelated to the contents of short-term memory or that contain clauses that satisfy goals no longer in short-term memory. These results support the claim that subjects engage in a form of causal reasoning when they read simple narrative texts.     

SEMANTIC INTEGRATION: A COMPARISON OF NORMAL AND DISABLED READERS

Abstract

This study examined comprehension of sets of related sentences from which inferences could be drawn. Subjects were 40 disabled and 40 normal readers (C.A. 10‐10 ‐ 11‐11) who were asked to read short paragraphs suggesting spatial relationships. After reading, they were given a recognition test containing items not previously read but which were structured in their relationship to the semantic content of the texts. Analyses of within and between group errors reveal significant differences due to sentence type and a strong tendency for both groups to “misrecognize” semantically congruent inferences. The results are discussed in light of past research on the memory capabilities of disabled readers and with respect to techniques of remediation.     

The effects of readers’ goals on inference generation and memory for texts

We investigated the effects of readers' goals on inference generation and memory for expository text. College students (N = 82) read texts for the purpose of either study or entertainment. On-line inference generation was recorded via think-aloud procedures, and off-line memory was assessed via free recall. Reading goal strongly influenced inferential activity: Readers with a study goal produced more coherence-building (i.e., backward/explanatory and forward/predictive) inferences, whereas readers with an entertainment goal produced more associations and evaluations. These differences were associated with superior memory for the texts in the study condition. The results indicate that inference generation during reading is partly strategic and is influenced systematically by reading purpose. We propose that reading goals influence readers' standards of coherence, which in turn influence the types of inferences that they draw and the final memory representations that they construct.     

Working Memory Capacity and Time Course of Predictive Inferences

A naming task assessed activation of inference concepts during reading. A predicting, or a control, context sentence was followed by a target word to be named, which represented the predicted event or an inconsistent event. The interval between the end of the context and the onset of the target word varied between 50 and 1050-msec. Individual differences in working memory capacity were assessed by the reading span task. As reflected by facilitation in naming latencies in the predicting condition, relative to the control condition, (a) inferences were not made within the first 50-msec after the context, regardless of reading span; (b) only the high-span participants drew inferences within a 550-msec interval; and, (c) both the high-and the low-span participants generated them within a 1050-msec interval. These results indicate that high working memory capacity accelerates the time course of predictive inferences, although they do not become automatic. We propose that this effect occurs because these inferences involve time-consuming elaborations that place demands on the effortful and limited resources of working memory. Deficiencies in word knowledge, speed of lexical access, or comprehension of explicit information do not account for low-span readers' difficulties in generating predictive inferences.     

Working memory capacity and time course of predictive inferences

A naming task assessed activation of inference concepts during reading. A predicting, or a control, context sentence was followed by a target word to be named, which represented the predicted event or an inconsistent event. The interval between the end of the context and the onset of the target word varied between 50 and 1050 msec. Individual differences in working memory capacity were assessed by the reading span task. As reflected by facilitation in naming latencies in the predicting condition, relative to the control condition, (a) inferences were not made within the first 50 msec after the context, regardless of reading span; (b) only the high-span participants drew inferences within a 550-msec interval; and, (c) both the high- and the low-span participants generated them within a 1050-msec interval. These results indicate that high working memory capacity accelerates the time course of predictive inferences, although they do not become automatic. We propose that this effect occurs because these inferences involve time-consuming elaborations that place demands on the effortful and limited resources of working memory. Deficiencies in word knowledge, speed of lexical access, or comprehension of explicit information do not account for low-span readers' difficulties in generating predictive inferences.     

Semantic involvement in the lexical and sentence processing of Japanese kanji

This study examined how skilled Japanese readers activate semantic information when reading kanji compound words at both the lexical and sentence levels. Experiment 1 used a lexical decision task for two-kanji compound words and nonwords. When nonwords were composed of kanji that were semantically similar to the kanji of real words, reaction times were longer and error rates were higher than when nonwords had kanji that were not semantically similar. Experiment 2 used a proofreading task (detection of kanji miscombinations) for the same two-kanji compound words and nonwords at the sentence level. In this task, semantically similar nonwords were detected faster than dissimilar nonwords, but error rates were much higher for the semantically similar nonwords. Experiment 3 used a semantic decision task for sentences with the same two-kanji compound words and nonwords. It took longer to detect semantically similar nonwords than dissimilar nonwords. This indicates that semantic involvement in the processing of Japanese kanji produces different effects, depending on whether this processing is done at the lexical or sentence level, which in turn is related to where the reader's attention lies.     

Working memory contributions to reading comprehension components in middle childhood children

This study examined working memory contributions to reading comprehension subskills in Greek children (mean age 9 years, 1 month). The phonological loop of the Baddeley and Hitch working memory model was assessed with 3 recall tasks (words, nonwords, and digits) and a word list matching task. The central executive (CE) was assessed with 3 tasks (listening, counting, and backward digit recall). Participants were also given a receptive vocabulary task, a reading fluency task, and written stories accompanied by comprehension questions. Canonical correlation analyses showed that the comprehension variables were related to the CE rather than the phonological loop measures. CE functions were more strongly associated with elaborative inference generation (involving significant offline processing) and comprehension control (involving metacognitive monitoring). Smaller yet significant associations were observed between the CE and the necessary inference and literal comprehension measures, whereas a moderate relationship was found in the case of the simile comprehension variable. Among the CE variables, listening recall demonstrated the highest loading on the canonical function, followed by moderate yet significant counting and backward digit recall loadings. Vocabulary was found to fully mediate several associations between working memory and comprehension measures; however, the relationship between listening recall and elaborative inferences was partly mediated. Reading fluency and, on several occasions, Greek vocabulary knowledge did not mediate the relationships between CE measures and comprehension skills assessed. This study demonstrates the usefulness of CE measures for identifying young children's possible difficulties in carrying out specific reading comprehension processes.     

The role of working memory in inferential sentence comprehension

Existing literature on inference making is large and varied. Trabasso and Magliano (Discourse Process 21(3):255–287, 1996) proposed the existence of three types of inferences: explicative, associative and predictive. In addition, the authors suggested that these inferences were related to working memory (WM). In the present experiment, we investigated whether WM capacity plays a role in our ability to answer comprehension sentences that require text information based on these types of inferences. Participants with high and low WM span read two narratives with four paragraphs each. After each paragraph was read, they were presented with four true/false comprehension sentences. One required verbatim information and the other three implied explicative, associative and predictive inferential information. Results demonstrated that only the explicative and predictive comprehension sentences required WM: participants with high verbal WM were more accurate in giving explanations and also faster at making predictions relative to participants with low verbal WM span; in contrast, no WM differences were found in the associative comprehension sentences. These results are interpreted in terms of the causal nature underlying these types of inferences.     

儿童句子错误记忆:发展性逆转现象及精加工推理效应

为探究文本背景下句子错误记忆的发展性逆转现象及精加工推理效应,95名有效被试学习3篇文本材料后,参与由学过句、内涵推理句、外延推理句和无关句组成的再认测验。结果发现：（1）高中二年级被试学过句的正确再认率和关键诱饵句的校正的错误再认率均显著高于小学五年级和初中二年级,后二者差异不显著;（2）内涵推理句错误再认率高于外延推理句,高中二年级被试尤甚。结论：（1）句子真实记忆随着年龄增长而增加;句子错误记忆存在发展性逆转现象,初中二年级到高中二年级是句子错误记忆发展相对迅速的阶段;（2）文本背景下,不同的精加工推理诱发了不同程度的句子错误记忆,这种精加工推理效应与一般世界知识的自上而下激活的程度有关。