24 ĐỀ THAM KHẢO KÌ THI TUYỂN SINH VÀO LỚP 10 MÔN TIẾNG ANH SỞ GIÁO DỤC HẢI DƯ...
THE EFFECT OF SYNTACTIC COMPLEXITY ON SENTENCE PROCESSING BY CHILDREN WITH READING DIFFICULTIES AND BEGINNING READERS
1. The effect of syntactic complexity
on sentence processing by children
with reading difficulties and
beginning readers
Maria Mastropavlou
Michaela Nerantzini
Theophano Christou
Marika Lekakou
University of Ioannina
This project has received funding from the European Union’s Horizon 2020 research and innovation programme
under grant agreement No 731724.
2. General aim of this study
Explore aspects of syntactic
complexity.
Investigate how they affect reading.
LDG7, Athens, June 2018 2
3. Introduction
Processes of reading
Processing and recognizing graphemes
Recognizing and accessing words in the
lexicon
Processing sentences
Integrating contextual information
LDG7, Athens, June 2018 3
4. Introduction
Development of reading
Two accounts on the difficulty of learning
to read:
The Structural Deficit Hypothesis:
The syntactic knowledge required for successful
reading is not yet available in beginning
readers.
The Processing Deficit Hypothesis:
The syntactic knowledge is there but cognitive
resources (i.e. working memory) are spent on
newly-acquired skills: decoding and processing
printed words.
LDG7, Athens, June 2018 4
5. Introduction
Syntactic complexity in sentence
processing:
Word order: subject-object asymmetries
(Gibson, 1998, Fiebach et al., 2002, Rosler et al.,
1998)
Number of verb arguments (Thompson
2003)
Embedding (Bock, 1982, Ferreira, 1991)
Locality – distance (Gibson, 1998, Zurif,
Swinney, Prather, Solomon, & Bushell, 1993)
LDG7, Athens, June 2018 5
6. Introduction
Syntactic comprehension in reading
difficulties
Poor readers are often found to have lower
sentence comprehension skills compared to
peers.
These difficulties have been attributed to
phonological processing limitations (i.e.
performance) rather than lacking knowledge
(i.e. competence) (Craint & Shankweilert 1986,
Leikin & Bouskila 2004, Brown 2006).
Children with dyslexia often present slow
acquisition of syntax during preschool years
but reach peers’ abilities after age 5
(Scarborough 1991).
LDG7, Athens, June 2018 6
7. The present study
Aims to explore how syntactic complexity
factors affect sentence comprehension
during typical and atypical reading
development.
Part of a broader aim:
Construction of iRead, an integrated system
for the enhancement of reading skills in
children with reading difficulties, beginning
readers and young EFL students.
LDG7, Athens, June 2018 7
8. The present study
Specific aim:
Explore the difficulty level of syntactic
features:
Passive morphology (passive, reflexive,
anticausative, reciprocal and deponent verbs)
Embedding (complement, adverbial and relative
clauses)
Discourse anaphors (definite and indefinite NPs,
proper names, personal and null pronouns).
LDG7, Athens, June 2018 8
9. Methodology: Participants
Three groups of children:
30 students with reading difficulties (20M,
10F), aged 11-12 years (mean=11,4 ,
SD=0,803012),
28 control students without reading
difficulties (19M, 9F), aged 11-12 years
(mean=11,56786, SD=0,466709),
28 beginning readers without reading
difficulties (11M, 17F), aged 7-9
(mean=7,982759, SD=1,004301).
LDG7, Athens, June 2018 9
10. Methodology: Materials
Non-verbal Tasks:
1. Digit span: Repeat a sequence of numbers
(2,3,4,5,6,7,8,9)
2. Memory span: Repeat a sequence of words
(2,3,4,5,6,7,8,9)
3. Sentences repetition (Length and
complexity of the presented sentences have
been manipulated)
4. non-word repetition: Increasing number of
syllables (2–3–4–5 syllable non-words)
LDG7, Athens, June 2018 10
11. Methodology: Materials
Experimental materials for self-paced
reading task
500 sentences including the following
structures:
Embedding (180 sentences)
Passive constructions (230 sentences)
Discourse anaphors (90 sentences)
LDG7, Athens, June 2018 11
12. Methodology: Materials
Item check: two sentence plausibility
tasks.
Plausibility task 1:
500 sentences given to 50 native speakers aged
18-24 to rate on a 1-5 Likert scale 35
sentences were changed (1SD below mean).
Plausibility task 2:
35 sentences were given to 25 native speakers
8 sentences were changed (1SD below
mean).
LDG7, Athens, June 2018 12
13. Methodology: Materials
Experimental conditions: Embedding
LDG7, Athens, June 2018 13
ADVERBIAL CLAUSES (TEMPORAL): 40 SENTENCES
ο Mathitis / pire / to vravio /
the student received the prize,
afu/otan / agaljase / ton diefthindi / sti jorti
when/after hug-3sg.past.perf. the principal at the ceremony
=when/after he hugged the principal at the ceremony
eno / agaljaze / ton diefthindi / sti jorti
while hug-3sg.past.imperf. the principal at the ceremony
=while hugging the principal at the ceremony
prin / agaljasi / ton diefthindi / sti jorti
before hug-3sg.nonpast.perf. the principal at the ceremony
=beforehugging the principal at the ceremony
14. Methodology: Materials
Experimental conditions: Embedding
LDG7, Athens, June 2018 14
COMPLEMENT CLAUSES: 60 SENTENCES
O maθitis /
the student
pistepse / oti / i kopela / htipise / ton papu / hθes to vradi
believed that the girl hit grandpa yesterday evening
rotise / an / i kopela / htipise / ton papu / hθes to vradi
asked If the girl hit grandpa yesterday evening
foviθike / pu / i kopela / htipise / ton papu / hθes to vradi
was afraid that the girl hit grandpa yesterday evening
15. Methodology: Materials
Experimental conditions: Embedding
LDG7, Athens, June 2018 15
RELATIVE CLAUSES: 80 SENTENCES
Right-branching: Subject (a), Object (b) with pu
(a) i kopela / ide / ton andra / pu / lerose / o maθitis / me boja.
the girl saw the man that stained-3sing.past the student-NOM with paint
=the girl saw the man that the student stained with paint
(b) i kopela / ide / ton andra / pu / lerose / ton maθiti / me boja.
the girl saw the man that stained-3sing.past the student-ACC
with paint
=the girl saw the man that stained the student with paint
Center-embedded: Subject (a), Object (b) with pu
(a) i kopela / pu / ide / o andras / lerose / ton maθiti / me boja.
the girl that saw-3sing.past the man-NOM stained the student with paint
=the girl that the man saw stained the student with paint
(b) i kopela / pu / ide / ton andra / lerose / ton maθiti / me boja.
the girl that saw-3sing.past the man-ACC stained the student with paint
=the girl that saw the man stained the student with paint
16. Methodology: Materials
Experimental conditions: Discourse
anaphors
LDG7, Athens, June 2018 16
SUBJECT POSITION (60 sentences):
I Anna ke o zoγrafos piγan stin kaferetia epidi /
=Anna and the painter-Masc. went to the café because…
(Personal pronoun) afti / ksexase / to portofoli / sto trapezi
she forgot the wallet on the table
(Definite NP) o zoγrafos/ ksexase / to portofoli / sto trapezi
the painter-Masc forgot the wallet on the table
(Indefinite NP) mia fili / ksexase / to portofoli / sto trapezi
a friend forgot the wallet on the table
(Proper Name) i Anna / ksexase / to portofoli / sto trapezi
Annaforgot the wallet on the table
17. Methodology: Materials
Experimental conditions: Passive
morphology
LDG7, Athens, June 2018 17
Passive constructions (60 sentences)
O andras / me ta mavra jalja / sinelifθi / apo tin astinomia / to vradi.
the man with the black glasses was arrested-pass. by the police at night
Reflexive constructions (30 sentences)
O γabros / me to aγonistiko amaksi / ksiristike / apo monos tu / to apojevma.
the groom with the race car shaved-pass. by himself in the afternoon
Reciprocal constructions (30 sentences)
Ta pedja / me tus kalus vaθmus / agaljastikan / metaksi tus / to proi.
the children with the good grades hugged-pass. each other in the morning
18. Methodology: Materials
Experimental conditions: Passive
morphology
LDG7, Athens, June 2018 18
Deponent verbs (20 sentences)
I kopela / me tin kordela / eroteftike / to aγori / ta Xristujena.
the girl with the ribbon loved-pass. the boy at Christmas
Unaccusative constructions (anticausative verbs) (40 sentences)
I tenda / me ta endona xromata / skistike / apo ton aera / to fθinoporo.
the tent with the bright colours was torn-pass. by the wind in autumn
19. Methodology: Experimental procedure
Self-paced reading task
Experiment broken into 4 separate lists using
Latin Square Design (20mins each),
8 lists for beginning readers (15mins each).
Reaction Times (RTs) were recorded at the
critical(s) segment(s).
LDG7, Athens, June 2018 19
24. 0
200
400
600
800
1000
1200
1400
1600
PROPER NAME DEF NP INDEF NP PRONOUN
Segment 2
Control
s
Beg.Re
aders
R.Diffic
ulties
Results: Discourse Anaphors
LDG7, Athens, June 2018 24
0
5000
10000
15000
20000
25000
PROPER
NAME
DEF NP INDEF NP PRONOUN
Discourse Anaphors (Total RTs)
Controls
Beg.Read
ers
R.Difficulti
es
seg1 seg2 seg3 seg4 seg5
Pr.Name I Anna ke o zoγrafos piγan stin kaferetia epidi i Anna ksexase to portofoli sto trapezi
Def NP I Anna ke o zoγrafos piγan stin kaferetia epidi o zoγrafos ksexase to portofoli sto trapezi
Indef NP I Anna ke o zoγrafos piγan stin kaferetia epidi mia fili ksexase to portofoli sto trapezi
Pronoun I Anna ke o zoγrafos piγan stin kaferetia epidi afti ksexase to portofoli sto trapezi
26. 0
500
1000
1500
2000
seg3 seg4 seg5
Reflexives (Beg.Readers)
REFL (by-phrase)
REFL (by himself)
AMBIG (adverb)
PASS_by-phrase
0
500
1000
1500
2000
seg3 seg4 seg5
Reflexives (R.Difficulties)
REFL (by-phrase)
REFL (by himself)
AMBIG (adverb)
PASS_by-phrase
Results: Passive Morphology
LDG7, Athens, June 2018 26
seg1 seg2 seg3 seg4 seg5
REFL
(by-
phrase)
O pappous
me ti mavri
mplouza
zigistike apo ti nosokoma to proi
REFL (by
himself)
O pappous
me ti mavri
mplouza
zigistike apo monos tou to proi
AMGIG
(adverb)
O pappous
me ti mavri
mplouza
zigistike pali me dyskolia to proi
PASS
(by-
phrase)
O andras
me ta mavra
jalja
sinelifθi apo tin astinomia to vradi
0
500
1000
1500
2000
seg3 seg4 seg5
Reflexives (Controls)
REFL (by-phrase)
REFL (by himself)
AMBIG (adverb)
PASS_by-phrase
27. Results: Passive Morphology
LDG7, Athens, June 2018 27
0
500
1000
1500
2000
seg3 seg4 seg5
Reciprocals (Beg.Readers)
RECIPR
(eachother)
RECIPR (by-
phrase)
0
500
1000
1500
2000
seg3 seg4 seg5
Reciprocals (Controls)
RECIPR
(eachother)
RECIPR (by-
phrase)
0
500
1000
1500
2000
seg3 seg4 seg5
Reciprocals (R.Difficulties)
RECIPR
(eachother)
RECIPR (by-
phrase)
seg1 seg2 seg3 seg4 seg5
RECIPR
(eachother) Ta paidia
me tous kalous
vathmous agkaliastikan metaksi tous to proi
RECIPR
(by-
phrase) Ta paidia
me tous kalous
vathmous agkaliastikan apo ti mitera to proi
28. Results: Passive Morphology
LDG7, Athens, June 2018 28
0
500
1000
1500
2000
seg3 seg4 seg5
Deponents (R.Difficulties)
DEP_trans
DEP_intra
ns
ACT_trans
PASS_by-
phrase
PASS_oth
erPP 0
500
1000
1500
2000
seg3 seg4 seg5
Deponents (Beg.Readers)
DEP_trans
DEP_intrans
ACT_trans
PASS_by-
phrase
PASS_other
PP
0
500
1000
1500
2000
seg3 seg4 seg5
Deponents (Controls)
DEP_trans
DEP_intran
s
ACT_trans
PASS_by-
phrase
PASS_othe
rPP
seg1 seg2 seg3 seg4 seg5
DEP trans
o andras me ta gkriza malia oneireftike ti mitera tou to vradi.
DEP intrans o andras me ta gkriza malia prosefxithike sto theo to vradi.
ACT trans
o andras me ta gkriza malia entopise ti mitera tou to vradi.
PASS by-
phrase o andras me ta mavra gialia sinelifthi
apo tin
astinomia to vradi
PASS other
PP o andras me ta mavra gialia sinelifthi se liga lepta to vradi
29. Results: Passive Morphology
LDG7, Athens, June 2018 29
0
500
1000
1500
2000
seg3 seg4 seg5
Anticausatives (Controls)
ANTIC_age
nt by-phr
ANTIC_non-
agent by-
phr
PASS_agent
by-phr
0
200
400
600
800
1000
1200
1400
1600
1800
2000
seg3 seg4 seg5
Anticausatives (R.Difficulties)
ANTIC_ag
ent by-
phr
ANTIC_no
n-agent
by-phr
PASS_ag
ent by-
phr
0
200
400
600
800
1000
1200
1400
1600
1800
2000
seg3 seg4 seg5
Anticausatives (Beg.Readers)
ANTIC_agen
t by-phr
ANTIC_non-
agent by-
phr
PASS_agent
by-phr
seg1 seg2 seg3 seg4 seg5
ANTIC
agent by-
phr To spiti me tin iperoxi limni erimothike
apo tous
katoikous
to
kalokairi
.
ANTIC non-
agent by-
phr To spiti me tin iperoxi limni erimothike apo to krio
to
kalokairi
.
PASS agent
by-phr o andras me ta mavra gialia sinelifthi
apo tin
astinomia to vradi
30. Discussion: Overview of results
Embedding
no group was affected by type of embedding
sentences with AN-complements were more difficult for
children with reading difficulties and beginning readers.
AN complementiser was more difficult for children with read.
difficulties only.
Discourse anaphors
children with r.difficulties had problems with Indef.NPs, which
were more persistent (till end of sentence).
Passive morphology
Read.Difficulties: PASS sentences were more difficult (overall
reading) than ACT + DEP. but not due to verb features: no RT
differences on Verb.
Reflexive+Reciprocals: R.Diff.s+Beg.Readers read default
options faster.
Read.Difficulties: DEP were harder at the post-verbal segment
than ACT. Beg.Readers: DEP were as easy as ACT.
LDG7, Athens, June 2018 30
31. Discussion: Preliminary assumptions
Reading difficulties:
slower reading than controls but not compared to
Beginning Readers.
most difficulties found emerged at the total RTs of
sentences rather than critical segments general
processing rather than syntactic processing of specific
structures.
affected by passive morphology more than beginning
readers but overcome difficulty by the end of the
sentence (Deponents).
Differences between R.Diffs and Beg.Readers cannot
be due to memory (they have similar performance in
cognitive assessment), probably due to general
processing limitations.
LDG7, Athens, June 2018 31
32. Conclusion
Findings point towards a general
processing difficulty in reading
difficulties rather than syntactic
knowledge limitations.
Default (or less marked) structures
seem easier (e.g. in
reflexives+reciprocals).
Statistical and more fine-grained
analyses are required to draw safe
conclusions.
LDG7, Athens, June 2018 32
33. Thank you for your
attention!
LDG7, Athens, June 2018 33
34. References
Brown, H. (2006). Syntactic processing in individuals with
dyslexia: Using ERP to address the debate. Theses and
Dissertations (Comprehensive). 789.
http://scholars.wlu.ca/etd/789.
Craint, S. and Shankweilert, D. (1986). Syntactic complexity and
reading acquisition. In A. Davison, G. Green, & G. Hermon (Eds.),
Critical approaches to readability: Theoretical bases of linguistic
complexity. Hillsdale, NJ: Erlbaum, in press.
Leikin, M. and Bouskila, O.A. (2004). Expression of syntactic
complexity in sentence comprehension: A comparison between
dyslexic and regular readers. Reading and Writing 17(7-8), 801-
822. Available at
https://link.springer.com/article/10.1007/s11145-004-2661-1.
Scarborough, H.S. (1991). Early syntactic development of
dyslexic children. Annals of Dyslexia 41(1), 207-220. Available at
https://link.springer.com/article/10.1007/BF02648087.
LDG7, Athens, June 2018 34
35. Results: Embedding
LDG7, Athens, June 2018 35
0
500
1000
1500
2000
2500
seg1 seg2 seg3 seg4 seg5 seg6 seg7
Embedding (R.Difficulties)
ADV_all
MEAN
COMPL_all
MEAN
CE.RCs_all
MEAN
RB.RCs_all
MEAN
0
500
1000
1500
2000
2500
seg1 seg2 seg3 seg4 seg5 seg6 seg7
Embedding (Controls)
ADV_all
MEAN
COMPL_all
MEAN
CE.RCs_all
MEAN
RB.RCs_al
l MEAN
0
500
1000
1500
2000
2500
seg1 seg2 seg3 seg4 seg5 seg6 seg7
Embedding (Beg.Readers)
ADV_all
MEAN
COMPL_a
ll MEAN
CE.RCs_a
ll MEAN
RB.RCs_
all MEAN
36. Results: Relative Clauses
LDG7, Athens, June 2018 36
200
700
1200
1700
2200
2700
seg1 seg2 seg3 seg4 seg5 seg6 seg7
Relative Clauses (Controls)
CE_SUBJ
_PU
CE_OBJ_
PU
RB_SUBJ
_PU
RB_OBJ_
PU
200
700
1200
1700
2200
2700
seg1 seg2 seg3 seg4 seg5 seg6 seg7
Relative Clauses (R.Difficulties )
CE_SUB
J_PU
CE_OBJ
_PU
RB_SUB
J_PU
RB_OBJ
_PU
Seg1 Seg2 Seg3 Seg4 Seg5 Seg6 Seg7
CE_S I kopela pu ide ton andra lerose ton maθiti me boja
CE_O I kopela pu ide o andras lerose ton maθiti me boja
RB_S I kopela ide ton andra pu lerose ton maθiti me boja
RB_O I kopela ide ton andra pu lerose o maθitis me boja
200
700
1200
1700
2200
2700
seg1 seg2 seg3 seg4 seg5 seg6 seg7
Relative Clauses (Beg.Readers)
CE_SUB
J_PU
CE_OBJ
_PU
RB_SUB
J_PU
RB_OBJ
_PU
37. Results: Discourse Anaphors
LDG7, Athens, June 2018 37
0
500
1000
1500
2000
2500
3000
3500
4000
seg1 seg2 seg3 seg4 seg5
Discourse Anaphors (Controls)
PROPER NAME
DEF NP
INDEF NP
PRONOUN
0
1000
2000
3000
4000
5000
6000
seg1 seg2 seg3 seg4 seg5
Discourse Anaphors (Beg.Readers)
PROPER NAME
DEF NP
INDEF NP
PRONOUN
seg1 seg2 seg3 seg4 seg5
Pr.Name
o janis ke I athlitria pigan sto nosokomio epidi o Janis xripise to amaksi sto fanari
Def NP o janis ke I athlitria pigan sto nosokomio epidi I athlitria xripise to amaksi sto fanari
Indef NP
o janis ke I athlitria pigan sto nosokomio epidi mia jineka xripise to amaksi sto fanari
Pronoun
o janis ke I athlitria pigan sto nosokomio epidi afti xripise to amaksi sto fanari
0
1000
2000
3000
4000
5000
6000
seg1 seg2 seg3 seg4 seg5
Discourse Anaphors (Red.Difficulties)
PROPER NAME
DEF NP
INDEF NP
PRONOUN
38. Results: Passive Morphology
LDG7, Athens, June 2018 38
0
500
1000
1500
2000
seg3 seg4 seg5
Passive (Controls)
with other
PP
with by-
phrase
without
by-phrase
0
500
1000
1500
2000
seg3 seg4 seg5
Passive (R.Difficulties)
with other
PP
with by-
phrase
without
by-phrase
0
500
1000
1500
2000
seg3 seg4 seg5
Passive (Beg.Readers)
with other
PP
with by-
phrase
without by-
phrase
seg1 seg2 seg3 seg4 seg5
With
other PP o andras me ta mavra gialia sinelifthi se liga lepta to vradi
with by-
phrase o andras me ta mavra gialia sinelifthi apo tin astinomia to vradi
without
by-
phrase o andras me ta mavra gialia sinelifthi apo tragiko lathos to vradi
The reading process is complex and multi-dimensional. Recent work identified a number of components that need to be considered. There are four processes of reading
Several models……
The SDH attributes difficulties in the acquisition of reading to syntactic processing deficiencies.
PDH ….. Ex. Readers have the knowledge necessary to correct for example textual errors but are not able to detect them within the text.
There can be no doubt that syntactic complexity is an important factor in sentence processing. Some syntactic factors that affect complexity of a sentence are: ….
Non verbal tasks were given to children
…..
With : Increasing number of syllables
We employed a self passed reading task which measures participant reaction times (RT) per sentence segment. After considering several phenomena, we decided to include embedded clauses, discourse anaphors and passives.
two sentence plausibility tasks were used in order to check our items
Starting with embedding, As you can see in the sentences, in the first type of embedded clauses we included adverbial, temporal clauses. The action/event of the adverbial clause occurred either before, after or simultaneous to the action of the main clause. They were introduced with afou/otan/eno/prin.
In the second type, the verb complement clauses were introduced with the subordinators oti, which means that, an which means whether and pu which means that.
As regards embedding, we examined relative clauses. We investigated two types, right branching and center embedding.
Next, we investigated discourse anaphors of various types: null subject pronouns, proper names, overt personal pronouns, proper names and definite & indefinite NPs (pes ena paradeigma ap;o to kathena)
In passives, we examined structures with passive verbs, reflexives and reciprocals as you can see on the slide
We also looked at deponents (verbs which do not have an active counterpart) and anticausatives (intransitive verbs assigning a patient-theta role to its complement)
Student saw on the computer screen only one word or a phrase each time and they have to press “space”, so that they can read the next word (as you can see in the example). Each sentence was followed by a question that requires children to select the correct answer out of two possible answers.
COMPLEMENTISERS: AN is difficult for Read.Diff. but not for other groups. difficulty with indirect questions?
CONNECTIVES: RED.DIFFS: FASTER READING OF AFU+OTAN SENTENCES, NO DIFFERENCES FOR OTHER GROUPS.
AN SENTENCES ARE MORE DIFFICULT THAN OTI AND PU FOR THE R.DIFFICULTIES GROUP.
Indef.NPs difficult for Beg.R.s and Read.Diff.s at the point of the NP. Difficulty is eliminated in overall RT for Beg.R.s, but not for Read.Diff. more persistent difficulty with new information in discourse.
READ.DIFFS: ACT+DEP SENTENCES ARE GENERALLY EASIER TO READ (WHICH IS NOT FOUND IN OTHER GROUPS). HOWEVER, THIS DIFFERENCE IS NOT DUE TO THE VERB, AS THE EFFECT IS ONLY FOUND IN TOTAL RTS AND NOT IN V SEGMENT.
EXPECTED STRUCTURE (BY HIMSELF) IS EASIER FOR BOTH BEG.READERS AND R.DIFF.S. ADVERB CONDITION IS MOST DIFFICULT AMBIGUITY OF VERB MEANING REMAINS (NOT RESOLVED BY PHRASE IN SEGMENT 4). REFL+AGENT SIMILAR TO PASS+AGENT. DIFFERENCE NOT DUE TO EXPECTATIONS CAUSED BY VERB BUT BY THE BY-PHRASE IN SEGMENT 4?
EXPECTED STRUCTURE (EACH OTHER) EASIER FOR BOTH BEG.READERS AND R.DIFF.S
BEG. READERS: DEPONENTS ARE AS EASY AS ACTIVE, PASSIVES ARE MORE DIFFICULT. BUT R.DIFFICULTIES: DEPONENTS ARE AS DIFFICULT AS PASSIVES. DIFFICULTIES WITH PASSIVE MORPHOLOGY, WHICH IS TEMPORARY, LASTS UNTIL BY-PHRASE, NOT UNTIL END OF SENTENCE.