A colleague of mine made a claim that the phonemic length of the root morphemes in whatever language does not usually exceed 5, as an average. I have some doubts about this unsubstantiated claim, based mostly on the "linguist's intuition". Does anyone have a good reference for a paper, or some other kind of source, dealing with the topic of the average root length cross-linguistically?

Notice that I am not interested in word-form length (polysynthetic languages would obviously show impressive figures), but just the length of the roots.

  • 2
    Averaging might be tricky: Do you use frequency-weighted average (that will reduce the length of roots because of Zipf's laws) or do you just take a dictionary based approach? Commented Jul 27, 2017 at 10:18
  • Dictionary based. It's for an introductory manual of linguistics and the author wants just to make some general, necessarily vague, but not too unsubstantiated, quantifications. Commented Jul 27, 2017 at 10:20
  • @jknappen Exactly. Also, as the answer from user6726 suggests, how to weight the languages? Should Luxembourgish count the same as Bengali? Commented Jul 27, 2017 at 10:45
  • @A.M.Bittlingmayer I am obviously assuming that there are no theoretical issues in using data from a arbitrarily large collection of languages. You might call it wishfull thinking, but without such an assumption things like www.wals.info would not have been realized. Commented Jul 27, 2017 at 10:53
  • @ArtemijKeidan Wals can be used to get metrics that are not plagued by this problem. For example, the distance between two languages will not be skewed because of arbitrary decisions about which languages to include. Yes, Wals can also be misused, but so can almost everything. Commented Jul 27, 2017 at 11:13

1 Answer 1


I am pretty sure that there is no "decent" study of the question, i.e. one that is non-anecdotal and is well-balanced in the languages considered, so it's unsubstantiated; but the feeling probably could be born out, within the margins of what's clearly establishable in morphology. For example, the majority of roots in Bantu languages are C(V)V(N)C, though there are some roots like *kamat "hold" which seem a bit longer: but this is usually analyzed as a CVC "core" root plus a non-productive extension -at. Austronesian languages have a considerable share of longer roots that look reduplicated, so one might reduce with "roots" to a true root plus a reduplication morpheme. If you have weak criteria for establishing what is a root, it's easier to eliminate longer roots.

The other problem has to do with claims about "on average", as a claim about human language . On average, languages are Niger-Congo or Austronesian, so if it turns out that most Niger-Congo and Austronesian roots are 5 phonemes or less, you'll probably get the desired result. It is probably true that in most Niger-Congo languages you have one root syllable with maximally CVC struction, and Austronesian roots tend, as far as I know, to be shortish.

  • What do you mean when you say that "on average, languages are Niger-Congo or Austrinesian"? Commented Jul 26, 2017 at 23:24
  • 3
    I mean that the majority of languages fall into those two phyla, so when you look at what "most languages" do, you're looking at properties of these two phyla. They have massive leverage in any statistical study, unless you have control over genetic relatedness.
    – user6726
    Commented Jul 27, 2017 at 0:36
  • 1
    New Guinea languages are another dense cluster, but they vary so much in features they probably wouldn't affect a generalization based on Austronesian and Niger-Congo. This reminds me of somebody's famous assertion, "To a first approximation, an animal is a beetle."
    – jlawler
    Commented Jul 29, 2017 at 18:00

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.