David T. Hakes

Effects of prior context upon lexical access during sentence comprehension

Abstract The effects of disambiguating prior contexts upon the processing of lexical ambiguities in sentences were investigated. Subjects listened to lexically ambiguous (or unambiguous control) sentences under one of three conditions: a neutral prior context, a disambiguating context occurring immediately prior to the ambiguity, or a disambiguating context occurring in a prior sentence. Subjects monitored for phonemes which occurred immediately after each ambiguous or control word. In the neutral context condition, reaction times were significantly longer following ambiguous words than following their controls. For both the immediate and distant context conditions, the ambiguity-control contrast was not significant The results were interpreted as support for a model of sentence processing in which a prior disambiguating context serves to restrict access to readings for an ambiguous word.

Understanding sentences with relative clauses

Sentences containing self-embedded relative clauses are generally believed to be difficult to understand because such clauses interrupt the clauses in which they are embedded. However, the experiments that purport to have demonstrated this have confounded the self-embedded or fight-branching location of the relative clauses with their internal structure, comparing self-embedded object relatives with right-branching subject relatives. In order to break this confounding, Experiment I compared the comprehension difficulty of self-embedded and right-branching object relative clauses on two measures of comprehension difficulty. Experiment II made the same self-embedded vs. fight-branching comparison for subject relative clauses. The results of both experiments consistently failed to support the interruption hypothesis.

Decision processes during sentence comprehension: Effects of surface structure reconsidered

Previous experiments have yielded conflicting results concerning the utility of relative pronouns as cues to the deep structure underlying self-embedded (SE) sentences. The present study investigated the usefulness of this cue. At the same time, it compared the relative sensitivity of two measures of sentence-comprehension difficulty: paraphrasing and phoneme monitoring. Forty Ss heard SE sentences, one-half of which had the relative pronouns deleted. The presence or absence of the relative pronouns in a particular sentence was counterbalanced across groups. The Ss also heard right-branching sentences. It was found that presence of the relative pronouns led to significantly faster phoneme monitoring times and to marginally better paraphrasing (i.e., to better comprehension) in the SE sentences. It was concluded that relative pronouns are effective cues and, further, that the phoneme monitoring technique is a better index of comprehension difficulty than the paraphrasing technique. Reasons for the previous experimental discrepancies were discussed.

Sentence comprehension and relative pronouns

Doubly self-embedded sentences, differing only in whether the relative pronouns were present or deleted, were presented to two groups of Ss who were required to respond to a word beginning with a particular letter in each sentence and to paraphrase the sentence. The results of both the phoneme monitor task and the paraphrase task indicated that comprehension was better when the relative pronouns were present than when they were deleted. These results are consistent with earlier results for the paraphrase task but are not consistent with earlier results for the phoneme monitor task. Possible reasons for the inconsistency were considered. It was concluded that the phoneme monitor task does reflect comprehension difficulty in a manner consistent with the paraphrase task and that deleting the relative pronouns does make comprehending self-embedded sentences more difficult.

Effects of reducing complement constructions on sentence comprehension

Two experiments were performed testing the hypothesis that deletion of an optional cue to sentence structure increases comprehension difficulty. Sentences were used that contained that -complement constructions where the complementizer was either present or absent. Both experiments used both paraphrasing and phoneme monitoring tasks. The monitoring data for both strongly supported the hypothesis; the paraphrasing data did so weakly for the second experiment.

Does verb structure affect sentence comprehension

Fodor, Garrett, and Bever have reported two experiments suggesting that sentences containing complex verbs are more difficult to comprehend than ones containing simple verbs. However, the tasks they used, paraphrasing and anagram solving, reflect comprehension difficulty only very indirectly. Both of the present experiments compared the difficulty of sentences containing simple and complex verbs using the paraphrasing task and also a task that assesses on-line comprehension difficulty, phoneme monitoring. The paraphrasing task yielded the predicted effect in Experiment 1 but not in Experiment 2. The phoneme monitoring task failed to yield the effect in either experiment. Two alternative accounts were suggested for the present data as well as those of Fodor et al.

Some psychological effects of violating linguistic rules

Chomsky has proposed that violation of linguistic rules affects the acceptability of the resulting sentence and, further, that the amount of loss of acceptability is a function of the generality and priority of the rule violated. To test this prediction, sentences embodying three types of rule violations, varying in generality, were constructed. The three types of rule-violation sentences, together with the normal sentences from which they had been derived, were studied in three tasks: (a) an acceptability rating task; (b) a free-recall learning task; and (c) a paraphrase task. In the rating task, the sentence types were ordered as predicted. In the learning task, however, the learning ease of two of the sentence types was the reverse of that predicted. Analysis of the strategies adopted in paraphrasing the sentences suggested an interpretation of this result.

Mediated transfer in a four-stage, stimulus-equivalence paradigm

The four-stage, stimulus-equivalence mediation paradigm was investigated in a paired-associate learning experiment. The first three stages were presented as self-paced PA learning tasks, and the fourth stage was presented as a matching task. All lists were composed of 6 pairs of 2-syllable nouns. Recall-practice trials were given on the first list following second-list learning, and on all three PA lists following third-list learning. The results indicated a significant amount of mediated transfer on the matching task and suggested that S s who mediate on any pair mediate on all pairs.

Serial learning in a continuous serial list

Learning of a continuous serial list with no intertriai interval (ITI) and no starting symbol was compared with learning of a standard serial list. As in a previous study by Lippman & Denny (1964), it was found that the continuous list was learned significantly more slowly and yielded a flattened serial position curve. There was a marked tendency for Ss in the continuous-list condition to respond correctly most often to the item which appeared second. When the item on which each S in this condition performed best was placed in the first serial position, a typical serial position curve was reinstated with some diminution of the recency effect. It was concluded that in the absence of explicit serial position cues Ss tend to provide their own.

Grammatical frames as stimuli in a four-stage, stimulus-equivalence paradigm

The four-stage, stimulus-equivalence paradigm was studied using sentence frames as stimuli and low meaningfulness paralogs as responses. The first three lists were presented as self-paced PA tasks, and the fourth, or test, list was presented as a matching task. In accord with the results of a previous experiment, highly significant mediated transfer was obtained. Unlike the earlier experiment, however, there was no evidence that mediation occurred as an all-or none phenomenon for all pairs in the test list. The difference in results was attributed to slower learning and poorer retention of the materials used in the present experiment.