Early effects of morphological complexity on visual evoked fields in MEG

38
Early effects of morphological complexity on visual evoked fields in MEG Eytan Zweig & Liina Pylkkänen New York University 80 th Annual LSA meeting, January 7, 2006

description

Early effects of morphological complexity on visual evoked fields in MEG. Eytan Zweig & Liina Pylkkänen New York University. 80 th Annual LSA meeting, January 7, 2006. Morphological decomposition - Two questions. Do all affixed words decompose? Does semantic opacity play a role? - PowerPoint PPT Presentation

Transcript of Early effects of morphological complexity on visual evoked fields in MEG

Page 1: Early effects of morphological complexity on visual evoked fields in MEG

Early effects of morphological complexity on visual evoked

fields in MEG

Eytan Zweig & Liina PylkkänenNew York University

80th Annual LSA meeting, January 7, 2006

Page 2: Early effects of morphological complexity on visual evoked fields in MEG

Morphological decomposition -Two questions

• Do all affixed words decompose? Does semantic opacity play a role?

• What is the timing of lexical decomposition?

Page 3: Early effects of morphological complexity on visual evoked fields in MEG

Semantic opacity• Transparent words generally taken to decompose

• Competing hypotheses about opaque words

“Farmer” Farm + -er

“Folder”Fold + -er

Folder ?

1. Opaque words decompose (Rastle & Davis, 2003; Davis et al., 2004)

2. Opaque words do not decompose (Marslen-Wilson et al., 1994)

Page 4: Early effects of morphological complexity on visual evoked fields in MEG

Semantic opacity

• Previous experiments have produced contradictory results (as reviewed by Feldman et al., 2004).

Page 5: Early effects of morphological complexity on visual evoked fields in MEG

Timing of decomposition

Decomposition Lexical access

1. Early decomposition (Taft & Forster, 1975; Rastle & Davis, 2003; Davis et al., 2004)

3. Race (Baayen, 1992)

Lexical access Decomposition

2. Late decomposition (Feldman et al., 2004)

farm + -er farm, -er

farmer farm + -er

Decomposition Lexical access

Lexical access

farm + -er farm, -er

farmer

Page 6: Early effects of morphological complexity on visual evoked fields in MEG

ERP evidence for early decomposition in sentence processing

• Word category violations elicit an early left anterior negativity (ELAN) (Friederici, 2000; Friederici et al., 2002).

• Category is determined through morphological cues.

• Indirectly supports an early effect of morphology.

Page 7: Early effects of morphological complexity on visual evoked fields in MEG

This study

• Takes advantage of the millisecond temporal resolution of MEG.

• Simple lexical decision task without priming.

Page 8: Early effects of morphological complexity on visual evoked fields in MEG

Magnetoencephalography (MEG)

http://www.ctf.com/Pages/page33.html

EEGEEGMEGMEG

Page 9: Early effects of morphological complexity on visual evoked fields in MEG

MEG analysis

Page 10: Early effects of morphological complexity on visual evoked fields in MEG

MEG analysis

Page 11: Early effects of morphological complexity on visual evoked fields in MEG

MEG analysis

Page 12: Early effects of morphological complexity on visual evoked fields in MEG

MEG analysis

Page 13: Early effects of morphological complexity on visual evoked fields in MEG

Typical MEG response to visual words

Page 14: Early effects of morphological complexity on visual evoked fields in MEG

Experiment 1Suffixed words

Page 15: Early effects of morphological complexity on visual evoked fields in MEG

Experiment 1 Stimuli

No SuffixNo Suffix Orth. -er

Opaque Suffix

Transparent

Suffix

SWITCH WINTER FOLDER FARMER

34 words per condition

Conditions were controlled for matched for length, surface frequency, orthographic neighborhood density and frequency, and syntactic category.

Suffixed conditions were further controlled for stem frequency and orthographic regularity.

Page 16: Early effects of morphological complexity on visual evoked fields in MEG

Testing for pre-lexical effects

• M350 is the first component to show effects of lexical factors (Embick et al., 2001;

Pylkkänen & Marantz, 2003).

• Fiorentino & Poeppel (2003) found that the M350 is sensitive to constituent frequency in compounds. – Decomposition likely to happen before M350.

Page 17: Early effects of morphological complexity on visual evoked fields in MEG

Prediction - Timing

• If decomposition is pre-lexical, we should find effects before the M350.

Page 18: Early effects of morphological complexity on visual evoked fields in MEG

M170

• The first component that has been found to be sensitive to the presence of letter strings (Tarkiainen et al., 1999).

• Most research has found that the M170 is not sensitive to lexical factors such as frequency.

Page 19: Early effects of morphological complexity on visual evoked fields in MEG

M170

• Fusiform gyri have been found to be the primary generators of M170 activity.

Page 20: Early effects of morphological complexity on visual evoked fields in MEG

Fusiform Gyri

• Functional asymmetry between hemispheres (Tarkiainen et al., 2002) :

– The left fusiform gyrus sensitive to letter strings (Cohen et al., 2000; Dehaene et al., 2002).

• “Visual Word Form Area”

– The right fusiform gyrus primarily sensitive to faces.

Page 21: Early effects of morphological complexity on visual evoked fields in MEG

Prediction – Semantic opacity

• If opaque words decompose, they will pattern with transparent words.

• If they do not, they will pattern with orthographic controls.

Page 22: Early effects of morphological complexity on visual evoked fields in MEG

-15

-10

-5

0

5

10

15

20

25

30

-50 -16 18 51 85 119 153 187 220 254 288 322 356 390

Suffixed

Not Suffixed

M170 - Left Hemisphere Grandaveraged waveform

n=16 no effect

Time

nAm

Page 23: Early effects of morphological complexity on visual evoked fields in MEG

10

12

14

16

18

20

22

24

26

28

30

No Suffix No Suffix, Orth. Match Opaque Suffix Transparent Suffix

M170 - Left Hemisphere Amplitudes

SWITCH WINTER FOLDER FARMER

nAm

Page 24: Early effects of morphological complexity on visual evoked fields in MEG

-15

-10

-5

0

5

10

15

20

25

30

-50 -16 18 51 85 119 153 187 220 254 288 322 356 390

Suffixed

Not Suffixed

M170 - Right Hemisphere Grandaveraged waveform

n=16 p < 0.001

Time

nAm

Page 25: Early effects of morphological complexity on visual evoked fields in MEG

M170 - Right Hemisphere Amplitudes

10

12

14

16

18

20

22

24

26

28

30

No Suffix No Suffix, Orth. Match Opaque Suffix Transparent Suffix

SWITCH WINTER FOLDER FARMER

nAm

Page 26: Early effects of morphological complexity on visual evoked fields in MEG

Other measures

• No M100 effects.

• No M350 effects.

• No response time effect.

Page 27: Early effects of morphological complexity on visual evoked fields in MEG

Conclusions

• M170 activity is influenced by the presence of derivational suffixes.• Early lexical decompositon.

• No effect of opacity• All affixed words decompose.

Page 28: Early effects of morphological complexity on visual evoked fields in MEG

Conclusions

• The right lateralization of the effect is surprising.

• Early visual word processing may be bilateral, with distinct functional roles for the left and the right hemispheres.

• However, there is a second hypothesis.

Page 29: Early effects of morphological complexity on visual evoked fields in MEG

Why the right hemisphere?F

AR

ME

R

FA

RM

ER

FA

RM

ER

Page 30: Early effects of morphological complexity on visual evoked fields in MEG

Experiment 2Prefixed words

Page 31: Early effects of morphological complexity on visual evoked fields in MEG

Experiment 2 Stimuli

No Prefix No Prefix Orth. re-

Prefix

ROTATE RESUME REFILL

32 words per condition

Conditions were controlled for matched for length, surface frequency, orthographic neighborhood density and frequency, and syntactic category.

Prefixed condition was further controlled for stem frequency and orthographic regularity.

Page 32: Early effects of morphological complexity on visual evoked fields in MEG

M170 - Right Hemisphere Grandaveraged Waveform

-10

-5

0

5

10

15

20

25

30

35

-50 -16 18 51 85 119 153 187 220 254 288 322 356 390

Prefixed

No Prefix, Ortho re-

No Prefix

Time

nAm

n=10 p < 0.02

Page 33: Early effects of morphological complexity on visual evoked fields in MEG

M170 - Left Hemisphere Grandaveraged Waveform

-10

-5

0

5

10

15

20

25

30

35

-50 -16 18 51 85 119 153 187 220 254 288 322 356 390

Prefixed

No Prefix, Ortho re-

No Prefix

n=10 p < 0.02

Time

nAm

Page 34: Early effects of morphological complexity on visual evoked fields in MEG

Other measures

• No M100 effects.

• No M350 effects.

• No response time effect.

Page 35: Early effects of morphological complexity on visual evoked fields in MEG

Conclusions

• Experiment 1 replicated.

• Stem lexicality cannot be the sole reason for the RH effect in Experiment 1.

• Possibly a combined effect of morphological complexity and stem lexicality.

Page 36: Early effects of morphological complexity on visual evoked fields in MEG

/ta-ba-ko/

KanaKanji

smoke-weed

Page 37: Early effects of morphological complexity on visual evoked fields in MEG

Nakamura et. al (2005)

FMRI activity in fusiform gyri

Page 38: Early effects of morphological complexity on visual evoked fields in MEG

Conclusions

• Morphological decomposition is a pre-lexical effect, and is not sensitive to semantic opacity.

• The RH M170 source is sensitive to morphological complexity irrespective of the linear ordering between a stem and an affix. • A crucial role for the right hemisphere in early

word processing.