Language Acquisition

Friday, April 25, 2003 1:03 pm

Here is more extensive description of the project. I am sending this now rather than talking about it in class because you should try to work things out this weekend, if you can at all.


The general question is: discuss the use of Root Infinitives (RI) and Null Subjects (NS) for each relevant verb type (clause type). Where possible, document the rate of null subjects, the interaction with wh-phrases and the co-evolution (or not) of RI and NS. Use this to address the following more precise questions:

1. being careful to only look at cases where the type of verb is clear (see below for which are clear), does a clear patterns emerge for the behaviour of RI and NS?

2. use your results in (1) above to make hypotheses about the ambiguous cases, if possible. Do they behave more like RS or NS (or mixed)? Also look at the aux/imperatives, how do they behave?

3. For those of you who have SLI kids: use both results in (1-2) above to determine whether the SLI kids have a profile comparable to the non-SLI kids, or whether they diverge in the nature of their development.


Here are the verb types for English:

verbal
..tensed (=finite)
....sForm
....unambiguousPast
..untensed
....bareStem (with third person subjects)
....    infinitive
....    participle
....    ing
..  murky
....    bareStem (with 1/2 or null subjects)
....    ambiguousPast
..  imperative
..  aux

All verb forms under 'tensed' and 'untensed' are clear and should be used for question 1. Those under murky, usually the most frequent ones, are ambiguous and should be used for question 2.

Each entry in the above tree is a Dochoa keyword. For third versus first/second person, use the keywords 'thirdPersonSubject' and 'firstSecondPersonSubject'. (Remember that the results fo these will probably be pretty approximate. If the numbers are not too big, you might want to check them by opening an editor on the results). A list of Dochoa keywords will be available shortly on the Dochoa website. I will send a separate message about that.

For French, the tree of possible verb types is simpler (compensating for the fact that the tagging was harder!):

verbal
..   tensed
..   untensed
....     infinitive
....     participle
....     untensedAmbiguous
..   imperative
..   aux

The tensed (=finite), untensed and participle cases are the clear cases. The untensedAmbiguous is the ambiguous case.


The project will be graded as follows:

1. Accuracy of tagging (20%)
2. Report on the difficulties while tagging and suggestions to improve Dochoa (10%)
3. Statistics and discussion for the questions above applied to your sessions only (20%)
4. Statistics and discussion for the questions above for all sessions of your group, compared with the profiles of the other kids (20%)
5. Personal project, see below (10%)

The project homeworks are worth 20% (for a total of 100% ;).

As you can see, 50% is tagging and homework, and 50% attributed to the final analysis. Within the latter, 40% goes to the RI/NS statistics and 10% to your project. If you found a personal project of particular interest to you, you however have the option to change the proportions to 30% going to RI/NS and 20% to your project. If you choose that option the figures above would be modified like this:

3. ... (15%)
4. ... (15%)
5. ... (20%)


The personal project: here you can choose any topic that caught your attention and count it, giving appropriate statistics and discussion. For this sub-part of the project, you are encouraged to do it in groups. If you do it in groups, you can either have a common writeup (graded with higher standards), or a submit each an individual writeup.

The topic is open to you, but I would like to hear about it before you spend too much energy on it. For those of you who have no idea for a topic, I will provide suggestions. Here are a couple, that I can explain in more detail in class or after class:

- look at absent preposition (eg. "I spoke him"). This has not been looked at, I don't know whether it occurs in any relevant quantity, but I suspect that it might be a relevant phenomenon.

- similarly, look at the rate of use of complementisers ("he said he came" versus "he said that he came"). It is sometimes said that initially they don't use any, even in languages that require them, such as French or in English context that require them ("he asserted he ate this", "the fact he ate this").

- try to code the null subjects for whether they are likely first person, second person or third person, and see if the use of NS/RI is different in person types. (That count was done on another French corpus, yielding no very clear results, but the situation might be different with English bare stems).

- in French, measure how many question words are moved ("ou il est?") or not ("il est ou?"). If you see unmoved wh-words in English, you can also do this with English kids.

These are "classical" suggestions. You are encouraged to come up with your own, which will undoubtedly be more imaginative :)


Timing and methodology: As I mentioned in class, you can easily drown in numbers, once you start counting various statistics. Don't :) I highly recommend that you schedule yourself some quality time just sitting down and thinking about what it is that you are trying to count and what the preliminary numbers are telling you.

Given the timing, it would be optimal if you could sit down with this during the weekend, extract the counts from Dochoa, and then have a couple of quiet thinking sessions about how to interpret the numbers. After that, give it some digestion time at the beginning of next week, and then turn to writing it up.

Also, I will enquire with the powers that be, to see if it is technically feasible for me to give you more time to do the projects (if you want that, of course). I fear however that the grades need to be turned in relatively soon.


That's it for now. Try to have fun unravelling fine-grained aspects of the speech of your kid!