Three theories of English plosives: the myth of deaspiration

Abstract: There are no unaspirated voiceless plosives in English, only unaspirated voiced plosives and aspirated voiceless plosives. All voiceless plosives in English are aspirated.

The standard theory we all learnt as undergrads tells us that English prevocalic voiceless plosives come in two flavors, aspirated when syllable initial and unaspirated otherwise. The standard theory we all learnt as undergrads can’t be right: it entails that voiced plosives cannot occur in syllable-second position. Such a prohibition a) has no justification and b) is absurd — the unaspirated voiceless plosives that the standard allophonic theory predicts do occur in that position sound just like voiced plosives. I’ve got two replacement theories to offer, neither perfect but both better than the standard.

The basic facts are these: a) English voiced plosives are unaspirated throughout the language; b) there is therefore no way to distinguish unaspirated voiceless plosives from voiced plosives in English — there is no difference in English between the sequences [skIl]; and [sgIl]: the word “skill” may as well be [s]+”gill” as [s]+”kill” with deaspiriation of the “k.” The former analysis is simpler — it requires no stipulation about deaspiration of /k/. English speakers’ bias towards the latter, more complex, analysis, reflects mere spelling convention. Yet the more complex, spelling-biased theory is the standard linguistics textbook theory.

Here’s the problem. The standard — spelling biased — theory has two rules:
1) voiceless aspirated prevocalic initials (e.g., “pill,” “till” and “kill”) are deaspirated when preceded by [s];
2) [b], [d] and [g] never occur preceded by [s] and followed by a vowel (they never occur in the deaspiration environment of 1).

On this theory, the phonemes /p/, /t/ and /k/ have two rule-governed allophonic pronunciations in prevocalic position: aspirated (syllable initial) and unaspirated (preceded by [s]).

Rule (1) is pedagogically useful for teaching the notion of the phoneme, complementary distribution and the allophone to students in their first exposure to linguistics. That may be why it persists. It has the unfortunate consequence of interpreting all plosives wedged between [s] and a vowel as deaspirated voiceless consonants, leaving no room for voiced plosives in that position. And there is no natural justification for a prohibition on voiced plosives in this position. Why shouldn’t the phonemes /b/, /d/ and /g/ occur there? After all, unaspirated /p/, /t/ and /k/ occur there and they sound exactly like [b], [d] and [g]. The prohibition is patently counterfactual: it says voiced plosives can’t occur in exactly the environment in which they do occur. This is not merely unjustifiable; it is preposterous.

A Simplest Answer

Consider a simpler theory without allophones: /p/, /t/ and /k/ are pronounced as aspirated [p], [t] and [k] wherever they are fully articulated; they are prohibited from occurring in the syllable second position by a double aspiration rule that is generally true of English consonant sequences. [fs] in Russian, not in English. There are only four English words with syllable initial double sequential aspirations and they are all Greek borrowings: sphere, sphinx, sphincter and sphalerite. I don’t know anyone but me who pronounces the labiodental initial of “phthisic” — but, then, I don’t know anyone else who pronounces the word at all. Even the lexicographers have dropped the [f]. Sequential onset aspiration is a foreign imposition on the natural phonology of English. This holds even in coda position: the aspirated /k/ of “ask” loses its aspiration when followed by a vowel (”ask anyone”).

This theory is considerably simpler and better than the standard. There’s only one rule, that rule has some natural justification and it does not lead to the standard’s absurdity. The rule: no voiceless plosives following [s] and preceding a vowel.

On this theory, in the onset /g/ is [g] and /k/ is [kh]. No allophones, no need, even, for phonemes. What you hear is what you get. It’s all phonetic, not phonological. If you want a justification for phonemes and allophones and complementary distribution, you have to look at environments where the plosives are not fully articulated. For example, the aspiration of /k/ is absorbed by [s] in “text” ([tEkst]); the /t/ loses is aspiration when followed by a vowel (”textile” — USA pronunciation, not where it’s divided into “tex”+”tile”) and the /t/ reduces to zero when followed by a consonant in “textbook.” These should suffice to serve the pedagogical function of introducing the notion of complimentary distribution, the phoneme and allophony.

A Problem for the Simple One-Rule Theory

There’s a serious problem with this theory. Consider “discourage.” On the standard theory, “courage” is underlyingly represented with an initial /k/ phoneme. Under the standard theory, allophonic variation in /k/ is induced by the presence of a syllable initial [s]. So, under the standard theory, “discourage” and “courage” are both represented with an underlying /k/ phoneme.

But under the simple theory, /k/ is prohibited from occurring after the [s]. These two words have different underlying representations, “courage” with a /k/ and “discourage” with a /g/.

Now, it does happen that morphology sometimes violates phonology. For example, the plural of “text” is required under English morphology to violate the syllable structure and consonant cluster constraints of English. Outside of such plurals and 3rd person singulars, English doesn’t permit consonant clusters of such length as [ksts] and any such concatenation of sounds would require division into two syllables.

An Unwelcome Solution

But the title of this entry refers to “Three Theories” so we’ve still got one to go.

Here’s a hybrid theory. Voiceless plosives become voiced when preceded by [s] and followed by a vowel. Here’s how this theory works:

The voiceless plosives share an allophone with the voiced plosives. /k/ and /g/, for example, are two distinct phonemes, but /k/ is sometimes pronounced the same as the /g/ phoneme is pronounced. So the underlying representation of “courage” and “discourage” both have the /k/, but in “courage” it is pronounced as an aspirated voiceless plosive, in “discourage” it is pronounced as a voiced plosive.

On this theory, there is no prohibition rule. Voiced plosives may appear before a vowel following [s]. So may voiceless plosives, but only in their unaspirated, voiced form.

Here is the unwelcome consequence: there is no way to determine on this theory whether “skill” has an underlying /k/ or an underlying /g/. In addition, the allophony requires a feature-changing rule and who really likes a feature-changing rule? Nobody talks to them at parties.


Cast in a more contemporary theory that regards features rather than phonemes, the three theories look like this:
Standard — plosives have place and voice features, but pick up aspiration by rule.
Simple — plosives are specified for place, voice and (redundantly) aspiration.
Unwelcome — plosives are specified for place and voice, but change voicing by rule.

The standard theory entails a preposterous prohibition. The simple theory entails changing underlying representations of identical morphemes. The unwelcome theory has a feature-changing rule. Take your pick.

An afterthought

The simplest analysis of “skill” under the unwelcome theory is [s]+/g/ — it requires fewer rules. But the theory already contains the feature-changing rule, so the more complex analysis is not ruled out even for economy’s sake. One may, however, assume that, ceteris paribus, a [g] is a /g/.


10 Responses to “Three theories of English plosives: the myth of deaspiration”

  1. Timay Says:

    can you please make a summary about this?:) I need it for our thesis. THANK YOU!! we liked your idea that’s why were considering it part of our study.

  2. Keri Says:

    Awesome post; very interesting. Thanks for writing =)

  3. Charley Aspacio Says:

    You actually make it appear so easy together with your presentation however I in finding this matter to be actually something which I feel I’d never understand. It kind of feels too complicated and extremely wide for me. I’m looking ahead in your next put up, I will attempt to get the hold of it!

  4. Peter Cyrus Says:

    Isn’t the S in discourage at the end of the previous syllable? So the k shouldn’t be deaspirated (but it is!).

    Are p t k ch also aspirated at the beginning of unstressed syllables? I thought not.

    Final p t k ch are always unreleased anyway, so aspiration is moot, right?

    • rob Says:

      The syllabification may explain why the /t/ in “distasteful” is not deaspirated. You’d expect the same with “discourage.” That’s what’s so interesting about “discourage.” Or maybe it’s a verb that self-refers to its phonology.
      Aspiration occurs in initials of unstressed syllables: “peculiar” “telepathy” “collateral” all can have aspirated initials, though obviously not as heavy as in a stressed onset. Aspiration in an unstressed onset might depend on the degree of destressing. If the unstressed syllable is reduced to zero (as if it were Greek pn or pt) it’s as if there’s no space left for aspiration, but that’s no doubt because it’s unreleased. Finals are an odd lot. I and my new neighbor aspirate /k/ final. He’s half my age and we’re from opposite coasts of the continent. But we’re both middle class, educated, white. Woody Allen also aspirates the final /k/. But I’m pretty sure not everyone does, otherwise I wouldn’t notice the aspiration when I hear it.
      If you forget about deaspiration and think of it as voicing, the data all make much more sense. Except “distasteful”/”discourage.” They’re a thorn in any theory.

      • Peter Cyrus Says:

        I don’t see too many problems with your theory. Disgust and discussed are perfect homonyms, with an unaspirated plosive, and the best explanation is that the plosive is a g, and that we have a rule that transforms unvoiced plosives into voiced ones after s. When the s is in a previous syllable, the rule may or may not apply, but that’s a minor quibble.

  5. rob Says:

    Which theory? I’ve got two. (-:

    Why imagine that there is a /k/ in “discussed” and not a /g/? Taking it at face value, /g/ would be simpler. Similarly with “skill” — why think it’s got a transformed /k/ rather than just a /g/?

    In the case of “discourage,” speakers have a sense that the word is directly related to “courage” which they know begins with [k]. But speakers don’t relate “discussion” to “cuss” (they are not historically related either, but that shouldn’t matter).

    That’s why “dis/courage” is important. It’s a case in which one can tell whether we’re dealing with transforms or not. The fact that the transformation is not regular (“distasteful” has a [t] not a [d] at the start of the second syllable) is then all the more troubling. The choice of best explanation, rather than simplest explanation, depends entirely on “discourage.”

    It can’t be the coarticulation of [s] and [t]. The [s] in “dispepsia,” as in “distasteful,” also fails to voice the plosive.

  6. Peter Cyrus Says:

    By the way, I incorporated your work into mine :

    • rob Says:

      Fascinating. I am particularly interested in the ternary logic, and sympathetic towards the project. I wrote an entry on the blog about ternary logic in relation to Lukasiewicz’ propositions about the future: For presupposition failures and maybe liar paradoxes trivalence are clearly useful, since bivalent mechanisms are not natural (Russell’s solution to the former, Kripke’s solution to the latter) or unsatisfactory (Russell’s solution to the latter). For epistemic modalities, I’m not sure, but there is a place for three options in
      I think it’s necessary
      I don’t think it’s necessary (meaning: I think it’s not necessary)
      I don’t think it’s necessary (meaning: I haven’t formed any judgment on it’s necessity).

      But the binary modal logic is pretty straighforward for it:

      Regarding the phonetic script, have you looked at Korean Hangul? Some aspects of it are visually universal, at least for humans.

      For a phonetic script, the theoretical phonology is irrelevant. But notice that English spelling manages to preserve the spelling shape of “courage” in “discourage,” which, for better or worse, cannot be done in a phonetic script. There is a trade-off with any phonetic script that language history is lost and the morphology can be lost as well, where the phonology interacts with it. It would be interesting to try to create a script that indicates the phonology, the morphology and the phonetic form, maybe with subscripts. To some extent, French and Greek accents have this function.

      How do you plan to promote schwa over IPA? (Feel free to contact me personally by e-mail at

  7. Judd Emery Says:

    I came across this while thinking about a student of mine who both devoices voiced plosives and deaspirates voiceless plosives in word-initial position. Seen through the prism of disordered phonology or articulation it seems very clear that deafrication is quite possible, in the sense that this student’s devoiced /d/ and deaffricated /t/, while level with each other, are distinguishable from the other two sounds. The sound is quite literally between the two – a plosive with no prevoicing and no aspiration.

    But the existence of this intermediary sound in normal English doesn’t tell me that the English speaker is unable to make a phonemic distinction without a spelling system. “Faddish” and “fattish” are homophones when spoken quickly with a medial flap but can be quickly disambiguated by pronouncing them, also correctly, with medial /t/ or /d/. In the same way I could clarify the difference between “disgorge” and “discourage” (not homophones) by prevoicing the /g/ and aspirating the /k/, while still pronouncing them correctly. Slightly more awkward but still within the rules, I believe, would be the disambiguation of “discussed” and “disgust” by the same method. One more evidence that the English-speaker hears the deaspirated devoiced plosive as distinct from the voiced plosive is that if I add aspiration to a word like “steam” it remains intelligible to everyone, whereas if I dramatically prevoice the plosive (“sdeam”) it becomes confusing to most listeners.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: