Ellipsis understood in its most capacious sense as an omission of linguistic material wherever it occurs includes phonetic phenomena such as syncope and voicelessness (= absence of voicing) in vowels. Typically, such ellipses occur in the so-called ELLIPTIC CODE of any language, by contrast with the EXPLICIT CODE, wherein the full variety of the relevant material does appear. Often the elliptic code version of a word is generalized as a matter of linguistic change and renders the explicit code version antiquated, then displaces it from the language altogether. Thus a word like listen––as the orthography, which still reflects an older period in the history of English, indicates––has two vowels but is typically pronounced either without the second vowel, viz. [ˈlɪsn], where the final consonant is syllabified, or with a schwa before the [n], i. e., [ˈlɪsən].
The “explanation” of phenomena such as the syncopation of the second vowel of listen may seem to be purely phonetic, i. e., to be couched in terms of the phonetic properties of the consonant involved and that of its context. Accordingly, the dropping of the sound /t/ here would be ascribed to (1) its definition as a “voiceless” (properly, tense) stop; and (2) the presence of the preceding (immediately contiguous) sound /s/ (a “voiceless” continuant) before an /n/ (a nasal continuant). This garden-variety appeal to phonetic context (both simultaneous and sequential), however, obscures the fact that phenomena of this kind have a PHONOLOGICAL FUNCTION, which has nothing to do with economy of effort or other such physical explanantia that have traditionally seduced linguists.
Phonological implementation rules (as they are called) make iconic reference to the distinctive (diacritic) feature values that constitute phonemes in the sound system of every language; and indexical reference to the sequential context in which phonemes occur in speech. Thus, in an item like listen (or whistle, for that matter), the fact of syncope in this context is a sign that makes reference to both the constitution of the sound syncopated and to the sounds of the context in which the syncope occurs. It has, in other words, essentially to do with semeiosis––with phonology as semeiotic–– and only secondarily with physical (= phonetic) reality.
Syncope is routinely aligned as a form of simplification with other linguistic phenomena where a sound is dropped or a feature elided. Accordingly, one should regard the “omission” of voicing in vowels in definable contexts as typologically homogeneous, hence an example of simplification as well. Thus in English, secondarily stressed or unstressed vowels in the context of immediately following nasals routinely appear as voiceless in the elliptic code. An example is the way the NPR reporters/hosts Eleanor Beardsley and Ira Glass pronounce the initial vowel of the words NPR and American (of “This American Life”), respectively––Beardsley with a voiceless [e] and Glass with a voiceless [ə] (schwa). The indexical function of vocalic voicelessness is triggered by (refers to) the voiced character of the neighboring nasals (/n/ and /m/, resp.), which (incidentally) belies the knee-jerk notion that this is a kind of phonetic “assimilation.”