RELATIONS BETWEEN LANGUAGE AND MEMORY:

Abstract— Immediate recall decreases for repeated relative to unrepeated words in rapidly presented spoken lists, a phenomenon known as repetition deafness (RD). The present study examines RD as a test case for a distributed memory account of relations between language and memory. Within the distributed memory framework, general connection-formation processes required for language comprehension influence repetition deficits in sentences. Thus, RD should increase as a function of factors, such as listlike sentence prosody, that disrupt the formation of word-to-phrase links. Also under this account, repetition blindness and RD in sentences should display fundamentally similar characteristics (e.g., relative insensitivtty to low-level sensory differences between the repeated words) Using innovative procedures for computer sound manipulation, the present study obtained data supporting both predictions RD increased for sentences generated with listlike versus normal prosody, but did not differ reliably for acoustically similar versus dissimilar repeated words. Implications of these data for the general issue of relations between language and memory are discussed.     

Repetition deficits, list context, and word-class interactions in the RSVP of words in sentences.

We report a failure to find a repetition deficit in recall following the rapid serial visual presentation (RSVP) of words within sentences, using adjectives rather than nouns as the critical items. In a series of experiments that ruled out participant and procedural differences as the source of the failure, both word class and list context were found to moderate the repetition deficit, but grammatical necessity did not. The presence in the list of sentences in which the repeated adjectives were separated by more than three words (i.e., more than 400 ms in RSVP) not only eliminated the repetition deficit for the recall of those sentences but also for the recall of sentences in which the repeated adjectives were separated by three or fewer words (i.e., less than 400 ms in RSVP). However, although substantially reduced, a repetition deficit with noun-based materials was still found in this list context. Matching the adjective-based sentences with the noun-based sentences in sentence length and position of the critical items revealed that the moderating effect of word-class on the repetition deficit was mediated by the biases in sentence structure that using different word classes tend to induce.     

When words collide: facilitation and interference in the report of repeated words from rapidly presented lists

Inhibited encoding is the basis of some accounts of repetition blindness--impaired report of the second occurrence of a repeated word in a rapidly presented word sequence. The author presents evidence for the claim that repetition effects arise from constructive processes of perception and memory that occur to some extent after the word sequence has been presented. Unpredictable postlist cues prompted subjects to report either the entire list or just the final word in the list. Repetition impaired the report of the second occurrence of a repeated word under full report but facilitated the report of such items when only the final word had to be reported. The author modulated this dissociation by presenting repeated words in sentences rather than unrelated word lists. The sensitivity of the effects of repetition to postlist cues supports a construction rather than an encoding inhibition account of repetition blindness.     

重复知盲效应产生的机制——来自汉语叠词加工的证据

以汉语中的叠词为实验材料,对重复知盲产生机制的类型标记个体化理论和建构理论进行检验。采用RSVP研究范式,在两个实验中让178名被试口头报告列表中出现的所有汉字（全部报告）或列表中最后两个字（部分报告）。结果发现：（1）在呈现速率为128ms和198ms时,完全重复的汉字在全部报告和部分报告任务中均存在RB效应,但在部分报告任务下出现的RB相比全部报告任务的减少了,符合建构理论的假设。（2）在全部报告条件下,不仅重复刺激的正确报告率会降低,而且与重复刺激相关的非重复刺激的正确报告率也会降低。研究表明建构理论比类型标记个体化理论更能解释RB效应的产生机制。     

Repetition blindness and bilingual memory: Token individuation for translation equivalents

The repetition blindness effect (RB) occurs when individuals are unable to recall a repeated word relative to a nonrepeated word in a sentence or string of words presented in a rapid serial visual presentation task. This effect was explored across languages (English and Spanish) in an attempt to provide evidence for RB at a conceptual level using noncognate translation equivalents (e.g.,nephew-sobrino). In the first experiment, RB was found when a word was repeated in an English sentence but not when the two repetitions were in different languages. In the second experiment, RB was found for identical repetitions in Spanish and in English using word lists. However, the crosslanguage condition produced significant facilitation in recall, suggesting that although conceptual processing had taken place, semantic overlap was not sufficient to produce RB. The results confirm Kanwisher’s (1987) token individuation hypothesis in the case of translation equivalents.     

Repetition blindness: The effects of stimulus modality and spatial displacement

Repetition blindness (Kanwisher, 1986, 1987) is the failure to detect repetitions of words in lists presented in rapid serial visual presentation (RSVP). Two questions were investigated in the present study. First, if repetition blindness is not found with auditory presentation, it would support a specifically visual account of the effect. Second, if displacement of the two instances in visual space eliminates repetition blindness, it would suggest that repetition blindness is restricted to instances in which identical stimuli are distinguished soley by temporal differences. In Experiment 1, the subjects omitted second occurrences of repeated words in verbatim recall of rapid sentences presented visually (in RSVP), but not auditorily (using compressed speech), indicating that repetition blindness is a modality-specific phenomenon. In Experiments 2 and 3, repetition blindness was observed even when two occurrences of a written word were presented in different locations, showing that distinct locations do not guarantee token individuation. The results are discussed within a model that distinguishes between processes of type recognition and token individuation.     

SEMANTIC BLINDNESS:

Abstract— This study demonstrates a recently predicted cognitive phenomenon known as semantic blindness, an inhibitory effect attributable to concept repetition in the serial recall of rapidly presented sentences Proficient bilinguals read mixed, Spanish-English sentences each including a target and a pretarget word Targets and pretargets were related in three ways They were identical(e g, like-like), semantically identical across languages (e g, gusta-like), and noridentical within or across languages (e g, read-like) Equivalent repetition blindness was found for targets with identical and semantically identical pretargets, indicating that repetition deficits were occurring solely at the semantic level, rather than at orthographic or phonological levels     

Interresponse times in serial recall: effects of intraserial repetition

The authors examined the effects of intraserial repetition on multitrial serial learning of random consonant lists, analyzing both learning rates and perfect trial interresponse times (IRTs). Lists varied along 3 dimensions: list length, presence or absence of a repeated element, and lag between repeated elements. After achieving a forward-recall criterion on a given list, participants (N = 20) attempted backward recall. At small lags, IRTs between the repeated elements were very short (compared with IRTs from identical positions in nonrepetition lists). At larger lags, the IRT to recall the second repeated item was substantially longer than in control lists. These results reveal a latency analogue of the Ranschburg pattern seen in accuracy data. A Ranschburg pattern was also found in participants' learning rates. These results both generalize the Ranschburg phenomenon and present further challenges to theories of serial order memory.     

Total retrieval time and hypermnesia: Investigating the benefits of multiple recall tests

Hypermnesia is an increase in recall over repeated tests. A core issue is the role of repeated testing, per se, versus total retrieval time. Prior research implies an equivalence between multiple recall tests and a single test of equal total duration, but theoretical analyses indicate otherwise. Three experiments investigated this issue using various study materials (unrelated word lists, related word lists, and a short story). In the first experimental session, the study phase was followed by a series of short recall tests or by a single, long test of equal total duration. Two days later, participants took a final recall test. The multiple and single test conditions produced equivalent performance in the first session, but the multiple test group exhibited less forgetting and fewer item losses in the final test. In a fourth experiment, using a brief delay (15 min) between the recall sessions, the multiple recall condition produced greater hypermnesia as well as fewer item losses. In addition, final recall was significantly higher in the multiple than in the single test condition in three of the four experiments. Thus, single and repeated recall tests of equal total duration are not functionally equivalent, but rather produce differences observable in subsequent recall tests.     

The contributions of encoding,retention, and recall to the Hebb effect

The article reports an experiment testing whether the Hebb repetition effect—the gradual improvement of immediate serial recall when the same list is repeated several times—depends on overt recall of the repeated lists. Previous reports which suggest that recall is critical confound the recall manipulation with retention interval. The present experiment orthogonally varies retention interval (0 or 9 s) and whether the list is to be recalled after the retention interval. Hebb repetition learning is assessed in a final test phase. A repetition effect was obtained in all four experimental conditions; it was larger for recalled than non-recalled lists, whereas retention interval had no effect. The results show that encoding is sufficient to generate cumulative long-term learning, which is strengthened by recall. Rehearsal, if it takes place in the retention interval at all, does not have the same effect on long-term learning as overt recall.     

Surface form and the spacing effect

Deficient processing theories of the spacing effect attribute poor recall of massed-repeated items to a failure to process one or both of the presentations fully. An implication of this approach is that anything that increases the probability that a repetition will receive full processing, or conversely, anything that decreases the probability that the item will be recognized as a repetition, should improve memorability of the item. The present set of experiments tested this prediction by manipulating the surface structure of repeated sentences. On the basis of previous research, it was assumed that memory for surface structure of sentences decays rapidly, and hence can contribute to initial identification of repetitions only at short spacings. Because this manipulation should hinder recognition of repetitions as repetitions, it was expected to induce full processing of massed repetitions, and thus facilitate recall of these items. This prediction was supported. When sentences were repeated verbatim (Experiment 1) or by the same speaker (Experiment 2), the typical spacing effect was obtained. However, when the surface structure or speaker changed at time of repetition, massed repetitions were recalled nearly as well (Experiment 1) or as well (Experiment 2) as their spaced counterparts.     

Calling all codes: interactive effects of semantics, phonology, and orthography produce dissociations in a repetition blindness paradigm

We used the repetition blindness (RB) paradigm to examine the roles of semantics, phonology, and orthography on report from rapidly presented 5-item word lists. Semantic primes in positions 1 and 3 in the list preceded homographic homophones, homographic heterophones, and heterographic homophones as critical targets in positions 2 and 4. All codes (i.e., semantic, phonologic, and orthographic) were repeated, or meaning changed while only phonology, only orthography, or both were repeated for the critical targets. Using a scoring procedure that considered order of report, we assessed facilitation for report of the first instance of the repeated target (a novel aspect of our procedure) and RB for report of the second instance of the repeated target. Except when accompanied by a change in orthography, a change in meaning reduced RB relative to when all codes were repeated. Facilitation in the report of the first instance of the repeated item occurred only if meaning changed and phonology was repeated (independent of whether orthography changed). Finally, recall was worse for the meaning change nonrepeated control lists (which instantiated 3 unrelated meanings in the list, including the last unrelated item) than for the no-meaning change control lists (which instantiated 2 unrelated meanings). We discuss the relevance of these findings for extant accounts of repetition blindness.     

An auditory repetition deficit under low memory load

Previous studies of the auditory analogue of repetition blindness have led to different conclusions regarding the nature of the effect (e.g., N. Kanwisher & M. C. Potter, 1989; M. Miller & D. MacKay, 1994). In the present study, recall accuracy for repeated elements was examined with lists of 2 or 3 items presented dichotically under high temporal pressure. When this procedure was used, a repetition deficit in recall was obtained for both vowels (Experiment 1) and consonant-vowel syllables (Experiment 2). Further experiments demonstrated that this deficit decreases as the stimulus onset asynchrony between the 2 critical elements increases (Experiment 3) and showed that the effect also occurs for words and not just nonsense syllables (Experiment 4). In all 4 experiments, estimations of guessing biases showed that responses to unrepeated lists were not artificially favored over responses to repeated lists.     

Spatial modulation of repetition blindness and repetition deafness

When two identical visual items are presented in rapid succession, people often fail to report the second instance when trying to recall both (e.g., Kanwisher, 1987). We investigated whether this temporal processing deficit is modulated by the spatial separation between the repeated stimuli within both audition and vision. In Experiment 1, lists of one to three digits were rapidly presented from loudspeaker cones arranged in a semicircle around the participant. Recall accuracy was lower when repeated digits were presented from different positions rather than from the same position, as compared to unrepeated control pairs, demonstrating that auditory repetition deafness (RD) is modulated by the spatial displacement between repeated items. A similar spatial modulation of visual repetition blindness (RB) was reported when pairs of masked letters were presented visually from either the same or different positions arranged on a semicircle around fixation (Experiment 2). These results cannot easily be accounted for by the token individuation hypothesis of RB (Kanwisher, 1987; Park & Kanwisher, 1994) and instead support a recognition failure account (Hochhaus & Johnston, 1996; Luo & Caramazza, 1995, 1996).     

Accessing integrated and nonintegrated propositional structures in memory

This paper investigates the access properties associated with different propositional structures. Two memory experiments are reported, in which the underlying structures of sentences were integrated or not. Some sentences tested had the same concept repeated across the propositions (integrated), whereas other sentences had no explicit repeated arguments (non-integrated). Accessibility to the memory traces of the sentences was manipulated through the acquisition and the testing conditions. In Experiment 1, subjects received either immediate or delayed recall tests, under free or cued conditions. Integrated sentences were recalled better than nonintegrated ones under conditions of high accessibility (immediate recall or delayed cued recall). In contrast, under the low-access condition (delayed free recall), nonintegrated sentences were recalled slightly better than the integrated ones. Experiment 2 confirmed and extended the results for delayed free recall. Here again, under conditions of low sentence access, nonintegrated sentences were recalled better. These results were interpreted according to theory dealing with the lag effect in list learning.     

REPETITION BLINDNESS INTERACTS WITH SYNTACTIC GROUPING IN RAPIDLY PRESENTED SENTENCES

Abstract— This study rested for predicted effects of syntax on a repetition deficit (RD) known as repetition blindness, the reduced probability of recall for repeated words in rapid serial visual presentation (RSVP) sentences The syntactic variable was phrase-congruent versus phrase-incongruent grouping within simultaneous RSVP displays With phrase-congruent grouping, each RSVP display contained a syntactic phrase (e g, "to play sports" in the sentence "They wanted to play sports but sports were not allowed"), whereas with phrase-incongruent grouping, RSVP displays contained nonphrases (e g, "sports but sports") RD was extensive with phrase-incongruent grouping (29%), but nonsignificant (6%) with phrase-congruent grouping, as if phrase-congruent groups free up the extra processing time needed to connect repeated words into phrases in rapidly presented sentences, enabling top-down retrieval of the phrases during recall The present results comport with effects of syntax and prosody on auditory RD, and suggest that visual and auditory RD are identical underlying phenomena.     

Evidence for alternative strategies of sentence retention

Measurements of pupil size were taken while subjects listened to sentences and either tried to repeat them (R) or answered a question about them (QuA), after either a 3 or 7 sec retention interval. Pupil dilations were larger for R than for QuA, both towards the end of presentation and during the retention interval. Similar results were obtained by Kahneman and Wright (1971) when comparing total and partial recall of word lists. They attributed the differences in pupil dilations to differences in rehearsal strategy. However, the interaction between recall condition and retention interval was less convincing with sentential material, and an alternative explanation is suggested in terms of the level of abstraction at which the sentences are processed. This interpretation is supported by other evidence relating to the existence of alternative sentence retention strategies.

Pupil dilations failed to reveal phrase juncture phenomena, although two levels of ambient illumination were used in the hope of detecting such effects. In a modified question condition (QuB), where the question was given before the sentence, pupil dilations varied as a function of the part of the sentence providing the answer. These data indicated that people did not begin to frame their answer until they encountered in the sentence those words used in the question.     

Recall of short word lists presented visually at fast rates: Effects of phonological similarity and word length

Phonological similarity of visually presented list items impairs short-term serial recall. Lists of long words are also recalled less accurately than are lists of short words. These results have been attributed to phonological recoding and rehearsal. If subjects articulate irrelevant words during list presentation, both phonological similarity and word length effects are abolished. Experiments 1 and 2 examined effects of phonological similarity and recall instructions on recall of lists shown at fast rates (from one item per 0.114-0.50 sec), which might not permit phonological encoding and rehearsal. In Experiment 3, recall instructions and word length were manipulated using fast presentation rates. Both phonological similarity and word length effects were observed, and they were not dependent on recall instructions. Experiments 4 and 5 investigated the effects of irrelevant concurrent articulation on lists shown at fast rates. Both phonological similarity and word length effects were removed by concurrent articulation, as they were with slow presentation rates.     

汉语颠倒词加工中重复知盲发生的水平

以汉语中的颠倒词为材料, 探讨汉语重复知盲的发生水平。采用RSVP方式呈现包含颠倒词对(C1和C2)的句子列表或词语列表, 让被试对列表中的词进行全部报告或报告是否存在重复词, 记录报告C2的正确率。三个实验采用3 (词语的重复性: 颠倒重复、完全重复、不重复) × 2 (颠倒词对的意义相似性: 意义不同、意义相似)两因素被试内设计, 结果在全部报告和报告是否存在重复词的实验中都出现重复性和意义相似性的交互效应, 但不同实验任务, 数据模式不同。该结果表明, 在RSVP任务中, 汉语颠倒词加工存在RB效应, 该效应产生的水平受制于不同的实验任务, 在部分报告任务中, 颠倒词的RB效应发生在知觉水平, 在全部报告任务中, 颠倒词的RB效应发生在语义水平。