1 CAS LX 500 Topics in Linguistics: Questions April 28, 2009 Spring 2009 14a: Processing questions How are questions processed? We’ve looked at the grammatical issues in questions, how is that grammatical knowledge put to use? The problem of parsing Most of what we’ve considered with respect to the grammar of questions has been more closely related to production of questions than with perception of questions. Given what you want to ask, what would it sound like? What’s involved in creating a structural analysis of what we’re hearing? I… IP 3 DP I! I 3 (I) (VP) I will… IP 3 DP I! I 3 I (VP) will I will meet… IP 3 DP 1 I! I 3 I VP will 3 V (DP) meet It seems reasonably likely that something like this is how we proceed. Take what we hear, as we hear it, construct the structure to the extent that we can predict it. (1) The editors said five words were misspelled. (2) The committee considered me to be qualified.
15
Embed
CAS LX 500 Topics in Linguistics: Questions April 28, 2009 ...ling-blogs.bu.edu/lx500b1s09/files/2009/04/lx500s09q-14a-proc.pdf5 (12) Implausible infinitival The school superintendent
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
CAS LX 500 Topics in Linguistics: Questions April 28, 2009 Spring 2009 14a: Processing questions
How are questions processed?
We’ve looked at the grammatical issues in questions, how is that grammatical knowledge
put to use?
The problem of parsing
Most of what we’ve considered with respect to the grammar of questions has been more
closely related to production of questions than with perception of questions. Given what
you want to ask, what would it sound like? What’s involved in creating a structural
analysis of what we’re hearing?
I… IP 3 DP I! I 3 (I) (VP) I will… IP 3 DP I! I 3 I (VP) will I will meet… IP 3 DP1 I! I 3 I VP will 3 V (DP) meet
It seems reasonably likely that something like this is how we proceed. Take what we hear,
as we hear it, construct the structure to the extent that we can predict it.
(1) The editors said five words were misspelled.
(2) The committee considered me to be qualified.
2
Given the fact that we are getting words one-by-one, the predictions can sometimes be
wrong. When the prediction is wrong, we need to reanalyze.
Some reanalysis is easy. Some reanalysis is hard.
(3) The mobster kicked in the side fell in pain.
We can get insight into the parsing process by looking at where people slow down when
reading. The Filled Gap Effect.
(4) Who did the children force us to sing songs for yesterday?
One big question: Much of our evidence about the grammar comes from consideration of
sentences that “sound good” and “sound bad.” But, to what extent can “*” be attributed to
parsing and to what extent can it be attributed to the grammar?
(5) a. The mouse stole the cheese.
b. The mouse the cat chased stole the cheese.
c. ?? The mouse the cat the child feeds chased stole the cheese.
d. * The mouse the cat the child the teacher scolded feeds chased stole the cheese.
Another way to think about this: What is the parser doing? What relationship is there
between the parser and the grammar?
For example—islands. Are they about grammar (as we’ve been assuming), or are they
about the parser? Are the islands native speakers linguists consult find problematic due to
parsing?
(6) Who did you hope that the candidate said that he admired __?
(7) a. * Who did the candidate read [a book that praised __ ]?
b. * Who did the candidate wonder [whether the press would denounce __ ]?
c. * Who did [the fact that the candidate supported __] upset voters?
d. * Who did the candidate raise two million dollars [by talking to __]?
An interesting construction in this context is the parasitic gap construction:
(8) a. * What did [the attempt to repair __ ] ultimately damage the car?
b. What did [the attempt to repair the car] ultimately damage __ ?
c. What did [the attempt to repair _pg_ ] ultimately damage __ ?
3
Look: Some gaps are possible inside islands. And if you’re hearing this word-by-word,
and evaluating it incrementally, (8a) is only really ungrammatical at the very end.
(9) What did the attempt to repair ultimately damage (*the car)?
So, certainly it can’t be the case that the parser just throws up its hands and storms away
from sentences with gaps in islands. But how does the gap in the island get there? Is it
posited immediately? If so, the parser has no problems putting gaps in islands, even if it
will turn out later to be ungrammatical. Does the parser wait until the good gap is
constructed, and then go back and put the pg in? But it’s clear that something is missing in
the subject as soon as ultimately is reached.
How quickly can the grammar be used? Does the parser create a rough structure that the
grammar later evaluates, or is the grammar actively involved in the parsing process? (And
if there is a “rough parse” what syntax is it built using? Why do we need two grammars?)
One way to look at this question is to see if grammatical violations are detected
immediately. Is the parser sensitive to islands? Binding theory? Etc.
An early oft-cited study (Stowe 1986) that indicates that the parser is sensitive to islands
found no slowdown in The teacher asked what the silly story about Greg’s older brother
was supposed to mean.
ERP study: McKinnon & Osterhout (1996): While a wh-dependency is pending, the
beginning of an island triggers P600 (syntactic anomaly?) and LAN. So, the parser notices