0

I am looking for a list of the most frequent syllables of all languages spoken on earth, sorted by frequency.

I found such lists for english and for german, but I want to get a list across all languages.

Suppose you can count every syllable that is spoken on earth within 24 hours. Syllables from languages with many speakers like Chinese (Mandarin), English or Hindi are more often used then syllables from rare languages like Esperanto or Klingon and so are ranked higher in the list. And within one language there are syllables that are more often used then others.

I know, what I want will be hard to realize. So I also would be happy with separate lists of the languages with many speakers, lets say with the top-200 syllables. (I'm not interested in rare syllables, my focus is on common syllables. For the same reason I'm not interested in rare languages.)

And I really want spoken syllables, not written di- and tri-graphs ("the" is a trigraph, not a syllable. It can be the written representation of the syllable /ðə/ but also part of /ðɛn/)


Addendum

Here are the lists that I've found:

German:
http://www.uni-potsdam.de/fileadmin/projects/treatmentlab/assets/Silbenfrequenzen_TreatmentLab.pdf

Columns from left to right:

  1. Ranging position.
  2. The syllable.
  3. This number tells you how often this syllable can be heard when one million words are spoken (which means, that frequent words are repeated many times and therefore theirs syllables are counted more often)
  4. Take a dictionary of all German words, where flexions of the same lexeme are distinct words (run, runs, ran and running would be listed as four different words). Count how often the syllable appears in this dictionary (the dictionary itself contains 365.530 different German words.)

This list is sorted by the 3rd column, and this is exactly what I want. This column answers this question:

When you listen to Germans speaking one million words, how often can you hear which syllable?

I'm not interested in the 4th column.

English:
https://medium.com/@rosson/high-frequency-syllables-in-english-ab75159618a0#.jaee1pxhq

This list has only 3 columns, where #1 and #2 are identical with the first two columns of the German list. Columns #3 is the same as the German col#3, but in the German list you have counts per million, in the english you have percent, which is just another representation of the same thing.

How would I use those lists to create an all-language-list?

In the english list I would convert the numbers in column 3 into counts per million like in the German list. Then I need the number of native speakers per language. Such a list can be found here: https://en.wikipedia.org/wiki/List_of_languages_by_number_of_native_speakers

As you see, English had 360 million native speakers in 2010, and German had 89 million native speakers in the same year. So I multiply the (converted) english numbers by 360, and multiply the numbers in columns #3 of the German list with 89, and then merge both lists. If there are syllables used in english as well as in German (like [ɪn]), I add both wighted numbers. Then I sort this list again by those numbers.

The result is a list of the most frequent syllables of the German and English spoken part of the world.

And because I am interested in the most frequent syllables, it is safe to ignore rare languages, among which you will find all languages, that not have been studied.

And to give you an example what I am not looking for: http://www.sttmedia.com/syllablefrequency-english There you find numbers of digrams and trigrams, but this are not phonetic syllables!


Addendum 2 (March 2018)

I think I better tell you the context. (What do I want to do with this list?)

I am writing a fantasy story. There was a small group of aliens traveling through space, and they got lost. They had technical problems and they stranded on a planet that turned out to be our planet earth. This happened thousands of years ago, and all humans on earth are descendants form this small group, and all languages spoken on earth today derived form the language of this small group of invaders.

I want to construct a language, that could be the proto-language of the first humans on earth.

My second goal is to make this language as easy to be pronounced by everybody on this planet, as possible. I want that people from China find it as easy to pronounce as people from Saudi Arabia, from Russia, from the USA, from Namibia, and so on.

I think that i don't need a set of common phonemes to combine, but a set of very common syllables, because even wide spread phonemes can be hard to be pronounced, when you use unusual combinations.

I want to try to build an artlang that is made of common syllables, and this is why I am looking for the most frequent spoken syllables on earth.

btw: When two syllables are similar, I want to count them as identical. For example the english words "I, eye, ay, aye" and the German "Ei" are all the same syllable (/aɪ/).

  • If you can provide links to the lists you found for English and German, we would be better able to understand what you are looking for. E.g. is the frequency of a syllable a function of the number of speakers in a language and the probability of the syllable in a random spoken corpus? – user6726 Dec 25 '15 at 0:09
  • 2
    90% of languages have barely been studied at all, let alone had a decent audio corpus collected for them. What you want will probably never be achieved. – curiousdannii Dec 25 '15 at 1:23
  • @user6726: I've added the lists that I've found and described them. – Hubert Schölnast Dec 25 '15 at 10:39
  • @curiousdannii: Languages that are not studied until now are languages with a small number of native speakers. Since I am not interested in all syllables, but only in the most frequent, it is safe to ignore languages with only a little number of speakers. The impact of the most frequent syllable of a language with only 5000 speakers in the top-200 of a weighted list of all languages is ignorable. – Hubert Schölnast Dec 25 '15 at 10:44
  • 1
    This souds like an XY Problem. Instead of "I want X" you might want to say "I want to accomplish Y and I think X would get me there," especially if X is odd or downright crazy. I'm thinking this is poorly defined because you seem to assume every language defines a "syllable" the same way, and misdirected in an undocumented assumption that you can conveniently ignore dialectal and idiolectal variations. – tripleee Mar 18 '18 at 12:44
3

The token frequency data you need is vanishingly rare. I think you could add Dutch, and perhaps you can add in a couple more languages like French and Spanish (although you could get pilot estimates for various languages, where a study reports frequency data for one or two non-random texts written in a language). The biggest effect that you would need to control for is the under-studied language problem. The top dozen world languages, population-wise, are in descending order Mandarin, Spanish, English, Bengali, Hindi, Portuguese, Russian, Japanese, German, Wu, Javanese and Korean. Together, 47% of the world's population speaks one of these languages, and the remaining 53% speak one of the other 6,772 languages. (This is based on population and language information that I acquired from SIL 15 years ago).

I suggest weighting the contributions of each language according to the number of speakers, so if you do get Dutch data, it contributes very little to the total count. Since Mandarin and Spanish have the highest weight in the global syllable-frequency competition (approaching 20%), it would be most important to cover those languages and assure that data from those languages are accurate / representative. I found a character-frequency table for Chinese here, which reports that the most common character is 的, which apparently has 3 pronunciations, although that is in Mandarin, and there is no way to know whether a given Chinese writer speaks Mandarin vs. Wu vs. Yue. But you could probably take the top 200 characters, research their phonetic values in the main Chinese languages and arrive at a weighted contribution.

Addendum: in light of the revised motivation for this question, the desideratum should not be finding the "most frequent" syllables. The underlying question is "which things exist in most languages", not "which thing occur most frequently". The goal should be to construct, first, sets of syllables that exist in all human languages, and subsequently, sets of syllables that exist in all but 1 language (all but 2, and so on). That way, you would not select syllables that are highly frequent in Chinese just because Chinese syllables are uttered very frequently.

It has to do with grammatical systems and not corpus studies of token frequency. The basic answer is, for the most part all CV, where C is a universal consonant and V is a universal vowel. The problem is that there are no truly universal consonants or vowels, just high frequency ones. If you select [a,i] as the vowels, [t k n] as the vowels, then there are 6 candidates, though [ki] is going to be a problem. Actually [ti] is also a problem, just less so. There is no repository of syllables in all languages, nor is there a respository of all vowels and consonants.

The question invokes the notion of a "rare language", which only makes sense as being about languages having few speakers. There are thousands of rare languages and only hundreds of common languages. You could use this list, or some subset of it, to find the language basis for the syllable compilation. Then all you would need is a list of the existing syllables of that top set of languages. The English syllable frequency list has a bit of a defect in that the underlying database (the CMU dictionary) does not assume syllable structure, so the author computed syllables in a manner that comports with other computational practices. We have no idea how that relates to speakers of English. Inspection of the list does reveal a number of putative syllables that fail the basic speaker test of syllable-hood (can speakers say it?): examples [æ, æmk, ɑfsk, ɑh, bæ, bɔj distinct from bɔɪ, bɛ, bɪ... θs (!?)]. So there is no syllable list for English. You might get such a list for Chinese languages (well, Mandarin and Yue), and Japanese should be pretty simple to do.

  • 2
    I think taking the top 200 characters would work especially poorly for high-frequency words, because those 200 characters are going to be top frequency for Mandarin, and many of them will be function words, which will be especially divergent across dialects. For example, 的 is going to be very infrequent in Yue compared to in Mandarin, and you would miss 嘅, the rough equivalent of 的 in Cantonese... For major dialects (which the OP would be satisfied with anyway), I think we can simply use dialect-specific corpora. I have done this for Cantonese before and it should work for others as well. – WavesWashSands Mar 18 '18 at 2:24
1

You are mixing popularity with universal acceptability, but they are two different things. Pork is a very popular staple of the diet in many parts of the world, but that doesn't mean it's acceptable to everyone, as you are well aware (vegetarians, religious reasons, etc).

There are modern languages with very constrained syllabic structure, so to construct a language which is equally easy to pronounce for speakers of contemporary Japanese, Hawai'ian, Rotokas, Pirahã etc etc etc you would be restricted to roughly something like CV structure with maybe eight consonants and four vowels. This sort of very limited repertoire typically leads to very long words (as seen in many languages of the Pacific) and/or heavy homonymy (cf Chinese).

But examining only modern languages completely misses language evolution; to the extent that we can recover, reconstruct, or reason about earlier language, it is not uncommon to find a different set of constraints which have changed over time. Chinese is a good example; modern Mandarin has a fairly constrained syllable structure, but earlier proto-Chinese apparently had a more complex one. Similarly, some modern Romance languages have acquired constraints which did not exist in Latin (word-initial sp- is forced to esp- in Italian and Spanish, for example, I believe). Reconstructed proto-Germanic word forms have many features which are outright alien to modern German; etc etc etc. This brief exposition barely touches the surface of language evolution; let's just note that it's constant, inevitable, arbitrary, and universal.

So in fact, for a work of fiction, no particular features of a constructed language would make it more credible or feasible to a linguistically informed reader. If anything, postulating that it has a direct connection to or influence over modern languages is the one thing that would make it hard for me personally to suspend my disbelief.

Exploring genuinely universal language features is interesting to linguists, because it can reveal things about our innate language mechanisms. However, to the extent that truly universal features have been identified across all known languages over time, these findings tend to be more on the level of "all languages have vowels and consonants", and even though many languages share a particular contrast, it's often hard to say unequivocally even that back vs front vowel contrast or voiced vs voiceless consonants are a universal feature of all human languages.

  • I think his goal is to make up a language which anybody could pronounce, regardless of their language. – user6726 Mar 18 '18 at 20:35
  • Yes, exactly. I'm addressing that directly in the second paragraph. – tripleee Mar 19 '18 at 4:13
  • Also came across this; sciencedaily.com/releases/2014/04/140401102126.htm – tripleee Mar 19 '18 at 4:56
  • @jknappen Thanks, updated. I didn't want to go into French because I don't know how systematic it is, though I know about examples like ésprit etc and also of course the broader pattern of s+clusile -> es+clusile or even ê + clusile. – tripleee Mar 19 '18 at 7:46
  • Thank you for your answer. you are right, and I agree with you in every sentence that you wrote. But »I am looking for a list of the most frequent syllables of all languages spoken on earth, sorted by frequency.« (this is the very first sentence in my question), and you didn't give an answer to this request, so I can't give you +1. – Hubert Schölnast Mar 19 '18 at 15:35
0

I could not find exactly what you were talking about. I was looking for the same thing. But this website has a list of the most frequent sounds across languages: http://web.phonetik.uni-frankfurt.de/upsid.html

  • Actually, it does not. It contains a list of segments in a small convenience sample of languages (viz. where the author could get data), and you can compute frequency of those segments within that set. – user6726 Mar 17 '18 at 16:12
  • Thank you. I know this site, but it contains only sounds, not syllables. Btw it seems to be old. At web.phonetik.uni-frankfurt.de/L/L2004.html there is a comment about German: "German is spoken primarily in West and East germany." But the both Germanys united 28 years ago (in 1990). – Hubert Schölnast Mar 17 '18 at 17:02

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.