Visual word recognition of morphologically complex words: Effects of prime word and root frequency Hélène Giraudo Karla Orihuela [nam Laboratoire CLLE (Equipe ERSS) Laboratoire CLLE (Equipe ERSS) [addr CNRS & Université Toulouse Jean Jaurès CNRS & Université Toulouse Jean Jaurès [addr giraudo@univ-tlse2.fr karla.orihueal@univ-tlse2.fr [addr [e-ma raudo & Voga, 2007; 2014; Giraudo & Grainger, Abstract 2000; 2001; but see also Aronoff, 1994 and Booij, 2010 for the same linguistic view) or for The present study aims to investigate the the access ways to the lexicon, morphology in- relative role of the surface frequencies fluencing the simple development of orthograph- (i.e., token frequencies) in base word ic representations (e.g., Duñabeitia et al., 2007; recognition. A masked priming experi- Rastle & Davis, 2003; Rastle et al., 2004 and see ment was carried using two types of suf- in the same vein Marantz, 2013). An interesting fixed French primes: the effects of words way to explore this issue is to use the masked having a surface frequency (SF) higher priming paradigm (Forster & Davis, 1984) which than their base (e.g., cigarette – cigare) has been designed to measure the qualitative and were compared with those produced by the quantitative effects induced by the prior pro- word primes having a SF lower than their cessing of a morphologically complex word pre- base (e.g., froideur-froid ‘coldness- sented visually on the subsequent processing of cold’). Results show that HighSF are another -target- word. Behavioural data obtained more efficient primes than LowSF rela- with the masked priming paradigm associated tive to both orthographic and unrelated with the lexical decision task revealed clear priming baselines. This suggests that de- strong morphological priming effects through spite a highly salient base, whole words various languages (Arabic: Boudelaa & Marslen- matter more than morphemes during the Wilson, 2001; Basque: Duñabeitia, Laka, Perea, early processes of lexical access. & Carreiras, 2009; English: Rastle, Davis, 1 Introduction Marslen-Wilson & Tyler, 2000; French: Giraudo & Grainger, 2000; German and Dutch: Drews Morphological complexity has been extensively and Zwitserlood, 1995; Greek: Voga & Grainger, explored by psycholinguists in order to shed light 2004; Hebrew: Frost, Deutsch & Forster, 1997) on the role of morphology in lexical structuring. but the results are still controversial when ma- Starting from the idea - inherited from the con- nipulating the relative frequencies of the prime nectionist theory of visual word recognition (see and the target. On the one hand, some studies Seidenberg, 1987) - that the lexicon is comprised (Giraudo & Grainger, 2000) have revealed that of different levels of interconnected representa- larger effects are obtained when using high in tions reflecting the linguistics characteristics of comparison to low frequency derived primes en- the words as well as the cognitive processes by couraging the lexeme-based approach; on the the which complex words are recognized, the other, some authors (McCormick, Rastle, & Da- main issue regarding lexical morphology con- vis, 2009) have failed to observe an interaction cerns its specific role relative to word forms and between the prime frequency and morphological semantics. Accordingly, morphology can be facilitation, strengthening the morpheme-based though as a structuring factor either for the lexi- approach. It has been suggested that these out- con, morphological relationships being expressed comes may be due to the fact that the methodo- by the mapping between from and meaning re- logical procedure among experiments varies flecting the construction of the words (e.g., Gi- (Amenta & Crepaldi, 2012), as they each use a Copyright © by the paper’s authors. Copying permitted for private and academic purposes. In Vito Pirrelli, Claudia Marzi, Marcello Ferro (eds.): Word Structure and Word Usage. Proceedings of the NetWordS Final Conference, Pisa, March 30-April 1, 2015, published at http://ceur-ws.org 128 different timing for their stimulus onset asyn- general average and outliers were removed chrony (SOA). The SOAs of the original experi- (1.4%). ments were of 57ms in Giraudo and Grainger (2000) and 42ms in McCormick, Rastle, & Davis Table 1: mean reaction times across the three (2009). priming conditions and the two targets condi- tions. Net priming effects are expressed in ms. 2 The study In order to disentangle such findings, the present RTs Net priming effects study was carried out using the same paradigm (U-M/O-M and U-O) and similar SOAs as the previous ones described. HighSF M 613 +45* / + 36* The main manipulation was to compare morpho- primes- primes logical facilitation when the frequency of the LowSF O 649 +9 complex words (used as primes) and their roots targets primes (used as targets) was inverted. More specifically U 658 we selected 60 base word targets from French, primes half having systematically a surface frequency LowSF M 572 +22 / +18 higher than their derived forms (55.82 primes- primes occ./million) and the other half a surface fre- HighSF O 590 +4 quency lower than their derived forms (10.15 targets primes occ./million according to Lexique database, New, U 594 Pallier, & Ferrand, 2001). Each target was primes primed by (1) a morphologically related word * : p < .05 (M, e.g., mariage-marier ‘wedding – to marry’), (2) an orthographically related word (O, e.g., The results show a clear pattern of a morphologi- marine-marier, ‘navy-marry’) and (3) an unrelat- cal facilitation effect (reaction times decreases ed word (U, e.g., courage-marier, ‘courage- when the prime-target relationship is morpholog- marry’). In both the HighSF condition and the ical, compared to orthographic and unrelated LowSF condition, primes were matched in num- control conditions). ber of letters (respectively 6.4 and 7 letters in A significant difference across conditions can be average) and surface frequency (respectively observed only when the target word has a lower 6.48 and 40.64 occ./million in average). Primes frequency than the primes. Statistical analysis were presented according to two frame durations showed that the critical net priming effects (dif- (SOAs), 48 and 66ms to examine the time-course ference between the reaction times for morpho- of priming. Three experimental lists were con- logical primes against orthographic and unrelated structed using a Latin square in order to present control ones) for HighSF primes - LowSF targets each target once only. was of 45* and 36*ms (respectively). Twenty-five students at the University of Tou- When looking at the LowSF-primes and HighFs louse participated in the experiment. All the par- targets the RTs differences of the net priming ticipants were native speakers of French and effects previously described, where not statisti- their average age was 26 (7.23 sd). They were all cally significant Morphological facilitation ef- right handed and had normal to corrected-to- fects seem to be larger when the frequency of the normal vision. The experiment lasted around 40 prime is higher than the frequency of the target, minutes and in exchange for their time, partici- regardless of the SOA used. pants received a 4 Giga USB key. The results are presented in Table 1. As we 2. Conclusion didn’t find any effect of the frame duration, we The results of the present study are in line with decided to present the averaged RTs. the previously found by Giraudo and Grainger Mistaken answers were not considered for the (2000), showing differential priming effects statistical analysis (2.8% of the data), neither when the surface frequency of the prime is ma- were reaction times lower than 250ms and over nipulated. The absence of a morphological prim- 1500ms (1% of the data). Cut-offs for the rest of ing effect in the High frequency M-primes/Low the data were set to 2.5 standard deviations from frequency targets contrasted with the strong sig- 129 nificant priming effects obtained with the Low Giraudo, H. & Grainger, J. (2000). Effects of Prime frequency M-primes/High frequency targets, word frequency in masked morphological and or- suggests competition effects to the detriment of thographic priming. Language and Cognitive Pro- an obligatory decomposition process. According cesses, 15, 421-444. to this view both low and high frequency targets Giraudo, H., & Grainger, J. (2001). Priming complex should have benefit from the prior presentation words: Evidence for supralexical representation of of a morphologically related word, but the results morphology. Psychonomic Bulletin & Review, revealed this was not the case. Only base targets 8(1), 127-131. having a surface frequency lower than the sur- Giraudo, H. & Voga, M. (2007). Lexeme-Based face frequency of their derivation were signifi- Model vs. Morpheme-Based Model from Psycho- cantly facilitated relative to both the orthographic linguistic Perspectives. In F. Montermini, G. Boyé, and the unrelated conditions (+45 and +36ms). and N. Hathout (Eds.), Selected Proceedings of the We interpret these data as an evidence of a com- 5th Décembrettes: Morphology in Toulouse, pp. petition process among the word forms at the 108-114. Somerville, MA: Cascadilla Proceedings Project. word level. Giraudo, H. & Voga, M. (2014). Measuring morphol- References ogy: the tip of the iceberg? A retrospective on 10 years of morphological processing, Carnets de Amenta, S. & Crepaldi, D. (2012). Morphological Grammaire, 22, xxx-xxx. processing as we know it: An analytical review of morphological effects in visual word identification. McCormick, S. F., Rastle, K. & Davis, M. H. (2009). Frontiers in Language Sciences, 3, 232. Adore-able not adorable? Orthographic underspeci- fication studied with masked repetition priming. Aronoff, M. (1994). Morphology by itself. Cam- European Journal of Cognitive Psychology, 21, bridge: MIT Press. 813-836. Booij, G. (2010). Construction Morphology. Oxford: McCormick, S. F., Brysbaert, M., & Rastle, K. Oxford University Press. (2009). Is morphological decomposition limited to Boudelaa, S., & Marslen-Wilson, W. D. (2001). Mor- low-frequency words?. The Quarterly Journal of phological units in the Arabic mental lexicon. Experimental Psychology, 62(9), 1706-1715. Cognition, 81, 65-92. Marantz, A. (2013) No escape from morphemes in Duñabeitia, J.A., Perea, M., & Carreiras, M. (2007). morphological processing, Language and Cogni- Do transposed-letter similarity effects occur at a tive Processes, 28:7, 905-916 morpheme level? Evidence for morpho- Marslen-Wilson, W.D., Ford, M., Older, L., & Zhou, orthographic decomposition. Cognition, 105, 691- X. (1996). The combinatorial lexicon: Priming der- 703. ivational affixes. In G. Cottrell (Ed.), Proceedings Duñabeitia, J.A., Laka, I., Perea, M., & Carreiras, M. of the 18th Annual Conference of the Cognitive (2009). Is Milkman a superhero like Batman? Con- Science Society, (pp. 223-227). Mahwah, NJ: Law- stituent morphological priming in compound rence Erlbaum Associates. words. The European Journal of Cognitive Psy- New B., Pallier C., Ferrand L., Matos R. (2001). Une chology, 21(4), 615 – 640. base de données lexicales du français contemporain Drews, E. & Zwitserlood, P. (1995). Morphological sur internet: LEXIQUE, L'Année Psychologique, and orthographic similarity in visual word recogni- 101, 447-462. tion. Journal of Experimental Psychology: Human Rastle, K., Davis, M.H., Marslen-Wilson, W.D., & Perception & Performance, 21, 1098-1116. Tyler, L.K. (2000). Morphological and semantic Forster, K.I., & Davis, C. (1984). Repetition priming effects in visual word recognition: A time-course and frequency attenuation in lexical access. Jour- study. Language and Cognitive Processes 15 (4-5), nal of Experimental Psychology: Learning, 507-537. Memory, and Cognition, 10, 680-698. Rastle, K., & Davis, M.H. (2003). Reading morpho- Frost, R., Deutsch, A., & Forster, K. I. (2000). De- logically complex words: some thoughts from composing morphologically complex words in a masked priming. In S. Kinoshita & S.J. Lupker nonlinear morphology. Journal of Experimental (Eds.), Masked priming: State of the art (pp. 279- Psychology: Learning, Memory and Cognition, 26, 305). New York: Psychology Press. 751-765. Rastle, K., Davis, M.H., & New B. (2004). The broth in my brother’s brothel: Morpho-orthographic 130 segmentation in visual word recognition. Psycho- nomic Bulletin and Review 11 (6), 1090-1098. Rastle, K. & Davis, M.H. (2008). Morphological de- composition based on the analysis of orthography. Language and Cognitive Processes, 23, 942-971. Seidenherg, M.S. (1987). Sublexical structures in vis- ual word recogni- tion: Access units or orthograph- ic redundancy? In M. Coltheart (Ed.), Attention and performance XII: Reading (pp. 245-263). Hills- dale, NJ: Erlbaum. Voga, M. & Grainger, J. (2004). Masked Morpholog- ical Priming with Varying Levels of Form Overlap: Evidence from Greek Verbs. Current Psychology Letters: Behaviour, Brain & Cognition, 13(2). 131