Priming asymmetry persists in German-English-French trilinguals: the sense model modified for the trilingual mental lexicon

ABSTRACT

The present study examined within – and cross-language priming patterns among German-English-French trilinguals in order to explore the lexico-semantic representation of L3 in relation to L1 and L2. The trilinguals participated in three lexical decision tasks within the masked translation priming paradigm. The results showed significant within-language repetition priming effects in all three languages, significant translation priming effects for L2-L1, L1-L3, and L2-L3, but no significant priming for L1-L2, L3-L1 or L3-L2. Our findings demonstrate that translation priming asymmetry persists in trilinguals and that the weakest L3 is integrated into both L1 and L2 conceptually (i.e. three languages have a commonly shared conceptual representation). In addition, our results showed a language dominance shift over lexical development between L1 and L2. We argue for a modified Sense Model as the best fit to explain the cognitive architecture of the trilingual lexicon.

KEYWORDS:

• Translation priming
• multilingualism
• priming asymmetry
• language proficiency
• language dominance
• trilingual lexicon

Acknowledgements

This paper is based on data from Christina Steinman's dissertation under the supervision of Xin Wang at Oxford University. We thank Omidreza Ghasemi for his help in data analysis, as well as Zoltan Denis and Serje Robidoux for the helpful discussions and comments on our Bayesian analysis.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 We used the “mixed” function in the afex package in our analyses, which generates different outputs in the predictor label level suffix from those when using “lmer”. This is due to the fact that “mixed” by default uses sum-to-zero contrasts whereas base R uses treatment contrasts (in which the first factor level is the baseline, as in “lmer”). Sum-to-zero contrasts are a better choice here to test interactions of categorical variables and Type III sums-of-squares tests of fixed effect terms. As a result, the interpretations of the fixed effect estimates are different: “lmer” gives the treatment-contrasts and reflects the difference of this factor level from the reference level; while “mixed” gives the sum-to-zero contrasts and reflects the difference from the grand mean.

Additional information

Funding

The paper is funded by John Fell Fund, University of Oxford OUP (EPD06550) from University of Oxford and Australia Research Council (DP 210102789) awarded to Xin Wang.