1

Wondering if there is any sort of list either across languages or for individual languages, either complete or partial, that list the sequences of vowels and/or consonants used in that language. If no comprehensive list exists, then just looking for one for English or other main languages.

Basically just looking for sequences that appear in the language, either vowel sequences, consonant sequences, or vowel-consonant/consonant-vowel sequences. Or perhaps maybe just vowel sound sequences. I think knowing the IPA sound sequences might be more relevant.

I would like to see which ones are the most and least common, as well as just see the scope of each language, and what's possible.

In English that list might look like:

ay
ai
ou
ow (vowel sound like cow)
oo
aia
...
ask
isk
is
...
Improve this question
2
  • Is this the kind of thing you're looking for? World Phonotactics Database.
    – Gaston Ümlaut
    Sep 6, 2018 at 22:23
  • Yeah I think that's close to it! A bit difficult to use, would be nice to just see all the data directly :) Hmm...
    – Lance
    Sep 6, 2018 at 23:39

3 Answers 3

Reset to default
3

If you are interested in letter sequences and are interested in any non-zero frequency of attestation, you may be able to find such tables for a particular language. I suspect that you can't find such a table for Luganda or Kven, and for most languages such a table would be impossible (or meaningless) since the languages aren't extensively-enough written that the tables could be constructed.

If you are interested in sound combinations, the available data becomes more scarce, and probably involves translation from written language to phonemic transcription. This is not too challenging for a language like North Saami where there exist programs to map written form to IPA, also for Luganda (and probably Kven) where the writing system is phonetically transparent, but for languages like English with suboptimal spelling systems, you may have to rely on a specially-created lexicon with a transcription (which exists for English). In the case of Tibetan, it's not clear what it would take to come up with such a mapping. What Tibetan and English have in common is that they have been written for hundreds of years (unlike Luganda, or North Saami in the current orthography).

There is a linguistic concept of "phonotactics", which refers to putative rules governing co-occurrence of segments within a syllable or word. There is no end of controversy over the validity of some or all phonotactic rules. The controversy arises because there is no clear and accepted criterion for saying that a given pattern is to be "allowed" or is "disallowed". For example, it is an absolute fact that no syllable of English begins with the sequence [sʃ] (a good candidate for exclusion); many syllables begin with [sp] (a good candidate for inclusion); and then there are sequences like [ʃp] which also don't exist in English as long as you exile words like shpiel, shpritz.

There are also claims to the effect that certain nonadjacent sequences may be phonotactically excluded, for example the onset and coda consonants in Cantonese can't both be labial. It is claimed that English excludes [sCVC] where the consonants are the same ("skick, spip, snan" aren't words). The problem is that we have not managed to convincingly distinguish between "is forbidden" and "is not attested" or "is very uncommon". There is thus a strong tendency in modern non-computational phonology to set aside claims about co-occurrence, unless there is compelling evidence that the pattern is part of the grammar.

Improve this answer
2
  • Well put, but your claim at the end feels too strong: there certainly are ways to figure out whether phonotactic rules are actual rules or not. For example, spiel with [ʃp] came into English intact, while tsumani with [ts] did not. This seems to indicate an underlying rule of English.
    – Draconis ♦
    Sep 5, 2018 at 23:20
  • In fact I think that we can demonstrate the validity of certain rules of syllable-parsing. The problem is that the ban on ʃp has been expressed as "about onset formation" and not generic phonotactics. For current purposes, distinguishing valid arguments from invalid arguments regarding syllabification would be "orthogonal". I am uncertain about [ts] because unsimplified and unmodified clusters persist for many people in Zeitgeist, Tzotzil – check the Google dictionary entry – and tsunami itself.
    – user6726
    Sep 5, 2018 at 23:32
4

Firstly we should clearly separate pronunciation of languages and orthography of written languages.

If you are interested in written languages then you basically just want character-level n-grams.

A list trained on a real text corpus will likely be too noisy to be readable, you want a frequency-ranked list.

Your intuition is correct that every language has a signature, which is the basis of the most common approach to language identification.

Improve this answer
3

The sequences of sounds allowed are called the phonotactics of a language. This is generally written in an abbreviated way:

Japanese syllable structure is (C(j))V(N|Q). /ji/, /wu/, /wi/, and /we/ are disallowed. /Q/ must precede a consonant.

This looks complicated, but it's explaining what a syllable in Japanese looks like: the onset is optional and always takes the form C(j) (in other words, a consonant, optionally followed by /j/), the nucleus is mandatory and always takes the form V (a vowel), and the coda is optional and always takes the form N or Q (the specific phonemes /N/ and /Q/).

/Q/ and /N/ are special phonemes in Japanese, the exact details don't matter here.

So what vowel sequences are allowed? Well, you can have as many V syllables in a row as you want, and this description doesn't say that any vowel sequences are disallowed, so you could string any vowels you want together. If you wanted you could write them all out, but this probably wouldn't be particularly useful.

What consonant sequences are allowed? The longest consonant sequence you can get, phonemically, is /N/ or /Q/ from the coda of one syllable, followed by /Cj/ in the onset of the next, for any consonant C. So for every C you have the potential sequences /NCj/, /NC/, /N/, /QCj/, /QC/, /Cj/, and /C/. You could write all of them out if you really wanted to.

Improve this answer
3
  • Wondering if every language has an explicit set of phonotactics or rules, or if they are nebulous.
    – Lance
    Sep 5, 2018 at 20:45
  • 1
    That depends on how they are described, and by whom. Native speakers have them specifically, in operational form, or they couldn't speak or understand others; but that's very different from specifying the rules in descriptive terms. That's mostly what experimental phonetics is about. Most languages have not been studied to a very deep level of phonotactics, though -- there are simply too many languages and too few linguists. Most importantly, there's too little time. It takes a lot of time and effort to work out even a sketchy phonotactics. I know; I've done it.
    – jlawler
    Sep 5, 2018 at 21:25
  • 2
    @LancePollard Every language has solid rules for this, the question is rather whether they've been set down in writing by linguists or not. And figuring out the phonotactics, as jlawler said, is a long and arduous process.
    – Draconis ♦
    Sep 5, 2018 at 21:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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