Published online by Cambridge University Press: 22 April 2026
Thanks to the means available for speech analysis and synthesis, there is at present considerable knowledge of the acoustic differences whereby we distinguish among the various sounds of English. We know, for example, what the more important acoustic cues are by which listeners decide whether a stop sound is velar, alveolar, or bilabial. In the case of the stops, moreover, we know by and large what the acoustic differences are between the members of each of the homorganic pairs p-b, t-d and k-g (in other words, between the so-called voiced and voiceless stops). But we are as yet unable to state with any exactness the extent to which each of these acoustic differences separately affects our ability to discriminate between the two kinds of stops. Moreover, where progress has been made in this direction it is certainly in some measure due to the fact that we have dealt chiefly with the stops in those positions where the voiced-voiceless contrast is phonetically maximal, that is, in initial and final positions.
1 Delattre, Liberaman, and Cooper, Acoustic loci and transitional cues for consonants, Jour. Acoust. Soc. Am. 27.769-73 (1955).
The experiments reported here were all conducted at the Haskins Laboratories in New York, where I benefited from discussions with more of my colleagues there than I have space to list here. The work was supported in part by the Carnegie Corporation of New York and in part by the Department of Defense in connection with Contract DA49-170-sc-1642.
2 See, for example, Cooper, Delattre, Liberman, Borst, and Gerstman, Some experiments on the perception of synthetic speech sounds, Jour. Acoust. Soc. Am. 24. 597-606 (1952).
3 The nonlinguist is often confused, and legitimately so, by the two different uses to which the term voice and its derivatives are put. In one usage the expression voiced-voiceless distinction is the name of each of the phonemic contrasts /b/-/p/, /d/-/t/, /j/-/c/, /v/-/f/, /z/-/s/, etc., where the first of each pair of opposed phonemes is called voiced and the second voiceless. In another usage, a speech sound is called voiced only if glottal tone is involved in its production; in the absence of audible glottal tone a sound is voiceless. That these two usages are not synonymous is shown, for example, by experiments whereby one and the same acoustic segment may be perceived as voiced or voiceless depending on the context. See P. Denes, Effect of duration on the perception of voicing, Jour. Acoust. Soc. Am. 27.761-764 (1955).
4 See reference cited in fn. 3.
5 Phonetic descriptions of the allophones of /p/ and /b/ proper to this position agree that these sounds differ in the matter of voicing. Trager and Smith, An outline of English structure 32, describe the p in this position as both voiceless and fortis relative to the b; other phoneticians, however, are not consistent in reporting a difference of the kind labeled by the terms fortis and lenis. Many do not apply these terms at all; and Heffner, General phonetics 132, reports without comment that Jespersen, Lehrbuch 107, described this p as lenis.
6 By its very definition, closure duration can be measured only when the stop is medial (and final, if audibly released). In a series of preliminary measurements, overlapping values of closure duration were obtained for the two stops (1) preceding main stress (abúse, repúte; rebél, repél); (2) in gemination (grab bag, cap pistol); (3) between unstressed vowels (sýllabus, éntropy).
7 In my dialect rupee has weak stress on the second syllable.
8 Each word has a duration of about 500 msecs.
9 Closure durations for the same words when included in longer utterances are generally shorter, but the same relation holds between p and b.
10 For ten minimal and near-minimal pairs the average difference was about 25 msecs. But despite the small size of the sample there were two cases where values clearly over-. lapped.
11 To eliminate any possibility that the low-frequency energy occurring within the p closure interval (Fig. la) might be a factor in any shift in the distribution of judgments, the portion of tape corresponding to this interval was first replaced by blank tape.
12 There is some basis for the expectation that the post-closure cues will outweigh the others. For example, in a tape-splicing experiment in which the words ragged and rabid were cut and recombined as rag-bid and rab-ged, the intervocalic stops reported by listeners were b and g respectively.
13 Glottal tone was introduced into the closure interval with generally negative effect, but certain problems encountered in this operation have as yet not been adequately enough resolved to permit full-scale testing.