How do you use WALS data?

We’re always curious to learn about how the WALS data is used by others.

Recently, a very interesting site came to our attention: The Conlang Atlas of Language Structures uses the WALS feature set as questionnaire, letting people fill in the values for various constructed languages.

This may be a model for one way to allow people to extend WALS by adding new languages.

4 Responses to “How do you use WALS data?”

  1. Thomas Hart Chappell Says:

    I have used WALS.info to come up with a list of 67
    languages in the WALS sample that have “free word-order” in
    5 or more of WALS’s 14 word-order features.

    I have looked at what WALS has to say about the 25
    languages in its sample that have “free word-order” in 6 or
    more of those features, to find out what correlates in
    morphology, nominal categories, nominal syntax, simple
    clauses, and verbal categories, go with “free word-order”.
    (Unsurprisingly, double-marking and polypersonal agreement
    top the list; but for the highly “free” languages, so do
    multiple genders and multiple cases.)

    I have looked at WALS.info to investigate the relationship
    between VO word-order and languages with much prefixal
    morphology on the verb and much suffixal morphology on the
    noun; and between OV word-order and prefixal morphology on
    nouns and suffixal morphology on verbs. (There’s not a lot
    of relation between where nominal affixes go and order of
    verb and object to be supported by WALS; but WALS does seem
    to support the idea that in VO languages verbal morphology
    is prefixal and in OV languages verbal morphology is
    suffixal.)

  2. Chris H. Says:

    Forget using it to catalog conlangs… I’ve used it to BUILD them. I assembled a spreadsheet of every feature that made reasonable sense to me and reformatted the info in terms of percentages, out of a hundred. For example, for the number of genders chapter, it became

    None: 1-57
    Two: 58-76
    Three: 77-86
    Four: 87-91
    Five+: 92-100

    Then, using a random number generator (http://random.org/) I generated a number between 1-100 for every category (about 60, all told), matching the random number to each subset of each feature. For example, generating a 69 on the above list would render a conlang with two genders, while a 2 would mean none. In total, it represents a LOT of information about a conlang, and based not on individual predilections or Indo-European leanings but on good, solid statistics.

  3. KANEKO Tohru Says:

    We use your atlas in our seminar examining the data by testing esp. with indigenous languages in the East Eurasian Continent. Sometimes we find the picture is too beautiful to accept the detailed structure investigation, e,g, the chapter 26 affixing.

    Anyway I remind me of your discussion as I attended when I was in the Leipzig University 1999-2000.
    Best Greeting to Martin Haspelmath and Bernaerd Comrie!

  4. Steffen Schaub Says:

    I have just started working on my master thesis on Word Order in the Native American languages (north of Mexico), and I am using the data in WALS as a starting point. I collect the existing information in WALS for the features

    – 81 (Word order S, V, O)
    – 82 (Order S,V)
    – 83 (Order O,V)
    – 86 (Genitive, Noun)
    – 87 (Adjective, Noun)

    and try to research these features for those languages where WALS is still missing data.

    I would be glad for suggestions or words of advice from experienced typologists.

Leave a Reply

*