8

Note: I am not a linguist, please provide any corrections for terminology.

I would like to find some approximate data (if it exists) comparing several languages with the number of different syllables in all used words of the language and in the X% most used words of the language.


I have been only able to find some (non-reliable) information for the number of different syllables in all used words of the language:

  • 6
    I think this is basically a question about typology and phonotactics, in case that helps your search. As you investigate, you may also want to make a distinction between the set of ‘theoretically possible syllables’, ‘phonotactically likely syllables’, and ‘actually/frequently attested syllables’. – Jeremy Needle Aug 7 at 15:28
  • 2
    @JeremyNeedle Thanks for the comment, I updated the tags and edited the question to make clear I mean actually attested syllables. – Puco4 Aug 7 at 15:35
16

Yoon Mi Oh's 2015 thesis (pages 44-45) provides estimates of the number of syllables for various languages, gathered by taking the 20,000 most frequent words in a corpus of each language and counting the different syllables that show up. Ordering them by increasing number of syllables:

Japanese: 643
Korean: 1104
Mandarin: 1274
Cantonese: 1298
Basque: 2082
Thai: 2438
Italian: 2729
Spanish: 2778
French: 2949
Turkish: 3260
Catalan: 3600
Serbian: 3831
Finnish: 3844
Hungarian: 4325
German: 5100
Vietnamese: 5156
English: 6949

For example, her count for English is 6,949 syllables—significantly less than the 15,000 you cited, because English has a lot of possible syllables, but most of them aren't actually used. ("Wug" and "strall" follow the rules of English syllable structure, but don't actually appear in any commonly-used words.)

Page 58 of the same thesis also gives the Shannon entropy of the syllable distribution in each language, which is an alternate way of measuring it; this takes into account the fact that English has a lot of syllables, but some of them are very rare (like the syllable "strengths", it only appears in a single word), while e.g. Japanese has fewer syllables, but they're more evenly distributed. By this measure, an English syllable in isolation conveys 9.51 bits of information on average, and a Mandarin syllable conveys 8.69 bits.

| improve this answer | |
  • Thank you a lot for the answer! I had read Yoo Mi Oh's very interesting Science article but I didn't now she also studied this. For completeness I suggested an Edit, feel free to accept it or change it as you wish. – Puco4 Aug 7 at 20:02
  • 1
    Doesn't a homophone of "strill" appear in the word "nostril"? Normally I would try to suppress the urge to be a smartass and point this out, but I'm curious to know if the definition of "syllable" you are using includes syllables differentiated only by orthography (e.g. are "lam" and "lamb" two different syllables?). – Paul Aug 7 at 23:47
  • 1
    @Paul Good point! I missed that one. Technically I'd put the S in the first syllable in "nostril" but it's still too close for comfort; I'll change it to "strall". (For the purposes of the answer, I'm going by phonemes, but spelling everything out in English orthography because OP said they're not a linguist and thus probably don't know IPA.) – Draconis Aug 8 at 0:21
  • @Draconis If you include reduced variants, strall could then be sort-of present in ancestral. Can’t think of any contexts where /strʌl/ or /stroɪl/ would appear, though. – Janus Bahs Jacquet Aug 8 at 9:36
7

Terminologically, I think you are interested in the number of "distinct syllables" in a language. "Syllabic phoneme" means, approximately, "vowels", but also syllabic consonant (as exist in some languages), and with the provision that the sound has to be contrastive and not allophonic (for instance, you would not count with [ə] and [ʌ] in English since the distribution is rule-governed).

This is not information generally available for languages, and Barker's underlying paper on English shows you why. For some languages it's a simply matter of computing and "trivial" searching, but even then, non-triviality lurks. Hawaiian has 8 consonants and 5, 10 or 25 vowels, depending on how you treat long vowels and diphthongs. If we take the smallest numbers, syllables are of the type V and CV, therefore there are 5+40 theoretically possible syllables. Then you would just search a dictionary to see if they all exist (there is a gap for [wū] and only 2 examples of [wu]). Taking the largest number, you get 25+200 (possibly minus 1 or 2). The count in English is much higher because we have more phonemes, and the possibilities for combination are greater = (C(C(C)))V(V)((((C)C)C)C). But not every C can appear in every positions: plus, as noted by Barker and every every other linguist, it's non-obvious where the syllable boundaries are in English, so you have to decide if [btʃɪk] is a syllable of English, given "Dabchick" (I would say no but I'm not here to argue with his algorithm). English is rife with syllabification restrictions (syllables can't start with [ŋ] unless you syllabify intervocalic [ŋ] as in the onset).

This person claims that Vietnamese has 17,974 syllables, but this is a product of combinatoric calculations, and it is noted that about half of them don't actually exist. That's a pretty high percentage, suggesting that there are non-accidental gaps. For instance, there is no syllable in English that begins with "bn", which linguists generally consider to be the result of a rules. There are no syllables composed of sCVC where the two consonants are the same and the vowel is short, with a tiny number of counterexamples like "stet", "stat": some linguists (Clements & Keyser, among others) have that as a systematic fact of English. There are no syllables having [bl] as the onset, [ɪ] as the nucleus, and a coronal non-sibilant as the coda (t,d,n,tʃ,dʒ): nobody has previously noticed that gap and proposed a rule. We don't know if that is just a gap in attestation, or an actual rule-governed gap.

In principle if you could get a comprehensive list of all words, including inflectional forms, for a language and could decisively parse words into syllables, you could count the number of actually attested syllables in a language. Most comprehensive word lists don't include inflected forms, and they are usually based on written forms of major national languages. Polish, for example, allows long sequences of consonants, but this extra combinatoric power is concentrated in word beginnings and word ends. In the context VC*V, it's controversial where to put the syllable break (this is the general problem of syllabification algorithms and the question of what to count).

In other words, no, not generally, there aren't many such enumerations.

| improve this answer | |
  • As for that case of no syllable in English that begins with "bn", which rules is that sound combination a result of (according to some linguists, as you put it)? In fact, only /s/ can be followed by /n/ in the syllable onset. – Yellow Sky Aug 7 at 17:25
  • Depends on how such rules are stated. One approach is admission-centric and the other is prohibition-centric. Usually, filters state sequential relations and don't have a focus whereas standard rules apply to a particular segment. So in a more rule-ish approach the input would be C[nV], and the onset-parsing would admit only [s] (maybe [ʃ]). Filter-type approaches have for example blocked onset sequences of [-continuant], thus also getting *pt. – user6726 Aug 7 at 19:19
  • @YellowSky How about the word "Mnemonic"? – Phil Freedenberg Aug 8 at 20:09
  • With an "m" in spelling, but not pronunciation. These rules are not about spelling. – user6726 Aug 8 at 20:11

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.