Transposition asymmetry in immediate serial recall

When participants confuse the position of items in immediate serial recall, they tend to recall transposed items too early rather than too late. This asymmetry of transposition errors was observed in four experiments. It increased as a function of list length, but was independent of report order, output position, cueing condition, and recall mode. The transposition asymmetry is consistent with error patterns in free recall and in regular speech production where transpositions are usually forward‐looking. The asymmetry of transposition errors is discussed in terms of models of serial memory.     

The impact of connectives on the memory for expository texts

Three experiments investigated the influence of connectives on memory for expository text. Subjects in Experiments 1 and 2 read and later recalled passages with either no connectives, temporal connectives (before/ and then), causal connectives (which caused/which enabled), or intentional connectives (in order that/so that). There was connective interference rather than connective facilitation in the studies reported here: the recall for passages without connectives was higher than the recall for passages with connectives. The results partially supported a semantic complexity hypothesis, which predicted that recall should increase with the semantic complexity of the connective (i.e. temporal < causal < intentional). In a third experiment, the semantic appropriateness of the connective was manipulated. Recall for passages without connectives was better than recall for passages with either appropriate or inappropriate connectives. In a fourth experiment, the subjects filled in missing connectives in natural expository texts. Subjects were able to select the connective category which was originally used by the author. The influence of connectives on text was discussed from the perspective of resource limitations, elaborations, and the semantic compatibility of the connectives to the text.     

Reading strategies of fast and slow readers

In three subject-paced experiments we evaluated reading patterns at the word, line, and sentence level for fast and slow readers. A moving-window method was used to collect word reading times for natural texts. At the word level, reading times of word N were influenced by features of word N-1 for fast readers but not for slow readers. The lag effect exhibited by fast readers indicates that they continue to process a word when it is no longer in view, thus limiting the notion of immediate processing. Contrary to our initial expectation that fast readers would process only a single new argument from a sentence, whereas slow readers would process several new arguments, we found that both reader groups adopted a many-argument strategy. However, fast and slow readers differed in terms of the text units (lines vs. sentences) defining the new-argument effects: Fast readers exhibited greater new-argument effects relative to lines, whereas slow readers exhibited greater new-argument effects relative to sentences. Specifically, slow readers integrated the new arguments primarily at the end of the sentence, whereas fast readers did so at line boundaries. These results are discussed in terms of a buffer-and-integrate model of reading comprehension.     

Forward and backward recall: different response time patterns,same retrieval order

How do people retrieve information in forward and backward recall? To address this issue, we examined response times in directional recall as a function of serial position and list length. Participants memorized lists of four to six words and entered responses at the keyboard. Recall direction was postcued. Response times exhibited asymmetry in terms of direction. In forward recall, response times peaked at the first position, leveling off for subsequent positions. Response times were slower in backward recall than in forward recall and exhibited an inverse U-shaped function with an initial slowdown followed by a continuous speedup. These asymmetries have implications for theoretical models of retrieval in serial recall, including temporal-code, rule-based, and network models. The response time pattern suggests that forward recall proceeds in equal steps across positions, whereas backward recall involves repeated covert cycles of forward recall. Thus, retrieval in both directions involves a forward search.     

Processing of new arguments at clause boundaries

In a subject-paced reading-time study, we examined the processing of new arguments at clause boundaries. Word reading times increased with the cumulative number of new-argument nouns at clause boundaries (as well as at sentence boundaries). New-argument nouns had a greater impact at clause boundaries than at nonboundary locations. In accordance with a buffer-integrate-purge model of reading (see Jarvella, 1979), the increase of reading times at boundaries was attributed to the integration of new information from the current sentence with prior information in the text representation. The increase at nonboundary locations was attributed to the growing load of buffering the new information. Reading times at clause boundaries were influenced to a greater extent by text-level integration than by such sentence-level processes as organization of words into clauses and linking of clauses within a sentence. The new contribution of this study was that it showed that clause boundaries provide an opportunity not only for sentence-level processes, but also for a text-level process, namely, the integration of text-new information with the growing text representation.     

Transposition asymmetry in immediate serial recall

When participants confuse the position of items in immediate serial recall, they tend to recall transposed items too early rather than too late. This asymmetry of transposition errors was observed in four experiments. It increased as a function of list length, but was independent of report order, output position, cueing condition, and recall mode. The transposition asymmetry is consistent with error patterns in free recall and in regular speech production where transpositions are usually forward-looking. The asymmetry of transposition errors is discussed in terms of models of serial memory.