Ignatius G. Mattingly

The motor theory of speech perception revised.

A motor theory of speech perception, initially proposed to account for results of early experiments with synthetic speech, is now extensively revised to accommodate recent findings, and to relate the assumptions of the theory to those that might be made about other perceptual modes. According to the revised theory, phonetic information is perceived in a biologically distinct system, a ‘module’ specialized to detect the intended gestures of the speaker that are the basis for phonetic categories. Built into the structure of this module is the unique but lawful relationship between the gestures and the acoustic patterns in which they are variously overlapped. In consequence, the module causes perception of phonetic structure without translation from preliminary auditory impressions. Thus, it is comparable to such other modules as the one that enables an animal to localize sound. Peculiar to the phonetic module are the relation between perception and production it incorporates and the fact that it must compete with other modules for the same stimulus variations.

Speech Synthesis by Rule

Intelligible English has been synthesized in part “by rule”, with a system consisting of a computer programme and an electronic analogue synthesizer. Methods for the synthesis of the various phonetic classes and their transitions are described.

Discrimination in speech and nonspeech modes

Abstract Discrimination of second-formant transitions was measured under two conditions: when, as the only variation in two-formant patterns, these transitions were responsible for the perceived distinctions among the stop-vowel syllables [bae], [dae], and [gae]; and when, in isolation, they were heard, not as speech, but as bird-like chirps. The discrimination functions obtained with the synthetic syllables showed high peaks at phonetic boundaries and deep troughs within phonetic classes; those of the non-speech chirps did not. Reversal of the stimulus patterns, producing vowel-stop syllables in the speech context and mirror-image chirps in isolation, affected the speech and nonspeech functions differently. An additional non-speech condition, presentation of the transitions plus the second-formant steady state, yielded data similar to those obtained with the transitions in isolation. These results support the conclusion that there is a speech processor different from that for other sounds.

Chapter 1 Linguistic Awareness and Orthographic Form

Publisher Summary This chapter discusses the linguistic awareness and orthographic form. It has been widely agreed that the notion of linguistic awareness is essential for an understanding of the reading process, the acquisition of reading and reading disability. This notion is likewise essential for an understanding of the invention and dissemination of orthographies. There are really only two possible ways to write, the syllabic method and the segmental method, because only by using one of these two methods is the writer assured of being able to write any word in his language. But for an illiterate to discover either of these methods, and thus be in a position to invent writing, requires awareness of the appropriate unit of linguistic representations. Awareness of syllables, or, on the other hand, of segments, is fostered by special morphophonological properties found in those languages for which writing systems were invented, though by no means in all languages. But once it has become established, the writing system itself shapes the linguistic awareness, and even the phonology, both of those inheriting the system and of those borrowing it to transcribe some other language. Thus, in the history of writing, syllabic and segmental traditions are clearly distinguished.

Phonetic Representation and Speech Synthesis by Rule

Publisher Summary While a computer model of performance in speech production is certainly conceivable, the title of the seminar, “Speech Synthesis Programs as Models of Speech Production,” seems incorrect as a characterization of existing systems for synthesis of speech by rule. Insofar as these systems have had more than the purely pragmatic goal of converting written utterances into spoken utterances, the objective has been a more modest one, but one that is nonetheless directly related to the theme of this Symposium, “The Cognitive Representation of Speech.” The objective has been to elucidate the nature of the phonetic representation shared by speaker and hearer (and recorded, after a fashion, in a conventional phonetic transcription) and the relationship of this representation to physical events. A simulation of the mental activity underlying actual speech production would be a far more ambitious project, though one well worth attempting. Synthesis-by-rule has also, of course, considered other interesting questions: the relationship of the conventional orthography to the phonemic representation, and the relationship of the latter to phonetic representation. However, its central concern, surely, has been to relate the phonetic representation either to the articulatory or to the acoustic parameters of continuous speech.

Synthesis by Rule of Prosodic Features

Synthesis by rule of a limited set of prosodic features of Southern English has been attempted as an extension of a previously reported system for synthesis of segmental phonemes. Methods used for synthesis of intonation features, pausal features and prominence are described.

Phonetic Activity in Reading: an Experiment With Kanji

Reading seems to require phonetic short-term storage even if the writing system is logographic. A short-term memory probe paradigm was used to test the ability of Japanese subjects to recall kanji characters. When pairs of characters in a silently read list were homophonous, recall was significantly poorer than when the silently read list included no such pairs, presumably because of confusion in phonetic short-term storage. Since the kanji have no overt phonetic structure, it would seem that this kind of storage is involved in the reading of kanji not because of the particular structure of the writing system but because of the essentially linguistic nature of the task. Evidently the linguistic process cannot readily disperse with phonetic short-term storage. It is suggested, however, that the design of a writing system does not depend directly on linguistic process but rather on the partial and uneven awareness that users of the system have of certain aspects of this process.

The Speech Code and the Physiology of Language

To the physiologist who would study language in terms of the interests represented at this symposium, the most obvious linguistic processes—the selection of words to convey meaning and the arrangement of words in sentences—must seem far removed from familiar concepts and methods. Surely, he would prefer to study processes that are physiologically more accessible, but are yet linguistic. We believe that the production and perception of speech, in the narrow sense, is one such process; we suggest, therefore, that the physiologist might do well to start there. The questions we would have him ask can be put very simply: How does a speaker convert the phonetic units—the consonants and vowels—to a stream of sound? On hearing that stream, how does a listener recover the phonetic units?

Did Orthographies Evolve

According to Gelb (1963), writing has “evolved” from picture writing to logography to syllabic writing to alphabetic writing. It is argued here that this widely accepted theory of orthographic evolution does not accurately fit the historical facts, and that the variety of orthographies is better explained on linguistic grounds. Orthographies have to be productive, and they can manage this only by providing devices for transcribing the possible words in the lexicon. The very limited number of different ways in which this is accomplished in different orthographies is accounted for by the structural peculiarities of the languages that the orthographies transcribe.

Phonetic Coding of Kanji

An experiment in the recall of visually presented Japanese Kanji ideograms suggest that Kanji may, like alphabetic words, be encoded phonetically, despite their lack of phonetic structure. The experiment, based on Kintschs and Buschkes (1969) paradigm, assumed that similarity of items in a list increased errors in recall. Four lists were prepared, each containing 16 different Kanji. The first included phonetically similar pairs of characters; the second, semantically similar pairs; the third, visually similar pairs; the fourth was a control list containing no similar pairs. The subjects, 10 native speakers of Japanese, were presented with randomly ordered versions of each list, at one character per second. After a subject had seen an entire list, he was presented with a cue character selected from the list, and asked to recall the character that had been presented immediately before the cue. Confusion in recall was significantly greater for the phonetic list than for the other lists. These results stren...