Sentence Perception Research Articles

Hierarchial chunking in sentence processing

In order to evaluate a left-to-right hierarchical chunking model of sentence perception, Johnson’s Hierarchical Clustering Scheme (HCS) technique was applied to data obtained from sentence intelligibility tests. One hundred and twenty Ss listened to sentences disturbed by white noise. After each presentation they wrote down what they had heard. For each sentence, a table of conditional probabilities p(j/i) was computed, where p(j/i) is the probability that word j had been correctly identified. given correct identification of word i. This was done for all i’s and j’s from the sentence. HCS analysis of the off-diagonal submatrices for which words i precede words j (“forward conditional probabilities ”) yielded satisfactory results. Apparently there is a latent hierarchical structure to these data. The large chunks that appear from these analyses do generally correspond to major syntactic constituents. Minor constituents, however, are very often not reflected in the chunking pattern.

Laterality Factors in the Perception of Sentences Varying in Semantic Constraint

The present experiment explored the differential rôle of the left and right ears in perceiving linguistic stimuli on the level of sentences. Sentences with identical syntactic structure but varying in semantic constraint [i.e., “The farmer plowed the field” (semantically well integrated) vs “The Indian hit the table” (semantically poorly integrated)] were presented to the left or right ear of 10 right-handed subjects. Listeners were required to listen to and then immediately repeat each sentence after presentation. A special speech-noise distortion circuit developed at the Communication Sciences Laboratory of The University of Michigan [M. A. H. O'Malley and G. E. Peterson, “An Experimental Method for Prosodic Analysis,” Phonetica 15, 1–13 (1966)] was used to produce a masking signal that provided a constant signal-to-noise ratio for the entire utterance. The circuit derives the speech masking signal from the utterance itself by destroying the segmental information although maintaining the instantaneous power spectrum and pitch information of the sentence. Masking output of the circuit was delivered to both ears of a listener while the speech signal was presented to only one ear at a time. This procedure provided an opportunity to examine the effect of speech distortion on the rôle of the two ears in perceiving sentences. Preliminary results indicate significant differences in immediate recall for ears—the right ear being more efficient than the left ear. Results of this pilot study indicate that: (1) differential ear preference is a function of cerebral dominance; and (2) there are differences between the two hemispheres in processing different types of linguistic information on the sentence level.

Compressed speech: correlates of listening ability.

To identify sources of individual differences in the ability to comprehend compressed speech, a battery of tests was administered to 52 college students. The criterion variable was performance on four multiple choice tests based on the content of four recorded history passages. Subjects listened to these passages at a normal rate of approximately 175 wpm, and at three compressed rates: 250, 325, and 450 wpm. Compressed versions were prepared using the Tempo-Regulator. Certain ancillary data on intelligibility and comprehension were separately obtained from the same subjects by means of a sentence perception task. The results of a multiple regression analysis of test score data indicated that the best predictor of comprehension at high rates of compression was the Best Trend Name Test, originally developed as a measure of the ability to evaluate semantic relations -one of the components in the Guilford structure-of-intellect model of abilities. The outcome of a factor analysis of the test data supported an interpretation that the ability to rapidly match or compare concepts on the basis of semantic relatedness was a determinant of performance on the Best Trend Name Test. In an attempt to account for the significance of this ability in the comprehension of compressed speech, Guilford's concept was juxtaposed with the test or comparison operation postulated to be one stage in the “analysis-by-synthesis” model of speech perception.

Role of surface and base structure in the perception of sentences

The Ss in this study listened to groups of ten syntactically homogeneous sentences mixed with white noise. An anomalous test sentence followed each group of ten. Perception scores for the test sentences showed that changes in both surface structure and base structure can significantly disrupt perception. Changes in surface structure have the stronger effect. These findings provide evidence for the psychological reality of some abstract properties of stimuli.

Aural Perception of Sentences of Different Syntactic Structures and Lengths

Eleven sets of 16 sentences, previously rated as natural English statements, were presented to two groups of listeners. Each set of sentences differed in syntactic structure and contained four sentences of each of four lengths: 8, 10, 12, and 14 words. The listeners heard the sentences either under a fixed condition of filtering or without filtering; their task was to write as much of each sentence as possible immediately after it was presented. The sentences were scored for the number of words correctly reproduced, and differences in aural perception among the various sentence types and lengths were tested on the basis of the derived intelligibility scores. The results of the study indicated that (1) perception of the sentences was affected by their syntactic structure and varied, depending upon the grammatical complexity of the statements as specified by a generative grammar; (2) the sentences of 12 words were more easily perceived than were the sentences of the other lengths; (3) no relationship was established between the relative perception of the various sentence structures and their degree of naturalhess.

Some perceptual consequences of linguistic rules

Summary Three experiments on the auditory perception of grammatical, anomalous, and ungrammatical sentences are interpreted to mean that both syntactic and semantic rules are normally involved in the perception of sentences.

Context in the Perception of Sentences

Context in the Perception of Sentences