The purpose of this study was to investigate the perceptual effects of three types of mispronunciations, affecting the voicing of obstruents, the front-back dimension of stressed vowels, and the stress pattern of words. Subjects were instructed to shadow prose passages containing mispronunciations. Words containing voicing mispronunciations typically were repeated in their original form; words with vowel and stress pattern mispronunciation led instead to other response types. Recent studies of the perception of fluent speech suggest that listeners use both phonetic ("bottom up") and contextual ("top down") information in word recognition. Much of these data are derived from an experimental paradigm in which subjects are asked to respond in various ways to speech containing mispronounced words. This research paradigm was first introduced by Bagley (1900) and reintroduced more recently by Cole (1973) (see also Cole & Rudnicky, 1983). Considerable research has shown that "top down" and "bottom up" information both contribute to word recognition. For example, Marslen-Wilson and Welsh (1978) found that subjects shadowing prose tended to correct mispronunciations without hesitation, particularly if the mispronunciations were relatively minor (changes of a single distinctive feature), if the mispronounced words were highly predictable from context, and if the mispronunciation occurred late rather than early in a word. Cole, Jakimik, and Cooper (1980) found that reaction time in detecting mispronunciations was influenced by the contextually suggested segmentation of a target word. For example, mispronunciations of "drift" were detected more quickly in "snowdrift," where they occurred in the second syllable of a single word, than in "snow drift," where they occurred in the second of two monosyllabic words. In addition, Cole and Jakimik (1978) found that mispronunciation detection was faster when target words were predictable from context-either from words explicitly present or from the general theme of a passage. The relative usefulness to a listener, or the salience, of various types of phonetic information has received somewhat less attention, even though researchers have hypothesized that not all phonetic information

The purpose of the study was to examine the perceptual effects of altering lexical stress during word recognition. A detection task was utilized to measure subjects' speed of response to target phonemes preceded by two-syllable homograph and nonhomograph words. These experimental words were pronounced with correct/incorrect lexical stress. When a nonhomograph was misstressed, a nonsense word was the result. When a homograph was misstressed, another English word resulted. Examination of subjects' speeds of response to target phonemes preceded by correctly/incorrectly stressed nonhomograph words indicated slower speed of response when the stimulus word was stressed incorrectly; word recognition appeared to be affected when a subject heard a misstressed nonhomograph (a nonsense word). However, subjects' speed of response to target phonemes following correctly/incorrectly stressed homographs were similar. Mispronounced homographs did not appear to impede word recognition. It was possible that when subjects heard a misstressed homograph (another English word), they relied more upon the prosodic contour for word recognition.Although it is generally known that prosodic information is essential during word recognition, it is not clearly understood how prosodic cues such as lexical stress and sentence accent operate in speech perception. The importance of stress in speech perception has been reviewed quite extensively by Lehiste (1970). Lehiste states that stress is perceived largely in terms of change in the speaker's fundamental frequency, with duration of speech segments playing a smaller role. According to Lehiste, listeners react to a speaker's vocal effort in perception of stress, as opposed to actual changes in vocal intensity.It has been suggested that the mental lexicon is organized, in part, by the stress characteristics of words (Brown & McNeill, 1966;Fay & Cutler, 1977). Fay and Cutler examined speech errors (malapropisms), involving the substitution of one word for another with a different meaning, for example, "equivocal" for "equivalent." The stress pattern was the same for the intended word and the substitute 98% of the time. Brown and McNeill, in exploring the "tip-of-the-tongue" phenomenon, discovered that when subjects were in a tip-of-the-tongue state, the words they offered for the intended target shared lexical stress 74 % of the time.Several studies have attempted to detail the importance of lexical stress during word recognition. Cutler and Foss (1978) observed that reaction times to a phoneme target were faster when the target-bearing word was stressed than when the target-bearing word was not stressed, regardless of the form class of the word, that is, whether the word was a content or function word. Cutler and Foss suggested that the most important element of a sentence usually receives stress, and that the most important element may be determined more by the semantics andpragmatics of a sentence than by the form class of the item. Likewise, Cole andJakimik (1980) reported that the number of lis...

No abstract

Lingual differential sensitivity for intensity was measured by determining difference limens (DLs) at three different frequencies and five sensation levels. DLs were obtained from five normal adult subjects. The results indicated a relationship between the DL values (expressed as a Weber fraction: AliI) and frequency. Overall, the data obtained indicate that the lingual sensory system can provide DL functions that are compatible with those of other tactile sensory systems, as well as those derived from other sensory modalities, such as audition.The measurement of differential sensitivity of both the auditory and tactile sensory systems has previously been examined in order to determine each system's ability to resolve slight differences between perceived stimuli differing in intensity. The difference limen (DL) is the unit typically employed as an index to express the differential sensitivity of a sensory system. Harris (1963) and Riesz (1928) examined the differential sensitivity of the human auditory system and determined that the DL or "just noticeable difference" Gnd) for intensity is approximately .50 dB. The results of these investigations also showed that the ability of the auditory system to detect .50-dB changes in intensity appears to be relatively independent of both sensation level and frequency, except for sensation levels near threshold.Differential sensitivity for intensity of the tactile sensory system yields somewhat more varied results than those found for the auditory sensory system. The size of the DL (at moderate to high intensities) as measured by various investigators has ranged from .30 (Sherrick, 1950), to .11 (Schiller, 1953), to .05 (Knudsen, 1928. These values were obtained by using the Weber fraction (M/I). Craig (1972) suggests that the discrepancies in the reported DL values for tactile sensitivity at moderate to high intensities may be due to differences in the various experimental methodolOgies. Even though these DL variations exist, Craig (1972) indicates that two generalizations can be made in relation to the factors affecting tactile DLs. First, at low intensities, M/I is large and decreases as intensity increases. Second, DLs for the sense of touch appear to be independent of frequency.The existing tactile DL information has been obtained primarily from the hand and index fmger. The purpose of the present study was to examine the differential vibrotactile sensitivity of the human tongue in order toThe authors would like to acknowledge the contributions of Neal Sloane to the development of this study.

A simplified modification in lingual vibrotactile instrumentation involving elimination of the tongue clamping procedure is presented. Reasons for this modification and the advantages are discussed in terms of facilitating clinical testing.