Aspiration is a secondary feature on specified obstruents, based on the language of use. The release of these obstruents is accompanied by a burst of air, which delays the onset of the vowel. This is referred to as ‘voice onset time’ (VOT).
In languages such as Salasaka Quichua, an aspirated /ph/ as opposed to an unaspirated /p/ can change the meaning of the word.
[pakak] ‘fear’ vs. [phakak] ‘one hundred’
However, this is not so in Standard American English (SAE), in which only voiceless stops can be aspirated in specific phonological environments. Furthermore, speakers of SAE are not aware of the difference unless they take a linguistics course! This means that the speaker believes s/he is always producing /t/, when, in fact, there are certain words in which /th/ is actually articulated. We say that /th/ is an allophone, or variation of /t/. And there are rules that constrain where /th/ can surface.
In SAE, only voiceless stops /p, t, k/ can be aspirated, and only in two phonological environments. First, all voiceless stops are aspirated in word initial position (2) a and b. The second environment is in the onset position of stressed syllables (2) c.
a [khæt] ‘cat’
b [khɹip] ‘creep’
c [əphɪɹ] for ‘appear’ or [pətheɪtoʊ] for ‘potato’
In all other phonological environments, /t/ will not be aspirated. For example, when a voiceless stop is the second consonant in a cluster, (in C2 position) then it is not aspirated (3).
Likewise, /t/ is not aspirated in word final position.