Linguistic Maps.

Y sent me a link to Matt’s 2017 post New Approaches to Ethno-Linguistic Maps at the linguistics blog Humans Who Read Grammars, calling it “neat,” and I felt the same way when I clicked the link:

One major issue with most modern maps of languages is that they often consist of just a single point for each language – this is the approach that WALS and glottolog take. This works pretty well for global-scale analyses, but simple points are quite uninformative for region scale studies of languages. Points also have a hard time spatially describing languages that have disjoint distributions, like English, or languages that overlap spatially. […] I believe that, thanks to greater computational efficiency offered by modern computers and new datasets available from social media, it is increasingly possible to develop better maps of language distributions using geotagged text data rather than an expert’s opinion. In this blog, I’ll cover two projects I’ve done to map languages – one using data from Twitter in the Philippines, and another using computationally-intensive algorithms to classify toponyms in West Africa.

Those maps are amazing! Then I thought “I should really investigate that blog,” and when I went to the main page I found a 2019 post by Annemarie Verkerk, Language family maps, that begins:

Last week, I assigned Bernhard Comrie’s (2017) chapter ‘The Languages of the World’ (from The Handbook of Linguistics, 2017) to a class. It’s a basic overview of the world’s language families, which is what I wanted them to read, but for one thing: there are no maps in it. I overcompensated in class by presenting a 30-item list of maps, because some things are just so much easier to understand using visual representations. I decided to post some of the best ones I could find here, for future reference and in order to invite you to post better ones in the comments.

It’s a very useful resource, as is the entire blog (here’s the About page). Thanks, Y!

Comments

  1. David Eddyshaw says:

    Horrifying to find that life-and-death decisions (literally) have been taken on the basis of Ethnologue:

    http://humans-who-read-grammars.blogspot.com/2014/10/scandal-linguistics-used-horribly-wrong.html

    The blog also seems to suggest that there has been published academic work based on the profound misapprehension that the (perfectly nice) Murdoch map of African languages

    https://upload.wikimedia.org/wikipedia/commons/4/49/Africa_ethnic_groups_1996.jpg

    is in some way complete (it may not be as bad as that: the articles linked are all paywalled.)

  2. David Eddyshaw: What did you think of the West African placename maps? Any surprises?

  3. David Eddyshaw says:

    No great surprises in the outcomes, though I’d like to know more about the methodology. In particular, ama seems an odd string to pick. A bit like doing the same thing in Europe with rde. It makes me wonder how many strings were tried before something came out that more or less matched the accepted language families.

    Large parts of inland West Africa share a lot of political and cultural history, so it’s not astonishing a priori that toponyms would show a lot of similarities over wide areas. Mooré and Gurmanche, for example, which cover much of the eastern part of Burkina Faso between them, share the element teŋa “village, town”, which unsurprisingly turns up in quite a few toponyms. All the more so for the huge area where descendants of the koine of the Empire of Mali are spoken.

    Most Ghanaian place names have fairly transparent meanings, especially those in the Kusaal-speaking areas; as you’d expect, given that villages and towns are relatively recent developments there. It’s different in the parts of West Africa where cities are much older: AFAIK the names of the Hausa Bakwai, the seven core Hausa city-states, don’t mean anything in Hausa apart from themselves.

    This is nice site for Ghanaian place names, maintained by John Turl as a labour of love:

    https://sites.google.com/site/ghanaplacenames/

  4. Ooh, very nice indeed! “The word Accra is derived from the word ‘nkran’ meaning ‘ants’ in Akan, a reference to the numerous anthills seen in the countryside around Accra.”

  5. The Ghanaian site is great. I can’t tell though if they have any way of sorting out folk etymologies. Does the name of Lake Bosomtwe really derive from ɔbosom ‘god’ + ɔtwe ‘antelope’?

  6. David Eddyshaw says:

    The classification of West African languages implied by the discussion is basically obsolete; the Senufo languages are not now thought to be “Gur”, even by those enthusiasts (unlike me) who believe that Gur is a valid subgroup in itself; and “Kwa” is not now thought to be a thing; the Akan languages don’t form a node with (say) Yoruba, which is probably actually closer to Bantu, despite the profound typological differences. Senoufo definitely isn’t particularly close to the Akan languages. Senoufo as a whole has been profoundly influenced by Mande, but any genetic connection between Senoufo and Mande is either very remote or nonexistent.

    In general, a methodogical weakness is that the creators of the map perhaps don’t appreciate sufficiently that multiethnic states have been a feature of West Africa for many centuries. West Africa has had politics for as long as Europe. It’s not hard to find places named in languages different from those used by their actual inhabitants (the Kusaasi area has, for example, “Garu”, from the Hausa for “city wall.”)

  7. Doesn’t it say that ama wasn’t a useful trigram because it was too widespread? Kro is the example of one that was useful.

  8. David Eddyshaw says:

    I can’t tell though if they have any way of sorting out folk etymologies

    A very good point. I can say that John Turl (who contacted me a while back about Kusaal names, and who communicates with me from time to time) is very aware of such issues, and adopts a fairly sceptical approach on the whole. It’s naturally more of an issue in regions where names are less transparent than in the Kusaasi area.

    One of the interesting things I’ve discovered from the correspondence is that quite closely related cultures can have surprisingly different approaches to creating place names. Kusaasi place names are mostly straightforwardly based on local geographical features of note, for example, whereas Mamprussi placenames often seem to commemorate some notable event in the history of the place; Mamprussi place names often have the form of complete statements, whereas Kusaasi placenames rarely do.

  9. David Eddyshaw says:

    @Keith Ivey:

    Yes, I wasn’t reading carefully enough.

    Akan (Twi) kurow is “town, village”, so no surprise there about the distribution.

  10. David Marjanović says:

    Horrifying to find that life-and-death decisions (literally) have been taken on the basis of Ethnologue:

    To be fair, the purpose of asylum bureaucracies is to deny every single asylum application, so they’ll reach for absolutely anything to justify that.

  11. David Eddyshaw says:

    Still, it looks better when they pretend harder.

  12. David Eddyshaw says:

    I spoke too soon about the Murdoch map, now that I look at the key in detail. A very non-standard classification (“Bantoid” doesn’t mean what he thinks it does at all*) and many outright errors. Still, it looks pretty.

    *To describe Dagbani (or “Dagomba”, as he calls it) as “Bantoid” is precisely as accurate as calling French “Slavoid”, for example. The word has an accepted (somewhat controversial) meaning in real African linguistics:

    https://en.wikipedia.org/wiki/Bantoid_languages

  13. John Cowan says:

    precisely as accurate as calling French “Slavoid”

    Or calling Baltic “Sand Slavic” (with reference to “Grasslands Bantu”).

  14. January First-of-May says:

    Or calling Baltic “Sand Slavic”

    Meh, probably more accurate than “Baltic”. Most of the languages of the Baltic Sea area are Germanic (in the west) or Finnic (in the east and north); the “Baltic” ones occupy only a relatively small area in the southeast.

    In general, a methodogical weakness is that the creators of the map perhaps don’t appreciate sufficiently that multiethnic states have been a feature of West Africa for many centuries. West Africa has had politics for as long as Europe. It’s not hard to find places named in languages different from those used by their actual inhabitants

    I wonder what would such a method produce for Europe.
    …On second thought, it probably won’t be entirely consistent with modern national borders (e.g., Alsace will pattern with Germany), but for the most part it would, given the longstanding European tradition of mildly divergent national versions of the same toponym (Temesvar and Timisoara would pattern very differently).
    OTOH, Russia (even European Russia) would probably be interesting, even if (perhaps especially if) the method would result in divisions that hadn’t reflected the situation on the ground in centuries (if ever).

  15. David Marjanović says:

    I wonder what would such a method produce for Europe.

    Well, that’s been done. There are Slavic place names and hydronyms all over the eastern halves of Germany and Austria, for example, and Celtic ones all over the place (the Isar is the river through Munich). And then there’s a sizable number of wholly unexplained names.

  16. January First-of-May says:

    Well, that’s been done.

    I know that much; the question was what such a simplistic method would produce for Europe, and to what extent it would correlate with actual toponym origin (to the extent that it is known).

  17. January First-of-May says:

    Akan (Twi) kurow is “town, village”

    Now that I’ve seen this with fresh eyes, I immediately remember the Polish town (or possibly village) of Kurów, which managed to get Wikipedia articles about it in something like 200+ languages. (236, it turns out. This used to be the record for a Wikipedia article at some point, but it’s no longer in the top hundred.)

    Turns out that Twi is in fact among those 200+ languages, but the article is quite underwhelming.

    I wonder if there would ever be a Kusaal Wikipedia…

  18. John Cowan says:

    the “Baltic” ones occupy only a relatively small area in the southeast

    True. But then again the Afroasiatic languages occupy only a relatively small area of Africa, never mind of Asia. “Names are but names.” I think the logic must have been that the Baltic languages are spoken nowhere else.

  19. David Eddyshaw says:

    the Afroasiatic languages occupy only a relatively small area of Africa

    I wouldn’t have said so myself:

    https://en.wikipedia.org/wiki/Languages_of_Africa#/media/File:Map_of_African_language_families.svg

  20. Yes, “relatively” is doing a lot of work there.

  21. John Cowan says:

    Well, fine, but a lot of that is uninhabitable desert. Besides, I have my suspicions about that projection: it looks rather Mercatorish, though I am hardly able to judge from appearance.

  22. David Eddyshaw says:

    I like the captions on that map. “Possibly a family”; “some areas may not belong.” I recognise a kindred spirit at work.

  23. January First-of-May says:

    Well, fine, but a lot of that is uninhabitable desert.

    I had recently finished(-ish) compiling a (mostly best-guess based) list of the largest (most populated) settlements in each 5×5 degree square [not actually square in most projections] of latitude and longitude. It turned out that a few areas that I expected to contain some settlements did not in fact have any.

    In particular, for two squares in the Sahara, the “largest settlements” were not even properly settlements as such: the military outpost of El Mreiti, Mauritania, and the desert irrigation project of Sharq El Owainat, Egypt.

  24. David Marjanović says:

    I think the logic must have been that the Baltic languages are spoken nowhere else.

    I’ve always taken for granted that the sea is named after the people, not the other way around.

  25. Trond Engen says:

    Baltic and (Store/Lille/Fehmarn) Belt are probably related, but exactly how is less clear than you’d think at first glance.

  26. gwenllian says:

    This used to be the record for a Wikipedia article at some point, but it’s no longer in the top hundred.)

    Turns out that Twi is in fact among those 200+ languages, but the article is quite underwhelming.

    I’m sure everyone will be glad to know Corbin Bleu’s path to dethroning Jesus Christ and achieving total Wikipedia domination is continuing apace, with an (admittedly similarly underwhelming) article in Twi about him, and none about his other main rivals, Barack Obama and Confucius.

  27. David Eddyshaw says:

    Corbin Bleu

    The Twi page impertinently tells me that I need to specify a valid dialect (which does not in fact seem to be possible), although the “Akan” page does grudgingly vouchsafe a sentence or two of actual information. Poor show, I say. How are southern Ghanaians supposed to keep up with current thinking in Corbinism? Northern Nigerians are equally ill-served …

  28. John Cowan says:

    As far as the record shows, the sea was named first, then the language and people: L Balticus ‘dweller by the Baltic Sea’. Wikt gives three etymologies s.v. Baltic:

    From North Germanic *balta (“straight”),[spurious reconstruction?] in reference to the narrow entranceway of the sea

    Related to Lithuanian baltas (“white”) and Proto-Slavic *bolto (“swamp, bog, mud”), which is from Proto-Indo-European *bʰel- (“white”)

    Related to Latin balteus (“belt”) (compare Proto-Germanic *baltijaz), referring to the Danish straits, “the Belts”. This is suggested by Adam of Bremen, who in the 11th century first recorded the name (Balticus, eo quod in modum baltei longo tractu per Scithicas regiones tendatur usque in Greciam).

  29. David Marjanović says:

    Latin balteus (“belt”) (compare Proto-Germanic *baltijaz)

    …which one of these is a loan from the other? They can’t be cognate.

    Proto-Slavic *bolto (“swamp, bog, mud”)

    Semantically perfect (perfect, I say), but geographically odd.

  30. John Cowan says:

    Semantically ‘white’ and ‘shallow water’ are associated in central IE, it seems; Wikt thinks it is because of the unmoving reflective surface of such water. In modern Latvian baltas is ‘mud(dy)’ as well as ‘Balt(ic)’, and in Albanian baltë is also ‘mud, swamp’. So I should have said Baltic was Mud Slavic rather than Sand Slavic.

  31. Trond Engen says:

    The usual derivation is belte “belt” < ON belti n. < PGmc *baltija(z) < Lat. balteus “sword belt” (possibly < Etr.).

    The complexities as I understand them:

    The word belti is neuter in ON, but *baltijaz is very much masculine in form. The alternative *baltija might work for belte, but that means an unexplained loss of Latin final s.

    There’s another Gmc. word represented by HG Balz/Palz m. “belt”, suggesting PGmc *baltaz, which is not so easily derived from the Latin word, and which yields neither Belt.nor belte in NGmc.

    At the face of it, belte looks like a derivation from a noun or adjective *balt- (or from a verb derived from it), parallel to feste n. from fast a. or feste v.

    Belt (Da. Bælt) without the final e doesn’t really work as a derivation from either of the suggested PGmc forms but rather yet another form *balti(z). This, incidentally, works better than *baltaz with the Latin source and could also be the source of the derivation of belte. Alternatively, it could be a back-formation from an umlauted i-stem plural of balt-. After all, “the Belts” often occur in the plural.

    Lat. balticum could be formed as an adjective to any of these PGmc formations, so not much help there.

    What I think makes most sense is that the Latin word (or a borrowing through some other language?) became PGmc *baltiz m. “type of wide belt”. This yielded forms with and without umlauted root vowels in the daughter languages, and also came to mean “straight, sound” in some coastal varieties. In North Germanic a secondary form belti n. “beltening” replaced it in the meaning “belt”.

  32. Trond Engen says:

    I forgot to add that belt “sound connecting the Kattegat with the Baltic” and belte “belt” could be unrelated, but one would still have to account for the phonological shape of belt. I don’t see anything in Baltic or Slavic that would encourage an i-stem in Germanic — unless it’s contamination from the near-homonym..

  33. Lars (the original one) says:

    For bælt ODS says “possibly a Wendish word for ‘sea,’ related to Old Slavic blato ‘lake’; according to others, related to bælte.”

    It’s attested from 1288, and I have a hard time seeing how people would look at a 17km wide body of water (where Storebælt is narrowest) and go “it looks like this thing that holds up my trousers.” The analogy is much more obvious when looking at a map but the first one was printed in 1570.

  34. Lake Biwa in Japan was said, in the 14th century, to be named after the musical instrument of the same name. The lake is 60 km long and up to 23 km wide. I don’t think they had maps then, the way we think of them nowadays.

  35. Proto-Slavic *bolto (“swamp, bog, mud”)

    Whence Lake Balaton in Hungary.

  36. Lars (the original one) says:

    @Y, that’s interesting. But Japan was a lot further ahead than Scandinavia.

    For comparison, the sagas tell that the witch Gefion / goddess Gefjun ploughed free an area of land that became Zealand and left lake Mälaren behind. But they look nothing like each other on a map (not even when correcting for the 5-10m change in land levels around Stockholm since then) so later redactions changed it to lake Vänern which is sort of a fit if you squint.

    I don’t know what the sagas call Storebælt, but my point is that knowledge of the shape of large geographical features was something that came later in Scandinavia.

Speak Your Mind

*