Document Sample
Reem_CNS-5.ppt Powered By Docstoc
					                                                    Neural correlates of diglossic code-switching:
                                       an ERP study of Modern Standard Arabic and Palestinian Colloquial Arabic
                                                                                               Reem Khamis Dakwar, Karen Froud
                                                                                              Teachers College, Columbia University
Introduction                                                                                           Experiment                                                                                                                                                                                            Results
Arabic Diglossia                                                                                         78 control sentences (39 in PCA, 39 in MSA)
  Primary dialect: everyday communication
                                                                 MSA/CA OR         Colloquial
                                                                                              MSA/CA                                                                                                                                                                                                            Significant N200 effect found in overall analysis (PCA & MSA collapsed)
  Superposed variety: formal settings and                                                                  Wen Na:m-at             ?il- binet         Laa    tukθir  mina        ?l-kala:m                                                                                                                      No significant N200 effect found for MSA to PCA codeswitches
  written literature (Ferguson,1959)                                                                       Where slept-2.Sg.Fem the-gir               Neg. excess-2.Sg.Masc. From the- talk
                                                                                                                                                                                                                                                                                                                Highly significant negativity in the early time window for codeswitched sentences
                                                                                                                                                                                                                                                                                                                - only for PCA to MSA codeswitches
  Distinctions between these two varieties exist in all language domains                                   PCA PCA                 PCA-PCA            MSA MSA                MSA-MSA
Aims of the study                                                                                          Where did the girl sleep?                  Don’t speak too much.
  The present study seeks to investigate the neurofunctional bases of diglossic code-
  switching in Arabic
  Such investigation may assist in determining whether MSA and dialectal varieties of                    78 sentences ended with a semantically anomalous word (39 in PCA, 39 in MSA)
  Arabic are neurally represented as distinct or inter-related linguistic systems
  also provides a bases for pedagogical approaches in teaching Arabic                                      Wen    Na:m-at          ?il- tawle         Laa Tukθir       Mina       ?al-malal
                                                                                                           Where slept-2.Sg.Fem the-table             Neg. excess-2.Sg.Masc. From the- bore                                                                                                                           N200: all sentences                               MSA to PCA switches                                 PCA to MSA switches
Background                                                                                                 PCA PCA                 PCA-PCA            MSA MSA                MSA-MSA
  No neurophysiological methods have been utilized in investigations of Arabic                             Where did the table sleep?                 Don’t bore too much.                                                                                                                                     N400 & LPC
  codeswitching                                                                                                                                                                                                                                                                                                   When all codeswitched and semantically anomalous sentences were
  EEG studies of Arabic (Boudelaa & Marslen-Wilson, 2002, 2004, 2005) have focused on                                                                                                                                                                                                                             compared to controls, N400 effects were visible but not significant for
  lexical representations in MSA                                                                                                                                                                                                                                                                                  semantically anomalous sentences; LPC was significantly more positive than
  Event-related potentials (ERPs) have the promise to provide information about the                      78 sentences ended with a semantically appropriate word spoken in the alternate language                                                                                                                 to control sentences, for codeswitched sentences only
  underlying neural representation of languages that exist in a diglossic situation                      variety (39 switched from MSA to PCA; 39 switched from PCA to MSA)                                                                                                                                       N400 was more significantly more negative to control sentences in PCA, but
  Jackson et al 2001, Moreno et al 2002: Late Positive Complex (LPC) associated with
  switching from one language to another                                                                                                                                                                                                                                                                          significantly more negative to semantically anomalous sentences in MSA
                                                                                                           Wen    Na:m-at         ?il- qiţa?          Laa Tukθir Mina        ?al-ħaki                                                                                                                             Late Positive Complex was statistically more positive than control sentences in
  LPC: positivity over occipito-parietal sensors, around 600-800 milliseconds after onset of
  a codeswitch                                                                                             Where slept-2.Sg.Fem the-cat               Neg. excess-2.Sg.Masc. From the- talk                                                                                                                       both languages, for the codeswitched sentences ONLY
  Moreno et al 2002: switching within a language – e.g. introducing a semantically                         PCA    PCA             PCA-MSA             MSA MSA MSA             MSA-PCA
  unexpected word – was not associated with LPC effects                                                    Where did the cat sleep?                   Don’t speak too much.

Stimuli: He heard a knock at the ….      door (expected ending) (solid line)
                                                       entrance (semantic violation)                                                         Computer-generated                                                         Attentional Task
                                                                                                            Digital recording
                                                                     puerta (codeswitching)                                                     Presentation                                                                 After
                                                                                                             Of sentences
                                                                                                                                                 (E-prime)                                                            Sentence Presentation
                                                                                                                                                                                                                                                                                                                 N400/LPC: All sentences                                N400/LPC: PCA only                                        N400/LPC: MSA

                                                                                                       Behavioral Results                                                                                                                                                                                        Our results support the view that PCA and MSA are represented in the brain as two
                                                                                                                                                                                                                                                                                                                 separate languages
                                                                                                                                                                                             50                                                   3000                                                           Moreover, N200 elicitation in PCA MSA switches only suggests that PCA is
                                                                                                         No significant effects of condition on reaction times                               45
                                                                                                                                                                                                                                                  2500                                                           represented as a first language, with MSA as the L2 (for these subjects)
                                                                                                         Significantly more errors were made in answering

                                                                                                                                                                                             35                                                                                                                  N400 results are less clear, and may be confounded – possibly by familiarity effects?

                                                                                                                                                                                                                                                         R ea c tio n T im e (m se c )
                                                                                                         the attentional question when the sentences had

                                                                                                                                                                 P erc e n ta g e E rro rs
                                                                                                                                                                                             30                                                                                                                  Major implications for educational approaches
                                                                                                         been codeswitched (false positive responses)                                        25                                                   1500

Hypotheses                                                                                               Semantic anomaly had no reliable effect on any

                                                                                                                                                                                                                                                  1000                                                       Future Directions
                                                                                                         behavioral measure                                                                  10                                                                                                                  Extend present study to include more subjects from uniform linguistic / educational
Based on these findings, we hypothesize                                                                                                                                                                                                           500
                                                                                                                                                                                             5                                                                                           Error Percentages       backgrounds; and duplicate study for different dialects
    N400 responses: only in within-language anomaly (unexpectedness)                                                                                                                         0                                                    0                                      Reaction Times
                                                                                                                                                                                                  Control   Semantically Anomolous   Codeswitch                                                                  Examine effects of codeswitching outside Arabic – e.g. switching between English
    N200 responses: MSA and PCA as distinct linguistic systems                                                                                                                                                    Condition
                                                                                                                                                                                                                                                                                                                 and Arabic, for bilingual speakers
      • Limited to L1   L2 switches
                                                                                                                                                                                                                                                                                                                 Examine effects of combining diglossic codeswitching and semantic anomaly in the
    LPC: MSA and PCA as distinct linguistic systems
                                                                                                                                                                                                                                                                                                                 same sentences
                                                                                                       EEG data analysis                                                                                                                                                                                         Investigations of structural constraints on diglossic codeswitching
EEG acquisition                                                                                                                                                                                                                                                                                                  Similar linguistic situations, such as SAE vs AAVE
   128-channel geodesic sensor net (Electrical Geodesics, Inc.[6]) was used to detect
                                                                                                         Pre-processing: 40 Hz low-pass digital filter applied, artifact rejection, baseline correction
    changes in electrical potential at the subject’s scalp                                                                                                                                                                                                                                                   References
                                                                                                         Epochs segmented 100 and +800 ms from target word onset
   Sensor impedances were kept below 50kΩ                                                                                                                                                                                                                                                                    Boudelaa, S., Gaskell, G.M. (2002). A re-examination of the default system for Arabic plurals. Language and Cognitive Processes, 17, 3, 321-343.
                                                                                                         Behavioral error response or timeout rejection                                                                                                                                                      Boudelaa, S. & Marslen-Wilson, W. D. (2000). Non-concatenative morphemes in language Processing: Evidence from Modern standard Arabic.
   Signal was continuously sampled at 250 Hz via a high input impedance amplifier                                                                                                                                                                                                                                Proceedings of Workshop on Spoken Word Access Processes, 1, 23,-26.
                                                                                                         Epochs associated with the same trial type were averaged within subjects                                                                                                                            Boudelaa, S. & Marslen-Wilson, W. D. (2004a). Abstract morphemes and lexical representation: The CV-skeleton in Arabic. Cognition, 92, 3, 271-303.
                                                                                                                                                                                                                                                                                                             Boudelaa, S. & Marslen-Wilson, W. D. (2004b).Allomorphic variation in Arabic: Consequences for lexical processing and lexical architecture. Brain and
                                                                                                         Data from all subjects were re-referenced offline to the average reference                                                                                                                              Language, 90, 106-116.
Participants                                                                                             Montage selection derived from previous studies (Moreno et al., 2002)                                                                                                                               Ferguson, C. (1959). Diglossia. Word, 15, 325-340.
                                                                                                                                                                                                                                                                                                             Jackson, G.M., Swainson, R., Mulin, A., Cunnington, R., & Jackson, S. R. (2004). ERP correlates of receptive language-switching tasks. The Quarterly
  5 native speakers of Palestinian Colloquial Arabic (Galilee dialect)                                                                                                                                                                                                                                           Journal of Experimental Psychology, 57A, 2, 223-240.
                                                                                                                                                                                                                                                                                                             Moreno, E.M., Federmeier, K.D. & Kutas, M. (2002). Switching languages, switching palabras (words): an electrophysiological study of codeswitching. Brain
  All right-handed                                                                                     N200:                                        N400 & P600:                                                                                                                                                 & Language, 80, 188-207.

  Four males, one female                                                                               Time window 100-300 ms                       Time window: 200-800 ms
  Mean age 30 years, 2 months (range: 25;10-36;2)                                                      fronto-central quadrant                      occipito-parietal region                                                                                                                                 Acknowledgements
                                                                                                                                                                                                                                                                                                             Thanks to everyone in the Neurocognition of Language Lab, all our participants, and especially
  Normal hearing and normal or corrected-to-normal vision
                                                                                                                                                                                                                                                                                                             Jamil, Fouad, Nadey, John & Sam. Contact us: /
  All exposed to MSA since first grade, in Arab schools in Israel

Shared By: