Author Topic: Expérience en grammaticalité  (Read 15746 times)

Mandos

  • Senior Member
  • *****
  • Posts: 5759
    • View Profile
    • http://politblogo.typepad.com/
Expérience en grammaticalité
« on: February 08, 2007, 12:21:38 AM »
Hello.  I had an idea while discussing something with a roomfull of colleagues and sketched it out on a paper napkin.  It has to do with French grammar, and by extension, human grammar.  And I'd like what we call in my business "grammaticality judgements" from French-speakers who are more genuine native speakers than I.

First, a reasonable amount background so you know what kind of answer I want.  Let's take a look at these three English sentences.

1. John has not eaten his fish, but I have.
2. John has not eaten his fish, but I've eaten my chicken.
3. *John has not eaten his fish, but I've.

To most native English speakers, 1 and 2 are grammatical.  Not in the textbook sense, but in the sense of what they learned from childhood.  Here we are using "grammatical" as a psychological descriptor, or better yet, a sensory descriptor analogous to, say, smell.  

Number 3 is bad.  It has a bad grammaticality "smell" to most native English speakers.  This is an odd fact, because "I've" is just a contraction of "I have".  It's legitimate to use it in that context, as we see in 2, EXCEPT that 2 does not have a deleted predicate.

So the puzzle is, why does this difference between 2 and 3 have the effect of invalidating 3, when everything else is the same?  Solutions to this class of problems (of which this is an example) have profound consequences for our understanding of grammar.

It happens that most of this kind of research into this specific type of phenomenon has been done largely in English, partly an accident of history that this type of work was developed in anglo-American environments, and partly because English provides these convenient and telling subject-auxiliary contractions ("He's", "I've", ...).  But to get a reliable picture of what's going on, it's good to have support from other languages.

Most of my colleagues with me speak languages other than English as their native language, but most of the languages available don't have these convenient contractions to test out these phenomena with.  I knew enough French, however, to wrack my brains coming up with a set of examples.  Here they are:

1. Il a mangé le gâteau, et je ne l'ai pas.
2. Il a mangé le gâteau, et j'ai aussi.
3. Il a mangé le gâteau, et j'ai.
4. Il a mangé le gâteau, et j'ai mangé le croissant.
5. Il a mangé le gâteau, et je n'ai pas.
6. Il n'a pas mangé le gâteau, et j'ai.

7. Il a marché au dépanneur, et j'ai aussi.
8. Il a marché au dépanneur, et j'y ai aussi.
9. Il a marché au dépanneur, et je n'ai pas.
10. Il a marché au dépanneur, et j'ai.
11. Il a marché au dépanneur, et je n'ai pas.
12. Il a marché au dépanneur, et je n'y ai pas.

So, it would be a big favour if someone could tell me which ones are grammatical (in the above sense) to you, and which do not.  Relative judgements of ungrammaticality are OK.  (ie, X is better than 9, but not great.)  Just keep it natural.  

Thanks.

lagatta

  • Senior Member
  • *****
  • Posts: 13093
    • View Profile
Expérience en grammaticalité
« Reply #1 on: February 08, 2007, 05:45:02 AM »
Maybe later. Right now that seems too much like work. I'm checking an English translation of a text translated from German into French - by a German, who speaks French fluently but has some odd syntax, against the original German text. My German is not good enough to do a straight translation from the original, but I can check the two.
" Eure \'Ordnung\' ist auf Sand gebaut. Die Revolution wird sich morgen schon \'rasselnd wieder in die Höhe richten\' und zu eurem Schrecken mit Posaunenklang verkünden: \'Ich war, ich bin, ich werde sein!\' "
Rosa Luxemburg

skdadl

  • Global Moderator
  • Senior Member
  • *****
  • Posts: 32874
    • View Profile
    • http://www.pogge.ca
Expérience en grammaticalité
« Reply #2 on: February 08, 2007, 05:52:59 AM »
NB: Sorry to be picky, but I have to point out that #1 is a bad sentence on grounds of logic, if not structure. "I have" is ambiguous there -- it could mean "I have eaten John's fish" (structurally more likely, because John's fish is the only antecedent object) or it could mean "I have eaten my fish" (logically more likely).

Your structural point is correct, though. A better example for #1 would be:

"John has not arrived at the office yet, but I have."

I would say of #3 that you could hear that from some dialect speakers, and I'm sure that Cole Porter produced at least a few of those in the service of witty rhymes.

Mandos

  • Senior Member
  • *****
  • Posts: 5759
    • View Profile
    • http://politblogo.typepad.com/
Expérience en grammaticalité
« Reply #3 on: February 08, 2007, 07:53:29 AM »
skdadl: You are right to point out the distinction between "structural" and "logical" correctness. It is the former that I'm after.  The referential ambiguity is an interesting question in itself, and it is possibly related to what *kind* of structure is in the deleted portion. but I don't think it's too relevant to my specific question.

Situations where people are deliberately futzing with the grammar (as in much poetry) to achieve a particular sonic effect must necessarily be abstracted out of this analyses.  You can generate any string you like, but most people will recognize a "ground state" for sentence structure.  Dialect is a more important point, but if such people exist then I suspect I can come up with examples that distinguish their idiosyncracies from the contractions that most English speakers use.

skdadl

  • Global Moderator
  • Senior Member
  • *****
  • Posts: 32874
    • View Profile
    • http://www.pogge.ca
Expérience en grammaticalité
« Reply #4 on: February 08, 2007, 08:10:27 AM »
*slight drift*

Mandos, I have a question to ask about the use of the word "morphology."

You know how people talk now about "memes"? Would it be correct to say that when people label something a meme, they are identifying underlying morphological parallels with other arguments -- ie, that all arguments that promote that meme are morphologically parallel?

*/slight drift*

brebis noire

  • Senior Member
  • *****
  • Posts: 4707
    • View Profile
Expérience en grammaticalité
« Reply #5 on: February 08, 2007, 08:22:06 AM »
Quote
1. Il a mangé le gâteau, et je ne l'ai pas.
2. Il a mangé le gâteau, et j'ai aussi.
3. Il a mangé le gâteau, et j'ai.
4. Il a mangé le gâteau, et j'ai mangé le croissant.
5. Il a mangé le gâteau, et je n'ai pas.
6. Il n'a pas mangé le gâteau, et j'ai.

Though I'm not a native speaker, I have developed a native ear for French, and only no. 4 makes grammatical and logical sense.

e.g. no. 3: He ate the cake and I have. Not even small native-speaking children would make this kind of error. (They might make others, but not these ones.)

Quote
It happens that most of this kind of research into this specific type of phenomenon has been done largely in English, partly an accident of history that this type of work was developed in anglo-American environments, and partly because English provides these convenient and telling subject-auxiliary contractions ("He's", "I've", ...). But to get a reliable picture of what's going on, it's good to have support from other languages.



I don't think you can get specific support from a different language for a particular argument in one language. IANALinguist (but my best friend nearly has her PhD and can beautifully refute Chomsky's theory of universal grammar as well as the basis on which his research was carried out) but from what I can tell, linguistic science has difficulty in making understandable and relevant arguments about languages other than the one being studied.

I guess what I am saying is that we are very far from "universal grammar" in human language, and I get edgy when it's approached from a soi-disant logical and scientific perspective.

Mandos

  • Senior Member
  • *****
  • Posts: 5759
    • View Profile
    • http://politblogo.typepad.com/
Expérience en grammaticalité
« Reply #6 on: February 08, 2007, 10:01:21 AM »
So would you also reject:

13. Il a mangé le gâteau, mais je ne l'ai pas.
14. John ate the fish, but I have not.

14 from a structural standpoint, not from a logical one.  ie, it's OK that it is ambiguous.

Quote
I don't think you can get specific support from a different language for a particular argument in one language. IANALinguist (but my best friend nearly has her PhD and can beautifully refute Chomsky's theory of universal grammar as well as the basis on which his research was carried out) but from what I can tell, linguistic science has difficulty in making understandable and relevant arguments about languages other than the one being studied.


Solike, these refutations are a dime a dozen and have been since the beginning.  And usually they're worth as much.  In my experience, the argument usually leaves nothing in its wake---no alternative proposals---except maybe a universal grammar-shaped hole, papered over by assumptions/commitments that are far more ambitious than what was being claimed before.  

If there is no universal grammar, there is no basis for acquisition.  Chomsky himself has refuted some of his original arguments, which is normal in science.  But to refute the entire approach to the problem, one has to challenge the Poverty of the Stimulus argument.  People claim to do this by various means.  For instance, I read a paper that claimed that the Poverty of the Stimulus argument was weak because he ran a simulation that learned some grammatical distinctions with a corpus of a couple of million word rather than hundreds of millions or billions...

brebis noire

  • Senior Member
  • *****
  • Posts: 4707
    • View Profile
Expérience en grammaticalité
« Reply #7 on: February 08, 2007, 10:16:34 AM »
Quote from: Mandos
So would you also reject:

13. Il a mangé le gâteau, mais je ne l'ai pas.
14. John ate the fish, but I have not.

I would reject it as ungrammatical and illogical in French, because the translation of 13. would be:
John ate the fish but I do not have it.


Quote
If there is no universal grammar, there is no basis for acquisition.  Chomsky himself has refuted some of his original arguments, which is normal in science.  But to refute the entire approach to the problem, one has to challenge the Poverty of the Stimulus argument.  People claim to do this by various means.  For instance, I read a paper that claimed that the Poverty of the Stimulus argument was weak because he ran a simulation that learned some grammatical distinctions with a corpus of a couple of million word rather than hundreds of millions or billions...


I wish I could reproduce my friend's very interesting refutation of the universal grammar theory, but I lack the detailed understanding of linguistics to do so. I do not think these refutations are a dime a dozen, and many of them have not even been adequately translated into English. The field of linguistics, especially non-English linguistics (and non-Bible-translating linguistics) is starved for funds.

One thing that particularly bothered me about Chomsky's attempt at universal grammar was that the funding and strategic interest behind it (at MIT) was a military context - i.e. If we can determine a universal grammar, we can use it to decode other languages and it therefore becomes a useful tool for spying an intelligence-gathering. This was, I guess, a naive belief from the 1950s postwar period that pretty well went nowhere.

shaolin

  • Full Member
  • ****
  • Posts: 495
    • View Profile
Expérience en grammaticalité
« Reply #8 on: February 08, 2007, 10:27:38 AM »
Is all of Chomsky's linguistics work based on the english language?  His father was a Hebrew scholar, and passed a lot onto his son.  Did none of this end up in his linguistics?  I admit to trying to muddle through the linguistics on a couple of occasions, only to get pulled back into his more interesting (to me) politics.

Mandos

  • Senior Member
  • *****
  • Posts: 5759
    • View Profile
    • http://politblogo.typepad.com/
Expérience en grammaticalité
« Reply #9 on: February 08, 2007, 10:39:26 AM »
Alright, so what I am trying to do is produce sentences with ellipsis deleting a past participle. This is a case of what is called "VP-ellipsis".  From some simultaneous seaches of the literature, I now realize that Romance languages don't use VP-ellipsis.  

Quote
I wish I could reproduce my friend's very interesting refutation of the universal grammar theory, but I lack the detailed understanding of linguistics to do so. I do not think these refutations are a dime a dozen, and many of them have not even been adequately translated into English. The field of linguistics, especially non-English linguistics (and non-Bible-translating linguistics) is starved for funds.

Yes, it's too bad you can present the arguments here. :)  Does your friend have publications whose references you can PM me?  At the moment, attempting to refute Chomsky is a popular endeavour.  His approach reached its "high-water mark" in the early 90s, I think, and while it's still the dominant paradigm in linguistics, lots of people attempt to make a name for themselves by refuting generative grammar, or at least his take on it, especially because...

Quote
One thing that particularly bothered me about Chomsky's attempt at universal grammar was that the funding and strategic interest behind it (at MIT) was a military context - i.e. If we can determine a universal grammar, we can use it to decode other languages and it therefore becomes a useful tool for spying an intelligence-gathering. This was, I guess, a naive belief from the 1950s postwar period that pretty well went nowhere.


...the US government currently spends a lot of money on exactly the reverse proposition, that we just find language-specific patterns based on a very simple "Markovian" model. (ie, what word is likely to follow the previous N words given what we've already seen...).  The logical problem this presents for natural acquisition are shunted aside as an expensive diversion, when it comes to funding.  Where do you think Google comes from?

Lastly, yes, there isn't a lot of money for linguistics, especially---and oddly enough---in Canada.  Though the University of Ottawa, McGill, UToronto, UQAM and a handful of others have relatively good departments.  The University of Ottawa does a lot of research in a universal grammar approach using a Romance language context.  And lots of work is done with Japanese and German and other languages.  But I actually don't believe that research into other languages can "refute" universal grammar as a general approach (as opposed to specific current proposals).  To do so requires a commitment to a different and to me unsupportable theory of acquisition.  (ie, that the child has access to at least a couple of million sentences before acquiring the grammar.)

Mandos

  • Senior Member
  • *****
  • Posts: 5759
    • View Profile
    • http://politblogo.typepad.com/
Expérience en grammaticalité
« Reply #10 on: February 08, 2007, 10:40:58 AM »
shaolin: So IIRC Chomsky did also work with Hebrew on some things.  But Cambridge, MA tended to have a lot of anglophones at the beginning, so a lot of early work was done using English as the example language.

skdadl

  • Global Moderator
  • Senior Member
  • *****
  • Posts: 32874
    • View Profile
    • http://www.pogge.ca
Expérience en grammaticalité
« Reply #11 on: February 08, 2007, 10:45:33 AM »
I see the problem with #13.

In English, we use two different verb tenses, the simple past and the perfect, to express that thought -- ate and have [eaten].

You're trying to translate that directly into French, and you can't because forms of the verb "avoir" play two different roles in that structure.

That is actually a well-known fact, I think. I can't imagine that figuring out a genuine universal grammar could ride on such banalities. As I recall, it is the way that English plays with combinations/successions of verb forms that is the peculiarity; the (correct) French combination would be more common.

Mandos

  • Senior Member
  • *****
  • Posts: 5759
    • View Profile
    • http://politblogo.typepad.com/
Expérience en grammaticalité
« Reply #12 on: February 08, 2007, 10:45:48 AM »
(triple post!)

So people can rest easy because I should have checked whether you can have this kind of ellipsis phenomenon in French at all, and apparent you can't. I know of non-English languages where you can, such as Japanese.  But I don't know any languages where which have that phenomenon and still use subject-auxiliary  contractions like English does.

Still, it was an interesting thought.  Back to the drawing board.

skdadl

  • Global Moderator
  • Senior Member
  • *****
  • Posts: 32874
    • View Profile
    • http://www.pogge.ca
Expérience en grammaticalité
« Reply #13 on: February 08, 2007, 10:47:58 AM »
Well, since you have to keep the tenses in French consistent, why not figure out a way to write

He [did something], and I not.

???

Wouldn't that be the way the French would have to do it?

Mandos

  • Senior Member
  • *****
  • Posts: 5759
    • View Profile
    • http://politblogo.typepad.com/
Expérience en grammaticalité
« Reply #14 on: February 08, 2007, 10:53:30 AM »
Quote
You're trying to translate that directly into French, and you can't because forms of the verb "avoir" play two different roles in that structure.

Well, "avoir" is used very similarly to "have" in English.  However, French has another interfering factor in that this can't work with any auxiliary, not just "avoir".  

Quote
That is actually a well-known fact, I think. I can't imagine that figuring out a genuine universal grammar could ride on such banalities. As I recall, it is the way that English plays with combinations/successions of verb forms that is the peculiarity; the French combination would be more common.


There are languages that behave like English, by the way, but they don't tend to be European languages, it seems.  

Well, nothing *hinges* on this kind of thing.  What a universal grammar *does* hinge on, I think, is an explanation of ellipsis phenomena, of which there are a number of different kinds, but most of them share certain "spooky" characteristics.  The fact that English *prevents* contractions in ellipsis phenomena for the most part is a very interesting fact, because it suggests that there is something *there* blocking the contraction but unpronounced.  (aka, a "trace".)  That it can't be found in French doesn't refute the theory---because it turns out that languages like French are preventing that entire general class of ellipsis phenomena.  And that languages that do have that phenomenon that I am *familiar* with don't use those contractions.

 

Return To TAT