Aspiration of Voiceless Affricate in English

Question

My question is about the voiceless affricate /tʃ/ ( CHair, maTCH, baTCH, strucTure) as it is used in ENGLISH:

English has two affricates: the /tʃ/ in "chair" and the /dʒ/ in "jar".

The /tʃ/ is a voiceless affricate.

My question is, since this is voiceless, does it follow the same rules of aspiration as voiceless stops in English? That is, like /p/, /t/, /k/?

/p/, /t/ and /k/ are usually aspirated when at the beginning of a word(except when preceded by s in the same syllable) and at the onset of a stressed syllable. Does the same hold true for /tʃ/?

So the /tʃ/ sound in "chair" is aspirated whereas in "catch", it is not? Or is it, but a little less in magnitude?

Answer 1

The affricate /tʃ/ does not behave differently from the stops /p t k/ w.r.t. aspiration. The relevant contexts for aspiration are bit more complicated and are best stated in terms of foot-initial (aspirated) vs. foot-non-initial positions, with some provision for C#ˈV contexts where there is no aspiration ("watch Oscar; set Oscar; stop Oscar"). In final position ("stop, hit, watch") there is often some degree of voiceless turbulence which is usually relegated to the category of release, not aspiration, but the physical substance of that noise is indistinguishable in type from that of aspirates.

The duration of the noise between the release of the stop is taken be the phonetic correlate of aspiration. That duration depends on contextual factors, for example /t/ versus /k/, being utterance-initial vs. medial, being pre-vocalic vs. pre-consonantal and soon. Given that, there are dozens of "degrees" of aspiration (defined as voice onset time or lag to complete silence in case the consonant is not followed by a voiced segment). As a statement of phonetics that's not a problem, but as a phonological generalization, that flies in the face of the premise that English just has a distinction "aspirated" and "unaspirated", which is furthermore rule-governed. So the ultimate answer depends on establishing what kind of rule governs aspiration. If it is due to a "rule" (equation) of continuous phonetic implementation, the aspirations of "pit" vs. "apex" are not the same – phonetic equations do not yield categories. In that case, asking if they are "the same" is kind of meaningless. If it is a categorial albeit non-neutralizing phonological rule, then all of the evidence says that the rule aspirating /tʃ/ isn't different from the rule aspirating /p,t,k/.

Answer 2

The fundamental (and contrastive) difference between phonologically aspirated stops and phonological affricates is the nature of the release. Aspiration is turbulent noise whose source is the glottis, thus aspiration has formant structure similar to the following sonorant, and a resulting low COG, and diffuse spectrum. Frication noise has its source in the oral constriction, and thus does not have the lower formants (because of de-coupling of the back constrictions) or diffuse quality, especially for sibilant affricates, which have greater amplitude noise in higher frequencies reflecting the front cavity constriction. In English, lexical stress applies, as well as focal stress, and the largest correlates of stress are amplitude, duration, and F0. Stress applies uniformly across the syllable, governed by the expandibility/compressibility of the segments in that syllable, so that vowels experience the largest effects because they can vary more. However, consonants also experience stress-based strengthening or reduction. While aspirated stops may have less aspiration in unstressed positions, being reduced in amplitude and duration, phonological affricates do not lose their frication portion in the same unstressed positions, even if their amplitude and duration may also be reduced. So, no, affricates do not follow a similar rule as aspirated stops, insofar as their loss of frication would result in a manner change. But they are governed by the same rules of stress reduction, so that in an unstressed syllable, they may have shorter duration and less amplitude, but the phonetic outcomes of a similar rule on different phonological targets are completely different.