The grapheme X can sometimes be pronounced with a two-phoneme sequence, such as the following:

  • /ks/ - taxi
  • /gz/ - exact
  • /kʃ/ - anxious
  • /ŋz/ - anxiety
  • /gʒ/ - luxury

For these two-phoneme sequences, I'm wondering how the 2 phonemes are split across syllables.

When the X comes at the end of the first syllable... does the second phoneme from the X get pronounced when you start the second syllable?

  • /ˈtæk.si/?
  • /ɪɡˈzækt/
  • /ˈæŋ(k)ʃəs/
  • /ˌæŋ(ɡ)ˈzaɪ.ə.ti/
  • /ˈlʌɡ.ʒə.ɹi/

Or is there another way to think about this?

  • 2
    That depends on the syllabic theory you follow. For my part, I would split all of these, though I wouldn't worry about unavoidable consonants like /(k)/. Commented Apr 1, 2022 at 16:39
  • Would love to read more on syllabic theory... any links or resources you can point me towards? Thank you!!
    – kanamekun
    Commented Apr 2, 2022 at 2:52

1 Answer 1


In the two cases you've mentioned where the syllable after the consonant cluster is stressed, it is fairly unanimously agreed that the syllabification goes between the two consonants. Thus:

  • /ɪɡˈzækt/
  • /ˌæŋˈzaɪ.ə.ti/

As I wrote in my answer to Why is Anxiety Sometimes Pronounced With a 'g' sound?, it's actually not typical for anxiety to have a [g] between the [ŋ] and [z]. That pronunciation would only be found in an accent where syllable-final [ŋg] exists—such accents exist, but the use of syllable-final [ŋg] instead of [ŋ] is considered a regionalism rather than a neutral feature of the English sound system.

Your other examples are harder.

In the case of taxi, an "onset maximizing" approach to syllabification would yield /ˈtæk.si/. However, there is a competing principle of syllabifying as many consonants as possible with the adjacent syllable with greater stress, explained by John Wells in "Syllabification and allophony" (originally published in Susan Ramsaran (ed.), Studies in the pronunciation of English, A commemorative volume in honour of A.C. Gimson (London and New York: Routledge, 1990)). That principle would yield the syllabification /ˈtæks.i/.

In my view, adopting Wells's principle makes it a little problematic to determine how anxious and luxury are syllabified, because it raises the question of whether /ŋ(k)ʃ/, /kʃ/ and /gʒ/ are ever possible coda clusters in English, which I find difficult to answer. On the one hand, no English words in common use end in /ŋ(k)ʃ/, /kʃ/ or /gʒ/. But on the other hand, I think they are easily pronounceable as-is in word-final position for most English speakers, and /kʃ/ at least would not tend to be replaced by any other sequence in proper names such as Baksh, Buksh or Anksh (I haven't found any proper names ending in /gʒ/). I can also imagine English speakers pronouncing these sequences at the end of words formed as clippings, e.g. when "anxious" is truncated to "anxsh" (compare "usual" > "yoozh"): here's some evidence that this kind of clipping is possible for English speakers:

Tiffany Schaefer @harpandsong

Jonathan came up with a new verb last night, "Anxsh." It's when you're >feeling anxiety for whatever reason and you just need to sit with it and "anxsh" for a bit. 🤣 It made me giggle.

6:40 AM · Apr 29, 2021 -- Twitter

Another example of a similar nature is mentioned in the paper "OMG the Word-final Alveopalatals are Cray-cray Prev(alent): The Morphophonology of Totes Constructions in English", by Lauren Spradlin (2016; Volume 22, Issue 1 Proceedings of the 39th Annual Penn Linguistics Conference): "perfecsh" as a truncated pronunciation of perfection (page 276).

Also see this relevant comment by Chris Button (November 30, 2017) on the Language Log post "Ask Language Log: Unnecessary disyllabism?" (by Victor Mair, November 27, 2017).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.