Based on my understanding, there is no universal vocabulary across languages, which is fine.

That said, there must be words that have a both highly correlated meanings and levels of usage.

Which leads to my question: what are the important concepts to account for in vocabulary comparisons across languages?

  • The closest to lexical universals would be mama-like words for mother and is it papa-like or dada-like words for father. There is even a famous exception to prove the rule with Georgian having the two with opposite the usual pattern. – hippietrail Sep 22 '11 at 20:35
  • @hippietrail: Hmm, maybe my question is not clear, or I'm misunderstanding your comment. I'm not in search of words per se, but meanings that are expressed via words in which the words would have the same meaning in both languages. Within a language such as English, the concept of synonyms might apply and concepts to take into account my be the gender, formality, use frequency, etc of a word when compared to others of similar meaning. Is there something I'm missing in your comment, or a way I might make my question more clear? – blunders Sep 22 '11 at 20:55
  • 2
    Ah well I believe the "same meaning" in this case is a very slippery concept. There are very few if any words that map perfectly 1:1 between any two languages. But some concepts do seem to be more fundamental and atomic, mostly ancient ones like body parts. But even then you find lots of cases like Spanish using the same word for "finger" and "toe" and Japanese not having separate words for "leg" and "foot". I think the Swadesh list is probably as close as it gets. – hippietrail Sep 22 '11 at 20:55
  • +1 @hippietrail: Awesome, thanks for the clarification! – blunders Sep 22 '11 at 20:59
  • Are you asking about -all- known world languages? or about comparing within a given family, say, just Slavic or just Polynesian? – Mitch Sep 29 '11 at 16:58

Things like the Swadesh List are an attempt at a vocabulary list along the lines you are talking about. The goal of such lists is to be:

  • universal
  • culturally independent
  • available in most languages

Another criterion that might matter for some applications is how stable the item is; i.e. how unlikely is it to be a loan word from another language. But this was not a stated goal of the Swadesh list, I don't believe.

| improve this answer | |
  • +1 @James Tauber: Thanks, never heard of that list Swadesh List. Guess my thought was that constructed languages such as Esperanto must have attempted to take this into account in creating a root vocabulary. – blunders Sep 22 '11 at 20:28

It really depends on how you want to establish the semantic similarity, but there are a variety of theoretical frameworks that you can apply – they all excel at teasing out different things. One is Componential analysis, which is just teasing out binary features – which was borrowed over from phonetics. As an example girl would be '+young +female' while boy would be '+young -female.' This analysis is really good for kin and pronoun terms that generally fit into matrices.

A second is Natural Semantic Metalanguage. This theory aims to simmer all languages down to a set of equatable "prime words" which can be built up into longer more complex explications. Aside from any theoretical issues you may have with this it can be a useful tool for teasing apart meaning, especially for complex and abstract things like emotion terms.

A final one I'll mention (although there are more) is Cognitive Semantics. George Lakoff who wrote the book "Women, Fire and Dangerous Things" is a major figure in this area. Those working in this type of cognitive semantics attempt to use abstracted diagrams. This is useful for teasing out spatial things – for example English uses "on" for things that are "on top of" and "on the side of" while other languages have other prepositions for each of these jobs.

There are other theories as well, with their own tools. The most important thing is to be aware of exactly what need you have, and to be aware of any short comings a theory might have.

| improve this answer | |

The list in "A Dictionary of Selected Synonyms in the Principal Indo-European Languages", Carl Darling Buck, 1949, has been used in several comparisons. It has 22 semantic categories of words, like bodyparts, clothing, time etc. AFAIK it includes all the words on the Swadesh list and some 1200 others in addition. The following two collections based this book are online: World Loanword Database and The Intercontinental Dictionary Series. The former has more words and the latter has more languages.

| improve this answer | |
  • +1 @kaleissin: Interesting. Do you think Buck's list is based on Roget's Thesaurus classification system, which I believe dates to 1805. Guess the only thing I can think of is that synonyms are the result of high frequencies of use, not sure if there's any direct meaning beyond that between languages, which I guess is a given. – blunders Sep 23 '11 at 17:25

I just read about a study on this which used a fascinating methodology. Here's a description of it from Arends et al. (1994) Pidgins and creoles, An introduction Johns Benjamins ch. 9 (p. 108):

Twenty Ndjuka [English-based creole, Surinam/Fr. Guyana – AE] morphemes were selected for which the 'area of meaning' [...] differs from the corresponding English morpheme, such as the word mófu, meaning both "mouth" and "pointed end of a pointed object." For each sense of each morpheme, an English sentence was constructed illustrating that particular sense. Correspondents for 43 languages [...] were asked to give the equivalent for each sense of each selected morpheme. Finally, by scoring the extent to which in each particular language one and the same morpheme was used to express different senses of each of the twenty morphemes, these languages could be arranged according to the degree to which their lexical semantic structure resembled that of Ndjuka.

The original study is Huttar (1975) "Sources of creole semantic structures" Language vol. 51 (link to JSTOR). The quantitative methodology is a little dated, but the basic approach to collecting the data is very interesting (and, so far as I know, not being practiced systematically these days).

| improve this answer | |
  • +1 @Aaron: I've used ranking systems like that before, and they're good for creating a ranking of relative similarity, but poor at establishing absolute comparions; or at least in my experience. That said, across a language a ranking like that would I believe be a great way to compare languages. If I'm missing something, please let me know, and thanks for postings, since I hadn't thought of using that method in this case. – blunders Sep 27 '11 at 2:23

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.