The concept of lexical set is a useful technique for differentiating accents or dialects within a language. A lexical set is a set of all words/syllables that are pronounced with the same vowel. These sets partition the syllables in the language, and then how different sets might merge (like the caught/cot merger) can help identify a particular dialect. A particular lexical set is labeled with one of its primary members. For example, there lexical sets

BATH including bath, pass, path, sample,...

PALM including palm, father, ...

In AmE these two sets are distinguished, but in BrE (RP) they are merged (supposedly; for me, palm and father are different).

There is a chart that lays out the merges among lexical sets for English dialects.

My question is this. Are there any such similar lexical sets done for other language families? I would think that German, Italian or Chinese would be amenable to such analysis having complex internal intelligible/unintelligible relations in their families.

(I'm looking for either references for such similar analysis in other language families, or an explanation of why the lexical set method is or is not a good tool for a particular language family).

Addendum: even though I accepted an answer, it didn't answer exactly what I"m looking for. I am hoping for, in the context of very similar varieties of a language, a list of vowel 'contexts' such that within one variety all instances of any particular context are pronounced with the same vowel (this seems to be the definition of Wells' tool. For example, in the set of German dialects there could be a list of words that help distinguish the dialects by a complete inventory of vowel contexts, i.e. not just a simple maken/machen line (a single distinguishing element), but the set of all vowel contexts that helps distinguish all the dialects.

  • Man, those two sets need to be split into four in my dialect of AmE. :P
    – tdhsmith
    Commented Sep 15, 2011 at 17:19
  • @tdhsmith: hm...yes, I took them verbatim from wikipedia; but the concept still works: for a bunch of words, for each speaker, the words can be separated by pronunciation into lexical sets. And two speakers with two different partitions come from different language communities.
    – Mitch
    Commented Sep 15, 2011 at 17:43
  • 1
    What is your question? Are you just looking for examples of such lexical sets in other languages? In that case, this is an open-ended question with no definitive answer...
    – Timwi
    Commented Sep 16, 2011 at 11:09
  • @Timwi: I'm looking for references to similar analyses for other languages. I've done a superficial search with no luck, but often things like this are helped immensely by local domain knowledge. I'm not asking for people to construct them off the top of their heads for an answer here, but if this question motivates such a construction, then great (and a link to a web site of such an initial speculation would be great too).
    – Mitch
    Commented Sep 16, 2011 at 12:15
  • 1
    Do you mean that languages such as German, Italian, or Chinese would not be amenable to such analysis? The varieties of Chinese are more like languages than mere accents, with words being replaced by non-cognates etc.
    – Zhen Lin
    Commented Sep 29, 2011 at 2:22

2 Answers 2


The wikipedia article on Lexical sets cites:

Armstrong, Nigel (2001). Social and stylistic variation in spoken French: a comparative approach. Amsterdam: John Benjamins. pp. 100ff.

  • hm...I missed that. Thanks. A quick google search also finds "French words: past, present, and future" by M. H. Offord, which uses sets for semantic comparison. That is, I was looking for phonologically defined sets, but this is interesting extension of the technique.
    – Mitch
    Commented Sep 29, 2011 at 15:24

Frank Blair's 1990 book "Survey on a Shoestring: A Manual for Small-Scale Language Surveys" gives a model for how to do this kind of analysis – it's the best non-computer based process I've found so far, although it does have limitations it its rather brute-force method. You might find some people who have used this method by checking who's cite Blair. I used this process in Lauren Gawne (2010) "Lamjung Yolmo: a dialect of Yolmo, also known as Helambu Sherpa" in Nepalese Linguistics, which is available online. This kind of process requires you to decide how you're dealing with tone, and closely related sounds so it's best to check people's methodology.

This kind of process is useful for determining how phonetically related two dialects, or closely related languages are. If you have more than two languages like I did in that paper this may help in establishing the historical relationship between groups of languages.

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.