When is a consonant simply followed by labliasing sound like /u/ or /w/ and when is the consonant itself considered palatised or labialised?

For instance, French point, in IPA, /pwɛ̃/ and Croatian pjevati /pjêʋati/. Contrast this with Polish śpiewać which may have IPA /ɕpʲɛvat͡ɕ/ or /ɕpjɛvat͡ɕ/ or English swoon may have IPA /sʷuːn/ or /swuːn/ but usually the latter.

The answer, I would guess, is when the phonemic difference is contrastive. This is problematic because in many phonetic transcriptions of words which do mark consonants with a raised modifier (see Polish and English above), the sounds are not contrastive in the language described.

/pʲ/ and /pj/
/sʷ/ and /sw/

If these are not interchangeable, when should I use either?

  • What are you asking, sorry?
    – Anixx
    Dec 9, 2016 at 17:36
  • First sentence. Dec 9, 2016 at 17:37
  • 2
    This does cause problems sometimes, but usually only in languages where both occur and contrast. Historically, the reconstruction of PIE *ekwos (instead of *ekʷos) is an example of the problem. In Lushootseed, which has both velar and labiovelar stops, and also velar stop plus /w/ clusters, one way to detect the labiovelars is to notice whether the consonant itself is rounded by listening for diphthongization of preceding vowels. Thus /-akʷ/ will be pronounced [aʊkʷ], because the [k] part is rounded.
    – jlawler
    Dec 9, 2016 at 17:38
  • @slipperyiron are you asking how to understand it based on IPA notation? If so, the following "j" in IPA means that the following consonant is /j/, while palatalization is never indicated this way. It is usually indicated either with superscript "j" or special character for the consonant or special character for the following vowel, depending or language.
    – Anixx
    Dec 9, 2016 at 17:43
  • What language is your last sentence referring to? "X and Y are separate sounds" is only meaningful in the context of a specific language. More generally, it's not clear to me which of the following two things you're asking: (a) how to distinguish between e.g. /pʲ/ and /pj/ in a language that contrasts the two, or (b) how to choose between the two notations in a language that doesn't contrast them.
    – TKR
    Dec 9, 2016 at 18:17

3 Answers 3


Deciding between a cluster analysis and a coarticulation analysis when the two do not phonemically contrast is always to some extent a matter of analytical choice, so there's no general answer to your question. Even in English at least one serious linguist has proposed that e.g. the initial cluster of a word like stop should be considered a single phoneme. Here are some possible factors that could influence your choice:

  • Does [j] occur outside of these sequences? If not, it's unlikely to be a separate phoneme.
  • Does your language generally allow initial clusters? If not, this is a point in favor of the palatalization analysis.
  • Is there evidence for a syllable boundary between (for example) the [p] and the [j]? If so, this points to the cluster analysis.
  • Can your sound occur in coda? If so, the sonority hierarchy would suggest it's a single phoneme.

  • In languages with distinctive palatalization, this generally affects whole series of consonants (e.g. all the velars, or even all the stops): is this true in your language?

  • How frequent are the sounds/sequences in question? If they're single phonemes, you should expect their frequency to be roughly comparable with that of other phonemes; if they're clusters, they're likely to be significantly less common than most phonemes.
  • This an excellent guide. Better than anything I would have hoped for. Thank you. For Anuak, Mabaan and Shilluk, this breaks 4:1 in favour of palatalization analysis. Palatalised sounds don't occur in coda in the la Dec 9, 2016 at 19:00
  • Multiple serious linguists have proposed the single-phoneme analysis for all onset clusters in English: Osamu Fujimura's C-D model features this as a means of disposing of linear order of segments.
    – user6726
    Dec 9, 2016 at 21:16
  • @user6726 Thanks, I didn't know that. It seems that if you treat all clusters as phonemes then you've just redefined "phoneme" as "allowed cluster" -- what's the theoretical payoff there?
    – TKR
    Dec 9, 2016 at 22:44
  • 2
    @TKR, disappointingly minimal. You don't have to say what order phonemes come in: it follows from prosodic structure. I'm not a huge fan, I have to say, but we just love exploiting "predictable" properties.
    – user6726
    Dec 9, 2016 at 23:08

IPA was not designed for languages with palatalized consonants, so it uses different signs when dealing with different languages. For instance, in Russian, Irish and German the IPA notation is different. In Russian and Irish it is mostly denoted by the IPA on the consonant, in German - with the following vowel. Even for Russian some palatalized consonants get superscript j as indication of palatalization, while others have separate characters, in some cases the both ways are permitted. IPA is just a mess.

If you are asking how to understand it based on IPA notation, the following "j" in IPA means that the following consonant is /j/, while palatalization should never be indicated this way. It is usually indicated either with superscript "j" or special character for the consonant or special character for the following vowel, depending on language.

  • What dialect of German has palatalized consonants?
    – TKR
    Dec 9, 2016 at 18:22
  • 1
    @TKR everyone, in words such as München (which has two palatalized consonants) upload.wikimedia.org/wikipedia/commons/c/cc/De-M%C3%BCnchen.ogg The consonant denoted as "ch" in German as well as "j" are always palatalized. Moreover, even English has palatalized consonants, but IPA does not indicate them in English.
    – Anixx
    Dec 9, 2016 at 18:27
  • 1
    OK, but there's no phonological contrast there -- it's a different beast from the Russian or Irish type of situation, where palatalization is contrastive. Transcribing non-contrastive elements is always a matter of choice to some degree. Anyway I don't see how "IPA is a mess" -- the [ʲ] symbol is perfectly clear.
    – TKR
    Dec 9, 2016 at 18:30
  • 1
    The contrast in German is in the vowels, not the consonants -- at least, that's the standard analysis. If your analysis places it on the consonants, IPA gives you the same tools to express that for German as it does for Russian.
    – TKR
    Dec 9, 2016 at 18:39
  • 1
    @TKR that's why it is bad to represent phones and why it represents the same sounds of different languages differently.
    – Anixx
    Dec 9, 2016 at 18:56

The application is that I'm trying to ascertain sound changes in the Nilotic Luo languages

I can tell you that when I read professional linguistic literature, I usually face with ad-hoc transcription systems, without any proper description. The professional comparative linguists even when dealing with Russian that contrasts the two, often use "j" where there is no such sound! For instance, I often see things like "bjelo" when dealing with Russian, even though this field requires precision in depiction of phonemes!

Below is an excerpt from Fortson's "Indo-European language and culture". All listed words are supposed to be Russian and used to analyze sound changes:

enter image description here

But what??? How does he indicate palatalization? In multiple different ways! At the end of the words he uses apostrophe, except the word kakoj where he does not indicate it at all. In the middle of the words he either uses "j" in words that do not have this sound (knjaz', desjat') or simply does not indicate palatalization at all (čtenie has all 3 consonants palatalized, but it is not indicated by Fortson, dennica has 2 palatalized consonants, d and n). In word desjat' all 3 consonants are palatalized, but Fortson does not indicate palatalization at all in the first case, uses "j" in the second case and apostrophe in the third case. Three different methods in one word!

How one can arrive at precise reconstruction of an ancient language if he does not distinguish different consonants in a living, widespread language, where these differences are phonemic and etymologically determined?!

  • The reconstruction isn't really meant to inform genetic classification—yet. Nilotic subclassification is very stable. The comparative method under neogrammarian assumptions can be applied to quickly identify likely errors so that we know what forms require new corroborating evidence. This is especially important for poorly attested spoken languages which don't yet have an orthography. Most fieldwork is done near blind which brings a sizeable risk of error. This is the intended use of my work. Dec 9, 2016 at 19:29
  • 2
    Downvoted because this doesn't address the question.
    – TKR
    Dec 9, 2016 at 22:09
  • 2
    I assume the palatalization is not marked in "čtenie" because it's predictable: č is always palatalized, consonants before e in native Russian words are always palatalized, and consonants before i are always palatalized. Dec 9, 2016 at 22:23
  • @TKR the question is, as understood it, how to convert some ad-hoc transliteration system to IPA.
    – Anixx
    Dec 10, 2016 at 3:58
  • 1
    The question isn't really about transcription but about analysis: how do you distinguish between a single coarticulated phoneme and a sequence of two phonemes? But even if it was about IPA, this answer has nothing to do with IPA.
    – TKR
    Dec 10, 2016 at 5:45

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.