Presentation is loading. Please wait.

Presentation is loading. Please wait.

Argument Structure violation

Similar presentations


Presentation on theme: "Argument Structure violation"— Presentation transcript:

1 Argument Structure violation
Processing of lexical category and argument structure information in deverbal adjectives: An MEG study on Greek Kyriaki Neophytou1*, Christina Manouilidou2, Linnaea Stockall3 and Alec Marantz1,4 1New York University Abu Dhabi, 2University of Patras, 3Queen Mary University of London, 4New York University *Corresponding Author: Visual, form-based morphological decomposition Left Fusiform Gyrus TP (p<0.0) from ms Temporal-Occipital lobe (Letter string – fROI, [2]) TP (p=0.05), ms Syntactic category check Anterior Temporal Lobe (ATL) Violation Type (p=0.144), ms Lexical access: Stem Look-up OF (Surface frequency - fROI) Stem frequency (p<0.0), ms Recombination – Semantic Composition MTG Surface frequency (p=0.01), ms OF(Surface frequency - fROI) Suffix Type (p=0.001), ms Violation Type (p=0.018), ms Research Question MEG Results – Across Suffixes MEG Results – Within Suffixes* How are lexical category violations and verb argument structure violations processed in deverbal adjectives in Greek? STG -menos: Gram > ArgStr (p=0.013), ms OF -menos: ArgStr > Cat (p=0.043), ms -menos: Cat > Gram (p=0.040), ms -simos: Cat > Gram (p=0.020), ms  TP from stem to word (and not surface frequency) modulate brain responses in the left fusiform gyrus, from ms  Not all suffixes show the same pattern of activity  -menos yields most of the statistically significant differences in the within-suffix comparisons ENGLISH Category violation *re-blue(Adj.) re- requires verb Argument Structure violation *re-sleep (Intransitive Verb) re- requires an internal argument taking verb GREEK lahano-tos (cabbage-able) (Noun) -tos requires a verb tremi-tos (tremble-able) (Intransitive Verb) -tos requires an internal argument taking verb Violation Type (ArgStr and Cat violations) show the predicted effects in the ATL (although not statistically significant) -menos, Cat -menos, ArgStr -simos, Cat -simos, ArgStr -tos, Cat -tos, ArgStr Hypothesis & Predictions The lexical category of the stem is assessed before its argument structure [1]. Thus, we expect increased early brain activity for category violations followed by increased brain activity for argument structure violations, cross-linguistically. Previous Work Obligatory, early decomposition [2, 3] ~170ms post stimulus onset (PSO) in the left Fusiform gyrus: significant effect of Transition Probability (TP) from stem to word and lemma frequency Syntactic Category Determination [4] ~200ms PSO in the Anterior Temporal Lobe (ATL): significant effect of Noun/Verb entropy Lexical access [5,6] ~ ms PSO in the left Superior/Middle Temporal Gyrus (STG/MTG): significant effect of stem frequency and derivational family entropy, distinct processing of syntactic category and argument structure constraints Semantic wellformdness [6,7] ~ ms PSO in the left STG/MTG and the Orbitofrontal cortex (OF): significant effects of semantic coherence    Stem frequency shows a significant effect in the OF, but not in the temporal lobe *No statistically significant effects were found for -tos Behavioral Results Acceptance Rates: ArgStr (30-38%) > Cat (6-10%) Response Times (RTs): ArgStr(m=1227.7ms)>Cat(m=1146.1ms)> Gram(m=984.1ms)  Surface frequency shows a significant effect in the MTG -menos, Cat -menos, ArgStr -simos, Cat -simos, ArgStr -tos, Cat -tos, ArgStr Conclusions ArgStr violations are harder to process, compared to Cat violations (higher acceptance rate and higher RTs) [1] Evidence (TP effects) supporting early visual morphological decomposition processes in the left fusiform gyrus (M170) Evidence that the first step in lexical access is stem look up (stem frequency effect), followed by recombination (surface frequency effect) Violation Type evokes the predicted brain activity pattern: Cat>ArgStr in the early processing window in the ATL, and ArgStr>Cat in the later window in the OF The differences between the Violation Types vary among the 3 suffixes, requiring further investigation Material Suffix Grammatical Ungrammatical Category violation (Cat) Argument Structure violation (ArgStr) -simos ananeo-simos (renewable) varel-imos (barrel-able) gela-simos (laugh-able) -tos sevas-tos (respectable) lahano-tos (cabbage-able) tremi-tos (tremble-able) -menos lavo-menos (wounded) ahino-menos (urchin-able) argi-menos (be late-able)  Both the Suffix Type and the Violation Type significantly modulate the activity in the OF: ArgStr violations evoke greater activity compared to the Cat violations -150 illegal & 75 legal items (25 items per category) -75 fillers (legal adjectives) References Methods | Data Analysis (MEG Data) [1] Manouilidou & Stockall, Italian Journ. of Ling., 26. [2] Tarkiainen et al., Brain, 122. [3] Gwilliams, Lewis & Marantz, NeuroImage, 132. [4] King, Linzen & Marantz, [5] Fruchter & Marantz, Brain and Language, 143. [6] Stockall et al., Poster at SNL 2015. [7] Pylkkänen, Oliveri & Smart, Lang. & Cogn. Processes, 24. 20 Greek native speakers Lexical Decision Task 208 channel KIT MEG system 1000Hz sampling rate 200Hz online LPF Within-Suffix and Between-Violation Types comparisons Eeelbrain - Python package -Spatiotemporal cluster tests over anatomical ROIs -Temporal Cluster tests over functional ROIs (fROIs) Effects of lexical variables Lme4 – R package -Linear Mixed-Effects Models    Supported by the NYU Abu Dhabi Institute, Grant G1001


Download ppt "Argument Structure violation"

Similar presentations


Ads by Google