print exposure and memory effects on pronominal resolution › langlab › pubs › files ›...

Fleva, E.1, Fotiadou, G.1, Katsiperi, M.1, Peristeri, E.1, Mastropavlou, M.2 & Tsimpli, I. M.1, 3

1Aristotle University of Thessaloniki (Greece), Language Development Lab

2University of Ioannina (Greece) 3University of Reading (UK), School of Psychology and Clinical Language Sciences

Print exposure and memory effects on pronominal resolution

EXCELLENCE I IDAR Project

Experimental Psycholinguistics, Madrid, 1-3 October 2014

• to examine the vulnerability of anaphora resolution in adult monolingual native speakers as a function of individual differences in:

• Memory abilities

(working memory and episodic memory)

• Language experience

(print exposure and education)

• Age

Aims


Anaphora Resolution

• Anaphora resolution (AR) is a linguistic phenomenon which describes the process of identifying the referent of anaphoric expressions into the context.

• AR is computed in syntax and at the interface with discourse (Tsimpli, et al. 2004; White 2008 a.o.).

• Recent research identified both linguistic and cognitive factors conditioning AR (Hendriks et al., 2014;

van Rij et al., 2013).


Overt vs. Null pronoun resolution

• Greek is a Null Subject Language (NSL)

• Ambiguity in antecedent preference:

(1) O Kostasi xheretise ton Giannik otan proi /aftosk ton plisiase.

the Kostas waved at the John when pro / he him approached

(Papadopoulou et al. in press; but Tsimpli et al. 2004)


Previous studies: Age effects on AR Older adults are less strategic in encoding /

retrieving the most important information from discourse

and overuse ambiguous pronouns

BUT

have no problems disambiguating a pronoun on the basis of information presented in an immediately preceding sentence

(Hendriks et al., 2014; Light & Capps, 1986; Titone, Prentice, & Wingfield, 2000 a.o.)


Internal Factors

Linguistic Factors

Discourse Effects

Age Anaphora Resolution

•Linguistic Form (overt /∅) •Syntactic Complexity (Word Order)

•Cognitive Abilities •Language Experience


Language experience

• may lead to variability in sentence comprehension performance.

how do we measure language experience?

• print exposure (Stanovich & West 1989)

• education level (Dabrowska 1997; Dabrowska & Street 2006)

Familiarity with print is positively correlated with better word recognition abilities (Chateau

& Jared, 2000; Fotiadou et al. 2014)


Research Questions and Predictions

• Age, Print exposure and Cognitive measures of memory will affect pronominal resolution

• Linguistic properties: Word-order and Context will affect pronominal resolution in all participants


METHODOLOGY

Participants

Materials & Procedure

Independent variables: Age & Education

10% 12,5%

2,5%

55,0%

15%

5%

0%

10%

20%

30%

40%

50%

60%

School Post-high School

College University Master PhD

Participants' Educational Level (%)

0

2

4

6

18 20 21 23 25 26 28 29 31 32 33 34 35 37 38 41 42 50 52 54 57 61 63 69 71 72 73 75

N o

f p

arti

cip

ants

Distribution of Participants' Age

Male and female monolingual Greek speaking adults (N= 40, M= 40.7; Age range: 18-75)

Backwards digit recall task (Alloway, 2007)

To assess verbal working memory: Oral presentation of series of numbers with gradient difficulty

Immediate and delayed recall story task (Rivermead:

adapted for Greek by Efklides et al., 2002)

To assess episodic memory: Oral presentation of a story with 21 ideas. Participant’s performance is scored according to the number of ideas repeated.

Language Experience

Author and Magazine Recognition Tests (Fotiadou et al.,

2014; modelled after Stanovich and West, 1989)

Education levels (questionnaire)

Results:

0%

5%

10%

15%

20%

25%

30%

35%

L3 L4 L5 L6 L7

WM (%)

0

5

10

15

20

25

30

35

40

ART MRT

ART-MRT scores (Mean)

0

5

10

15

20

tell retell

Rivermead


Word-order (topicality) & context effects

Topicality & word order the first out of two or more entities in a sentence – often the

grammatical subject and the continuing discourse topic – is the most salient one in the preceding discourse

(Ariel’s accessibility theory 1998; 1990; Arnold et al. 2000; Reinhart 1981)

Context contextual information prior to the target sentence helps

resolution in comparison with isolated sentences (eg. van Berkum, Brown & Hagoort ,1999: ERP findings)


Linguistic tasks self-paced listening antecedent identification

sentence-picture matching task

(E-prime: Schneider et al., 2002)

The effect of Word Order : SVO (unmarked) vs. OVS (marked) sentences in null (Condition 1) OR overt (Condition 2) pronouns

The effect of a preceding context with or without explicit reference to the actors: SVO sentences in null (Condition 1) OR overt (Condition 2) pronouns

The SPL experiments described include:

• 10 experimental items per condition

• 20 filler sentences

• 5 practice sentences


Duration: 10-20 min approx.

Subject Referent

Object Referent

Other Referent

WORD ORDER

CONDITION 1

a. The old lady -NOM / waved at/ the pupil -ACC / when / she / was crossing/ the street.

b. The pupil -ACC/ her –Cl waved at / the old lady -NOM/ when / she / was crossing/ the street.

CONDITION 2

a. The old lady -NOM / waved at/ the pupil -ACC / when / Ø already/ was crossing/ the street.

b. The pupil -ACC/ her –Cl waved at / the old lady -NOM / when /Ø already/ was crossing/ the street.

Who was crossing the street?

(a) Because the show was not going well, during the break a ballerina FEM got angry and

splashed the singer FEM.

(b) Because the show was not going well, during the break the members of the group got very

angry.

CONDITION 1: The ballerina FEM/ approached / the singer FEM / when / she / was entering / the

stage.

CONDITION 2 : The ballerina FEM/ approached / the singer FEM / when / Ø slowly/ was entering/

the stage. Who was entering the stage?

Subject Referent

Object Referent

Other Referent

Various combinations of pictorial stimuli Various syntactic structures in the sentences used


3 measures in the self-paced listening referent-matching task

online listening times (in msecs) per segment

offline RTs (in msecs) on matching decisions

Preferred referent (%)

The Analyses


• Aftos Pronoun

• Null Pronoun

Word Order

0

500

1000

1500

2000

2500

The old ladyNOM The pupilACC

waved at herCl waved at

the pupil ACC the old ladyNOM

when she was crossing the street

svoaftos oclovsaftos

* *

*

0

500

1000

1500

2000

2500

The old ladyNOM The pupilACC

waved at herCl waved at

the pupil ACC the old ladyNOM

when ∅ already was crossing the street

svonull oclvsnull

*

• Aftos Pronoun

Context

• Null Pronoun

(a) Because the show was not going well, during the break a ballerina FEM got angry and splashed

the singer FEM.

(b) Because the show was not going well, during the break the members of the group got very

angry.

500

1000

1500

2000

The ballerina approached the singer when she was entering the stage

* *

500

1000

1500

2000

The ballerina approached the singer when ∅ already was entering the stage

* *

Integration of known information slow down

RESPONSE LATENCIES PREFERRED REFERENTS

0

20

40

60

80

100

svo_aftos oclvs_aftos

Word Order - Overt Pronoun Reference

Subject

Object

other

0

20

40

60

80

100

svo_null oclvs_null

Word Order - Null Pronoun Reference

Subject

Object

other

3000

3200

3400

3600

3800

4000

4200

4400

4600


aftos

*

3000

3200

3400

3600

3800

4000

4200

4400

4600

svo_null oclvs_null

∅

Word order markedness Overt P

0

20

40

60

80

100


Word Order - Overt Pronoun Reference

Subject

Object

other

0

20

40

60

80

100

svo_null oclvs_null

Word Order - Null Pronoun Reference

Subject

Object

other

WM r=-.443; p=.004

Lang.exp. r=-.406; p=.011

WM r=.560; p=.002

Episodic (immediate & delayed)

r=.389; p=.041 & r=.458; p=.014

Experimental Psycholinguistics, Madrid,

1-3 October 2014

0

10

20

30

40

50

a_aftos b_aftos

Context - Overt Pronoun Reference

Subject

Object

Other

0

10

20

30

40

50

a_null b_null

Context - Null Pronoun Reference

Subject

Object

Other

2500

2700

2900

3100

3300

3500

a_aftos b_aftos

aftos

2500

2700

2900

3100

3300

3500

a_null b_null

∅

No reference slower Response

*

*

*

(a) Because the show was not going well, during the break a ballerina got angry and splashed the singer.

(b) Because the show was not going well, during the break the members of the group got very angry.

0

10

20

30

40

50

a_aftos b_aftos


Subject

Object

Other

0

10

20

30

40

50

a_null b_null


Subject

Object

Other



r=.402; p=.011

r=-.322; p=.046

Correlation with AGE:


0

10

20

30

40

50

a_aftos b_aftos


Subject

Object

Other

0

10

20

30

40

50

a_null b_null


Subject

Object

Other



r=-.376; p=.018 r=-.320;

p=.047

Lang. Experience (ART)


0

10

20

30

40

50

a_aftos b_aftos


Subject

Object

Other

0

10

20

30

40

50

a_null b_null


Subject

Object

Other



Episodic M r=-.436; p=.033

Educ. Level: r=-.441; p=.031


Summary and Conclusions • Linguistic properties : a. Word-order affected antecedent preference and RTs in

the overt pronoun conditions only b. Context affects pronominal resolution in the overt

condition (more subject antecedent in the +repeated NP condition)

Age effects in the neutral context: older participants selected ‘other’ referent more than object in the overt pronoun condition

Language experience affects antecedent preference in the overt pronoun condition (more language experience increases ambiguity in the subject/object antecedent preference)

Memory ability: Lower scores in episodic memory task more selection of ‘other’ referent in neutral context


Overall, • Context : Pronominal resolution shows effects of

‘internal factors’ (age, language experience and episodic memory) in selected conditions only and primarily with overt pronouns

• This is consistent with work on attrition and second language learners showing that overt pronouns in NSL are affected

• Word-order : Memory and language experience affects the degree of overriding the default subject antecedent preference in the null pronoun condition


Selected References Alloway, T.P. (2007). Automated Working Memory Assessment. London, UK: Harcourt Assessment Arnold, J. E., Brown-Schmidt, S., & Trueswell, J. (2007). Children’s use of gender and order of mention during pronoun comprehension.

Language and Cognitive Processes, 22, 527-565. Van Berkum, J. J. A., Brown, C. M., & Hagoort, P. (1999). Early referential context effects in sentence processing: Evidence from event-related

brain potentials. Journal of Memory and Language, 47, 147–182. Carminati, M. (2002). The Processing of Italian Subject Pronouns. PhD dissertation, University of Massachusetts Amherst. Chateau, D., & Jared, D. (2000). Exposure to print and word recognition processes. Memory and Cognition, 28, 143–153. Corbett, A. T., & Chang, F. R. (1983). Pronoun disambiguation: Accessing potential antecedents. Memory & Cognition, 11, 283- 294. Dimitriades, A. (1996). When Pro-Drop languages don’t: overt pronominal subjects and pragmatic inference. In: Dobrin, L.M., Singer, K. &

L.McNair (eds.) CLS 32: The Main Session, 33-47. Dwivedi, V., & Goldhawk, M. (2009). Underspecification of scope ambiguity and anaphora: Evidence from self-paced reading. Poster

presentation for the 22nd annual meeting of the CUNY Sentence Processing Conference, University of California, Davis, USA. Fotiadou, G., Fleva, E., Katsiperi, M. & Tsimpli, I. M. (2014). Exploring the effects of print exposure on lexical processing among Greek adults. In

Proceedings of the International Psychological Applications Conference and Trends (InPACT), Porto, Portugal., 310-312. Efklides, A., Yiultsi, E., Kangellidou, T., Kounti, F., Dina, F., & Tsolaki, M. (2002). Wechsler Memory Scale, Rivermead Behavioral Memory Test,

and Everyday Memory Questionnaire in Healthy Adults and Alzheimer Patients. European Journal of Psychological Assessment, 18(1), Page 63-77

Hendriks, P., Koster, C., Hoeks, J. (2014). Referential choice across the lifespan: Why children and elderly adults produce ambiguous pronouns. Language and Cognitive Processes

Kaltsa, M. Tsimpli, I.M., Rothman, J. (in press) Exploring the source of differences and similarities in L1 attrition and heritage speaker competence: evidence from pronominal resolution. Lingua

Miltsakaki, E. (2003). The Syntax-Discourse Interface: Effects of the Main-Subordinate Distinction on Attention Structure. PhD dissertation, University of Pennsylvania.

Myers, J. L., & O’Brien, E. J. (1998). Accessing the discourse representations during reading. Discourse Processes, 26, 131-157. Nieuwland, M. S., & Van Berkum, J. J. A. (2006). When peanuts fall in love: N400 evidence for the power of discourse. Journal of Cognitive

Neuroscience, 18(7), 1098-1111. Papadopoulou, D. Peristeri, E., Plemmenou, L., Marinis, T., Tsimpli, I.M. (in press) Pronoun ambiguity resolution: Evidence from monolingual

adults and children. Lingua Tsimpli, I.M., Sorace, A., Heycock, C. & Filiaci, F. (2004). First Language Attrition and Syntactic Subjects: A Study of Greek and Italian Near-Native

Speakers of English. International Journal of Bilingualism, 8, 257-277 van Rij J., van Rij, H., Hendriks, P. (2013). How WM load influences linguistic processing in adults: a computational model of pronoun

interpretation in discourse. Topics in Cognitive Science, 5, 564–580

Special thanks to the participants

&

Thank you for your attention!

This research has been co-financed by the European Union (European Social Fund – ESF) and Greek national funds through the Operational Program "Education and Lifelong Learning" of the National Strategic Reference Framework (NSRF)

print exposure and memory effects on pronominal resolution › langlab › pubs › files ›...

Documents