Native language processing of cognates in general academic vocabulary in multilinguals
Vedrana Gnjidić
Marijan Palmović, PhD
Anita Peti-Stantić, PhD
Starting point
Cognate word pairs share meaning and form in two (or more) languages. Depending on the extent of orthographic (and phonological) overlap, cognates can be either identical or form-similar. Cognates are also a form of affordances (Singleton & Aronin, 2012) which means that, if perceived, they have the potential to facilitate language learning and processing.
The cognate facilitation effect has been reported in tasks such as lexical and semantic decision, word association, or sentence reading, i.e., in studies exploring both language production and comprehension in L2/L3 or L1. Lexical decision experiments show that the facilitation effect is even stronger with triple cognates than with double cognates and present both in the native language and the weaker FL, thereby supporting the language non-selective hypothesis. L1 cognate processing has mostly been investigated with relatively proficient FL speakers due to the limited sensitivity of behavioural methods in registering the cognate facilitation effect in less proficient speakers. EEG provides a higher level of measurement sensitivity in that respect (e.g., Bice & Kroll, 2015).
Researchers predominantly use concrete and relatively frequent cognates as experimental stimuli. Identical cognates are typically chosen over form-similar cognates since the strength of the cognate facilitation effect diminishes with a decrease in orthographic similarity. However, the facilitation effect with form-similar cognates has been recorded even in the most dominant language, L1 (e.g., van Hell & Dijkstra, 2002). It seems that the facilitation effect might be stronger for form-similar cognates with differences in the suffixal part than for those with differences in the word root (e.g., Font, 2001).
Research problem and motivation
Cognates are frequent in formal registers of many languages spoken in Europe. By extension, they also make up a substantial part of general academic vocabulary (GAV). GAV words have abstract meanings and are used across disciplines. Despite the abstractness of the register, late multilinguals might process GAV form-similar cognates (such as parcijalan/partial/partiell) more easily than GAV noncognates, just as it is typically the case with more concrete words.
Based on the nonselective language activation hypothesis and the notion that multilinguals make use of the features cognates share on different representation levels bidirectionally, the main hypothesis is that learning foreign languages facilitates processing of cognates in an abstract register of the native language. Additionally, cross-linguistic awareness could intensify the activation potential of the shared features during L1 processing (e.g., Otwinowska-Kasztelanic, 2011).
Experimental evidence of facilitated processing in that register would point towards the purposefulness of a systematic, cross-linguistically integrative approach to abstract (general academic) vocabulary in native language instruction. A better comprehension of GAV words would in turn facilitate deep reading and learning in young adults.
Approach and methods
GAV cognates and noncognates used as experimental stimuli were collected from popular science literature and supplemented by entries from the Croatian Psycholinguistic Database. Experimental focus was on the influence of L2 on L1 comprehension in multilinguals and the contribution of cross-linguistic awareness to that process. Participants were recruited among university students of various profiles. Nonspeeded lexical decision was used to assess their FL proficiency in English and German (Dialang), a standardized test to assess their knowledge of internationalisms in Croatian (VerT), and a researcher-designed battery of tasks to assess their crosslinguistic awareness through active use of internationalisms in written tasks (synonyms, translation).
A semantic categorisation task was created in E-prime 2.0., containing 38 cognate (e.g., modificirati ‘modify’), 38 noncognate (e.g., obrazložiti ‘justify’) GAV verbs, and 39 concrete verbs related to food preparation (e.g., zapeći ‘bake’). Participants categorised the items as academic vocabulary or food preparation while a 32-channel EEG was recorded with the BP software. The groups of GAV cognates and noncognates did not differ statistically significantly in terms of abstractness ratings, corpus frequency, or length (mm). The task was piloted online (Psytoolkit; Stoet, 2010, 2017) with 28 participants (M [age] = 22.7; self-reported L2 proficiency – B2; L3 – A2) to verify that the stimuli would be categorized unambiguously. We expected that GAV cognates would be categorized faster than GAV noncognates, and that the fastest categorization (shortest reaction times) would occur with food-related verbs.
Results
The average percentage of incorrectly categorized verbs per participant in the pilot experiment was 2,9%. Most of those verbs were categorized incorrectly by only one or two participants. There were three verbs categorized incorrectly by more than three participants which were therefore excluded from the final stimulus selection for the ERP experiment.
The average reaction time to cognates, noncognates, and verbs related to food preparation was 715.91 (SD = 98.19), 745.60 (SD = 96.92), and 716.61 (SD = 79.70) milliseconds, respectively. Repeated measures ANOVA revealed a main effect of word group (F [2, 54] = 7.55, p < .01, ηp² = 0.22). Post hoc tests using the Bonferroni correction revealed that the differences in RTs to cognates and noncognates were statistically significant (p < .01), as well as the differences in RTs to noncognates and verbs related to food preparation (p < .01). The difference in RTs to GAV cognates and verbs related to food preparation was not statistically significant (p = 1.0).
Data collection for the ERP experiment is currently ongoing and preliminary results will be presented at the conference.
Discussion and conclusions
The results of the online semantic categorisation task revealed the presence of the cognate facilitation effect in an abstract register with L1 form-similar cognates. However, it is possible to claim that internationalisms are perceived as more characteristic of (and therefore predictable in) the academic register – i.e., the categorising advantage might have arisen from recognizing morphological cues revealing cognates as internationalisms (suffixes, prefixes, letter clusters). Consequently, since such cues are not available for noncognate GAV verbs, they might require a deeper level of processing (a more detailed access to meaning) in order to be categorized. We expect that the pattern of RT results from the pilot experiment will be replicated in the ERP experiment.
Event related potentials enable comparing the processing of cognates and noncognates before the categorisation decision has been made, i.e., during lexical access. The mean amplitude of the N400 component is expected to be less negative for GAV cognates than for noncognates, reflecting facilitated processing of cognates in an abstract register of the native language.
The results of this experiment are expected to contribute to the understanding of the cognate facilitation effect with form-similar cognates during native language processing in an abstract register and the understanding of the connection between crosslinguistic awareness and cognate processing in that register. Participants included in this study represent the most prevalent group of multilinguals in Croatia, unbalanced FL learners of two languages who learn their FLs in an institutional context.
References:
Aronin, L., & Singleton, D. (2012). Affordances theory in multilingualism studies. Studies in Second Language Learning and Teaching, 2(3), 311-331. https://doi.org/10.14746/ssllt.2012.2.3.3
Bice, K., & Kroll, J. F. (2015). Native language change during early stages of second language learning. Neuroreport, 26(16), 966–971. https://doi.org/10.1097/WNR.0000000000000453
Font, N. (2001). Rôle de la langue dans l’accès au lexique chez les bilingues: Influence de la proximité orthographique et sémantique interlangue sur la reconnaissance visuelle de mots (Unpublished Ph.D. dissertation) Montpellier, France: Université Paul Valery
van Hell, J. G., & Dijkstra, T. (2002). Foreign language knowledge can influence native language performance in exclusively native contexts. Psychonomic Bulletin & Review, 9(4), 780–789. https://doi.org/10.3758/BF03196335
Otwinowska-Kasztelanic, A., (2011). Chapter 1: Awareness and affordances: multilinguals versus bilinguals and their perceptions of cognates. In G. De Angelis & J.-M. Dewaele (Eds.), New Trends in Crosslinguistic Influence and Multilingualism Research, 1–18. https://bit.ly/2STfqvF
Stoet, G. (2010). PsyToolkit – A software package for programming psychological experiments using Linux. Behavior Research Methods, 42(4), 1096-1104.
Stoet, G. (2017). PsyToolkit: A novel web-based method for running online questionnaires and reaction-time experiments. Teaching of Psychology, 44(1), 24-31.
205 total views, 1 views today
This post is also available in: 
Hrvatski (Croatian)
