Thursday, September 08, 2011

Why German is strange

Following up on comments to the previous post, some readers may be interested in the following list of the top ten rarest typological features of Northwestern European languages (on WALS), ordered from most to least unusual:
  1. Polar Questions - coded through word order (Did he? He did.); very unusual outside Europe.
  2. Uvular Consonants - continuants only (French/German/Dutch "r"); usually languages with uvulars have a uvular stop.
  3. The Perfect - coded with a word meaning "have" (I have done it); unparalleled outside Europe.
  4. Coding of Evidentiality - using a modal verb; unusual outside Europe
  5. Demonstratives - no distance contrast (German); rare worldwide.
  6. Negative Indefinite Pronouns - used without a predicate negator (I saw nothing, instead of I ain't seen nothing); very rare outside Europe.
  7. Front Rounded Vowels - high and mid (ü, ö); unusual outside northern Eurasia
  8. Relativization on Subjects - using a relative pronoun; most of the world's language use non-pronominal strategies.
  9. Weight-Sensitive Stress - Right-oriented, antepenultimate involved; unusual.
  10. Order of Object and Verb - alternates depending on clause type (German, Dutch); most languages keep this fixed irrespective of clause type.
This is from: Cysouw, Michael. 2011. Quantitative explorations of the world-wide distribution of rare characteristics, or: the exceptionality of northwestern European languages. In: Horst Simon & Heike Wiese (Eds.) Expecting the Unexpected. Berlin: De Gruyter Mouton. 411-431.

For a list of some linguistic features common in Europe more generally but rare outside it, see Haspelmath 2001 (summarised here.)

18 comments:

  1. Another strangeness is that German has no coding of aspect at all (I think it was David Marjanovic who pointed this out at LH).

    ReplyDelete
  2. David Marjanović08 September, 2011 21:19

    :-)

    I suppose the phonemic /pf/ is so strange it's not even in the WALS in the first place? :-)

    Pretty often, I'm still not sure whether to better use this or that in English. Fortunately for me, French has gone pretty far in the German direction.

    The Perfect - coded with a word meaning "have" (I have done it)

    Of course, some verbs take "be" instead of "have". And calling it "perfect" is silly (see below), though that's what everyone does...

    Front Rounded Vowels - high and mid (ü, ö); unusual outside northern Eurasia

    To be fair, within northern Eurasia, they extend from sea to shining sea, and then south all the way to Hongkong.

    It's funny how they're all over Germanic now but were completely absent from Proto-Germanic, Gothic, and the runes.

    Weight-Sensitive Stress - Right-oriented, antepenultimate involved

    What do you mean?

    Another strangeness is that German has no coding of aspect at all

    There is the rheinisches Verlaufspassiv – in parts of western Germany, "I am/was working" is increasingly often expressed as ich bin/war am Arbeiten, and some linguists now say this should be put in the textbooks –; but it's not spreading, as far as I can see.

    Then there are the two different passives: with "become" for processes, and with "be" for states (like English "it is done"). Does that count? The "be" passive strikes me as rare...

    All other aspect-like features are lexical. Examples off the top of my head:

    – The verb prefix er-, which indicates successful completion*, is hardly productive, if at all.

    – In Germany (roughly speaking), verbs of motion (roughly speaking) form the so-called perfect with "be", and all other verbs form it with "have". In Austria, this has been reinterpreted: verbs that describe states take "be", verbs that describe actions take "have". So, "stand"**, "sit" and "lie" take "be" in Austria. When people (including myself) use "be" with "sleep", they always immediately correct themselves, but it happens way too often to be random; I give "have slept" a few more decades. I'm sure the fact that einschlafen "fall asleep" takes "be" (even in Germany; it's a motion into sleep, I suppose) helps.

    * Specifically successful completion. It's not simply "perfective aspect".
    ** Famously (in Austria), this allows Austrians to distinguish stehen "to stand" and gestehen "to confess a crime" in the "perfect": ich bin gestanden "I stood", ich habe gestanden "I confessed". In Germany, the latter form is used for both.

    I think it was David Marjanovic who pointed this out at LH

    I've probably done that, but it's not my discovery. :-) There's a lot of literature on it, and I haven't read most of it.

    ReplyDelete
  3. David Marjanović08 September, 2011 22:36

    Oh. The paper explains what it means by evidentiality. :-)

    Interestingly, it doesn't mean the uvular r when talking about "uvular continuants", at least in the text. It means the voiceless fricative, present in Dutch, northern (as an allophone) and southwestern German, and Yiddish.

    I still don't understand the stress thing. In German, syllable weight depends more on stress position than the other way around.

    ReplyDelete
  4. David Marjanović09 September, 2011 01:18

    I tried to read up on the stress question in WALS and failed. Maybe it's too late at night, but I don't understand it.

    Anyway, here's another strange characteristic that's so strange it's not in WALS: while German does distinguish adjectives and adverbs, it* treats "be" as a verb, so it goes with adverbs, not adjectives. Adverbs are marked by the lack of a gender/number/case/definiteness ending, so they look like English adjectives.

    * Exception: the Highest Alemannic dialects of central Switzerland and beyond.

    ReplyDelete
  5. I don't think that's German treating "to be" as a verb and thus using an adverb - I think it's just that predicative adjectives don't use endings.

    For example, you get the same endingless thing after other verbs which would take predicative adverbs, such as "become" ("Er wird groß") or "consider" ("Ich finde sie hübsch").

    Compare the difference between "Ich finde ihn schnell" = I find him quick, I consider him to be quick, I think he is quick and "Ich finde ihn schnell" = I find him quickly, my search for him was fast. I think that in the former case, "schnell" is an adjective, not an adverb, despite having no inflectional ending, and that this is the form in sentences such as "Der Baum ist groß".

    ReplyDelete
  6. David Marjanović09 September, 2011 17:38

    I don't think that's German treating "to be" as a verb and thus using an adverb - I think it's just that predicative adjectives don't use endings.

    Point taken.

    That just doesn't take any of the weirdness away. What other languages are there where predicative adjectives look like adverbs and don't look like normal adjectives? The WALS doesn't say.

    (There are of course languages where all three look the same. Mandarin is one, and I bet that's very common outside Indo-European.)

    BTW, I looked it up, /pf/ isn't in the WALS.

    ReplyDelete
  7. That just doesn't take any of the weirdness away.

    True.

    ReplyDelete
  8. In Haspelmath-speak, capitalized "Perfect" means a language-specific notion conventionally labeled that way, whether or not it's connected with the universal semantic notion of the perfect. Since "perfect" is what it's called in German grammar books, "Perfect" is what Haspelmath calls it.

    Of course this distinction doesn't work in German — I wonder what he does there, or maybe he only writes professionally in English anyway. It's a good thing Chomsky didn't wind up working mostly on German (in a different world where his parents never left Russia?), or generative syntax would be even stranger than it is!

    As for the be-passive, it's no passive at all, any more than in English The door was closed when talking about a state, which is the usual case.

    ReplyDelete
  9. David Marjanović10 September, 2011 00:46

    Of course this distinction doesn't work in German — I wonder what he does there, or maybe he only writes professionally in English anyway.

    In German, that tense is called Vergangenheit ("past") or... wait for it... Perfekt.

    The tense that looks like the English past tense is called Mitvergangenheit (an awkward made-up word without a transparent meaning; "co-past"???), Präteritum ("past" – quite literally "passed", in Latin) or, and I hope that's no longer done, Imperfekt in a really desperate attempt to interpret the Latin aspect distinction into German. Especially desperate because it's backwards in one respect: like in English, this is the default time for narration, like the Latin perfect and very much unlike the Latin imperfect.

    The best analogues are actually the French passé simple and passé composé.

    As for the be-passive, it's no passive at all

    Well, yeah, but that's what the textbooks call it: Zustandspassiv ("state passive") as opposed to Vorgangspassiv ("process passive", the one with werden "become").

    Of course, the textbooks of German also insist that "it's true" in "I know it's true" is an "unintroduced dependent clause" (uneingeleiteter Gliedsatz), even though the verb in it comes second rather than last like in an ordinary independent clause. That's simply because if you add an introductory conjuction, "I know that it's true", the verb suddenly does go last (ich weiß, dass es wahr ist/dass es stimmt).

    ReplyDelete
  10. Ah, I wasn't clear enough.

    Haspelmath and his school make an explicit distinction between "perfect" and "Perfect", "present" and "Present", "adjective" and "Adjective", etc. The first names the universal function, the second names the form usually or traditionally so named in the language under discussion. Thus in contemporary English the Present expresses the present only for stative verbs; for active verbs, the Present expresses the habitual, and the present must be expressed by the Present Progressive (which can also express the near future). In German, of course, the Present expresses the present for all verbs. This convention is typographically disturbing, at least to my eyes, but not as confusing as if the capitals were left off altogether.

    My somewhat jocular point was to wonder how this distinction could be expressed when doing linguistics in German, since obviously capitalization is not available. You could go with PERFEKT, PRÄSENS, ADJEKTIV, usw., but my ears would be hurting by the bottom of the second page. The alternative, of course, is to do all your linguistics in English.

    ReplyDelete
  11. Etienne here. Interesting paper and comments. On page 13 the author suggests that in some cases rare features arising separately in languages in contact will come to reinforce one another. I would like to propose that front rounded vowels in Modern Germanic languages and French/Gallo-Italian varieties may be an instance of this. This is because in the relevant Romance varieties the shift from Latin /u:/ to /y/ was unaffected by the quality of the vowel in the following syllable, whereas in the relevant Germanic varieties the sound change to /y/ was triggered by an /i/ in the following syllable.

    But it is easy to see how contact might have aided both changes.If, as I suspect, it started in Romance, we can visualize this: in the process of shifting to a Germanic variety which has /y/ as an allophone of /u/ it would be easy for Romance speakers (with an /y/ phoneme) to perceive this /y/ as a separate phoneme, thereby contributing to making it a phoneme in the Germanic variety being shifted to.

    ReplyDelete
  12. David Marjanović12 September, 2011 01:12

    In German, of course, the Present expresses the present for all verbs.

    (And the habitual, and the future at least when that's not too confusing.)

    My somewhat jocular point was to wonder how this distinction could be expressed when doing linguistics in German, since obviously capitalization is not available.

    A genuine problem! In a translation of a Dave Barry book, I've seen Italics And Capitalization Of Every Word...

    Etienne here.

    Why don't you simply comment using the "Name/URL" option? That's what I do. You don't have to enter a URL...

    On page 13 the author suggests that in some cases rare features arising separately in languages in contact will come to reinforce one another.

    I think this happens a lot.

    Off the top of my head:

    – Basque has merged /w/ and /b/. Spanish has merged /v/ and /b/. In both cases, the result is the same. And some wonder if Gothic has played a role; word-final devoicing regularly turned Gothic b into f, never into p.
    – Basque (like Gascon or part thereof) has shifted intervocalic /l/ to /r/; the previous /r/ got out of the way by being lengthened, so it stayed a separate phoneme. Basque now has one single length contrast in its sound system. While Latin consistently distinguished long and short consonants and Italian still does, Spanish has lost all of these contrasts... except for the one between long and short /r/. There is thus one single length contrast in the sound system, and it's the same one as in Basque.

    we can visualize this: in the process of shifting to a Germanic variety which has /y/ as an allophone of /u/ it would be easy for Romance speakers (with an /y/ phoneme) to perceive this /y/ as a separate phoneme, thereby contributing to making it a phoneme in the Germanic variety being shifted to.

    In this case, isn't the traditional explanation (apocope and mergers of unstressed vowels removed the conditioning factors and left the former allophones stranded) enough?

    It's sad, though, that nothing seems to be known about the language(s) of the Romance-speakers that persisted in a few cities and villages all over Austria for centuries after Odo(w)aker had ordered all Romans to move south of the Alps.

    ReplyDelete
  13. David: Oops, hadn't realized that in the Name/Url option the url was optional. Well, to misquote a certain STAR TREK character, "I'm a linguist, not a computer geek, dammit".

    So, back to being a linguist...the reason why a contact explanation seems required in the case of the rise of /y/ on both sides of the Romance/Germanic border is because such a phoneme is cross-linguistically rare: add to that the fact that the phoneme arose on both sides of the border at about the same time, when both language families were in intense and intimate contact (it's often forgotten that everything South of the Danube and West of the Rhine which is now Germanic-speaking was once Romance-speaking), and some kind of contact explanation seems likely.

    As for the Basque-Romance instances you mention: I'd be careful here. We have little evidence indicating what Basque was like before it came into contact with Latin: hence, if we find features whereby Basque and neighboring Romance varieties, *today*, align with one another and stand opposed to other Romacne varieties, this *might* indicate that Basque influenced its neighbors. It could also indicate that an innovation which began in a neighboring Romance variety spread to its neighbors, including Basque. It has been shown that some non-Latin vocabulary in Basque and its modern neighbors cannot be indigenous to Basque and must have entered Romance first (from some other, now-extinct language) and thence spread to Basque.

    As for the now-extinct Romance varieties of Austria, Bavaria and other now-germanicized areas: actually, we can make some educated guesses: on the basis of loans and place-names many of their salient phonological innovations can be established, and these often link these extinct varieties to living ones (the extinct Romance variety of the Moselle, for instance, seems to have shared several phonological innovations with its Lorrain and Walloon neighbors).

    ReplyDelete
  14. David Marjanović13 September, 2011 01:01

    the reason why a contact explanation seems required in the case of the rise of /y/ on both sides of the Romance/Germanic border is because such a phoneme is cross-linguistically rare:

    Well, it's very common in northern Eurasia... of course, this is a contact explanation.

    add to that the fact that the phoneme arose on both sides of the border at about the same time

    Absolutely.

    ReplyDelete
  15. Etienne, to what extent is the development of /y/ a part of the general fronting in French? Or is thta general fronting not realy so general, really it is all conditioned by a /i/ somewhere?

    Something else, that fronting of/u/ to /y/ occurred in Welsh to and something like that, by way diphthongization, is goingon in Irish and varieties of English influenced by it. I don't know about the situation in Welsh, where the only example that comes ot mind is Boudicca -> Buddig, where the /u/ is now pronounced as /i/, but /i/ is not conditioning it in Irish.

    ReplyDelete
  16. Jim: I'm not sure what other types of "fronting" you're talking about: I have seen the shift of Latin /kt/ to /jt/ referred to as part of a "palatalizing/fronting" tendency, but this shift covers a much broader spectrum of Romance languages/dialects than the /u/ to /y/ shift, so I don't quite see how they could both be linked.

    ReplyDelete
  17. Etienne, I mean where you find honore > honneur, aria > aire... wait - I see that I am having a hard time thinking of examples that don't involve an -i. I wonder if there are any.

    ReplyDelete
  18. David Marjanović20 September, 2011 18:52

    Over on Language Hat we just ran across more German strangeness: compounds that contain a meaningless -s-, -n- or -en-. Historically these elements descend from genitive endings, but their distribution is now chaotic – lexically determined and with plenty of regional variation. What's going on in my head is that all nouns in a compound except the last are turned into prefixes, and the prefix form of a noun may look like the nominative singular, the genitive singular, the nominative plural or occasionally something else.

    Donau-0-dampf-0-schiff-0-fahrt-s-gesell-0-schaft-s-kapitän-s-kajüte-n-schlüssel-0-bund...

    Case in point, -schaft (a suffix for abstract nouns, much like English -ship) is feminine, so it never takes -s as an ending in its declension, yet its prefix form has -s-.

    ReplyDelete