Sentence-picture matching Comprehension questions Repetition of

The comprehension, repetition, and production
of Wh movement in Syntactic SLI
Rama Novogrodsky and Naama Friedmann
The poster can be downloaded at
http://www.tau.ac.il/~naamafr
Tel Aviv University, Israel
Abstract
Results:
The comprehension tasks indicated that reversible non-canonical structures that are
derived by Wh movement, object relatives and object questions are a source of
considerable difficulty in Syntactic-SLI. The results in the repetition task were similar:
subject relatives and non-reversible object relatives were repeated more accurately than
reversible object relatives. The elicitation tasks showed that children with S-SLI can
produce relative clauses, but prefer to produce subject relatives instead of object
relatives and tend to omit one of the arguments to reduce the number of -roles.
The comprehension, repetition and production of sentences derived by Wh
movement were tested in 20 School-age (4th-8th grades) Hebrew-speaking children
with Syntactic SLI. The comprehension of relative clauses and referential Wh
questions was tested using a sentence-picture-matching task. Relative clauses were
also tested using a comprehension question and written relatives paraphrase tasks.
The repetition task included reversible and non reversible relative clause sentences.
The production of relative clauses included two elicitation tasks.
Participants:
The Syntactic SLI participants met all the exclusionary criteria for SLI (Leonard, 1998).
Their nonverbal intellectual functioning was within the age-appropriate level.
The participants in the control group were typically developing 4th graders.
Comprehension questions
Conclusion:
The deficit in Wh movement structures in school-age Syntactic-SLI relates to an inability
to transfer thematic roles to moved constituents in reversible non-canonical sentences.
Sentence-picture matching
Relative clauses: 60 sentences:
30 sentences: subject- or object- center embedded
relatives, and sentential complements. In center
embedding both subject and object relatives represent
non-canonical Wh movement structure. After each
sentence a question was asked (Who fell asleep?)
Subject relative:
The-giraffe that-licked ACC the elephant fell-asleep.
Object relative:
The-giraffe that-the-elephant licked fell-asleep.
Sentential complement:
The-giraffe saw that-the-elephant fell-asleep.
20 simple SVO, 20 subject relatives, 20 object relatives.
The participants were asked to point to the matching picture.
SVO:
The dog licks ACC the cat.
Subject relative:
Show me ACC the cat that licks ACC the dog.
Object relative:
Show me ACC the dog that the cat licks
80%
80%
*
70%
Control
S-SLI
Control
*
70%
60%
50%
Subject Relative
Object Relative
Sentential Complement
Comprehension of relative clauses was impaired,
Sentential complements of verbs were unimpaired.
*
70%
Control
60%
50%
*
S-SLI
S-SLI
60%
SVO
80%
Object relative:
The guy that the boy loved cut old newspapers.
Control sentence:
The student from fourth grade cut old newspapers.
90%
90%
90%
24 written sentences, 12 object relatives and 12 length-matched
control sentences. The participants were asked to read the
sentences aloud and then paraphrase them.
100%
100%
100%
Paraphrasing written relatives
Subject Relative
50%
Object Relative
Control Sentences
Object relatives
Wh questions: 80 questions: 20 who subject, 20 which
subject, 20 who object and 20 which object
Non-referential subject question:
Who is-licking ACC the-dog?
Referential subject question:
Which cat is-licking ACC the-dog?
Non-referential object question:
ACC who the-dog is-licking?
Referential object question:
ACC which cat the-dog is-licking?
The paraphrasing task also showed the same impairment in
object relatives. Errors were mainly thematic role reversals.
For example:
“The boy cut old newspapers, that’s why the guy liked him”
100%
Syntactic SLI and Agrammatism
90%
*
80%
Elicitation of relative clauses
32 subject and target subject and object relative clauses:
20 in a picture elicitation task and 12 in a preference task.
Picture description task
Experimenter: In this picture the dog licks the cat. In that picture
the cat licks the dog. Which dog is this? (Pointing to the dog in
the bottom picture for a subject relative).
*
70%
S-SLI
Control
50%
Who Subject
Which Subject
Who Object
Which Object
Object relatives and referential object questions were
significantly harder for the children with syntactic SLI than
for typically developing children.
Target object relative: This is the girl that the grandma kisses.
100%
90%
80%
70%
60%
*
*
S-SLI
50%
Repetition of reversible and
non-reversible object relatives
Reversible and non reversible object relative sentences were
presented aurally. The participants were asked to count out loud to
10 and then repeat the sentence. Thematic role errors in repetition
were compared to other repetition errors. For example: switching
two NPs, reducing one of the arguments, or replacing it with a
resumptive pronoun.
Reversible object relative:
I saw the horse that-the-donkey pushed.
Non-reversible object relative:
The teacher read the story that-Uri wrote.
40%
35%
30%
20%
25%
10%
20%
0%
1
3
4
Subject Relative
Object 2Relative Subject
Relative Object
Relative
Picture Description
Preference Task
S-SLI
Control
15%
10%
5%
0%
In production the participants with Syntactic SLI tried to
avoid sentences that contain two thematic constituents
in a Wh movement operation. They either produced a
subject relative for an object relative, produced a simple
sentence, or omitted one of the arguments.
Reversible OR
Comprehension
of reversible
non-canonical
sentences
Production
Deficit in object relatives.
Deficit in both subject and
object relatives.
Complementizer omissions.
Syntactic
structure and
trace
Possible locus of
the deficit
No complementizer
omissions.
Grammatical sentences
Ungrammatical productions.
intact
impaired
Syntactic chain
Syntactic structure
(CP), trace deletion
Summary
The results of this set of studies suggest
that school-age children with syntactic SLI
have an impairment in the comprehension
and production of reversible non-canonical
sentences that are derived by Whmovement.
Control
30%
Agrammatism
60%
Target subject relative: This is the dog that the cat is licking.
Preference task:
Experimenter: There are two boys. One boy gives a present,
and one boy receives a present. Which boy would you rather
be? (start with “the boy…)
Target subject relative: The boy that gives a present.
Experimenter: There are two children. Daddy wakes up one
child and the clock wakes up one child. Which child would you
rather be?
Target object relative: The child that daddy wakes up.
Syntactic SLI
Non-reversible OR
There were more thematic errors than other errors in both
groups in reversible compared to non-reversible object
relatives. Importantly, the Syntactic SLI group produced
significantly more thematic errors than the control group.
The deficit in Wh movement relates to an
inability to transfer thematic roles to moved
constituents,
possibly
over
another
argument of the same type.
This is the zebra that the horse pushed t