Age-related Differences in Interlingual Priming: A Behavioural and Electrophysiological Investigation

ABSTRACT

Reaction time (RT) and the N400 ERP component were measured to examine age-related differences in bilingual language processing. Although young bilinguals appear to access both languages simultaneously (i.e., non-selective access), little is known about language selection in older adults. The effect of language context on language selectivity was investigated using interlingual homographs (IH; i.e., words with identical orthography but distinct semantic features in two languages, e.g., coin meaning ‘corner’ in French and ‘money’ in English). Younger and older French/English bilinguals were presented with triplets of letter strings comprised of a language context cue, an IH, and a target word, in a lexical decision semantic priming task. RT and ERP results support non-selective language access in young adults; however, the older bilinguals used the language context cue to bias their reading of the IH. Results are discussed in terms of age-related changes in language processing and context use in bilinguals.

Keywords:

• Aging
• Bilingualism
• Context
• Event-related brain potentials
• N400

ACKNOWLEDGEMENTS

The authors would like to thank the members of the Cognitive Psychophysiology Laboratory for their help with stimuli creation, data collection and processing, specifically, T.L. Choy, S. Gagnon, D. Cateni and S. Einagel. Thank you to Drs D. Poulin-Dubois, N. Segalowitz, V. Taler, and A. Winneke for helpful comments throughout this project. Preliminary versions of this work were presented at the IXth International Conference on Cognitive Neuroscience, Havana, Cuba, September 2005 and the 15th annual meeting of the Cognitive Neuroscience Society, San Francisco, CA, April 2008. This work was supported by a Canadian Institutes for Health Research New Investigator Award and grant #203751 from the Natural Sciences and Engineering Research Council of Canada to N.A. Phillips.

Notes

¹Note that the unambiguous language context cue was designed only to bias the reading of subsequent words in the triplet; it did not bear any semantic relationship to the IH or the target.

²One older adult had some knowledge of Italian but was not fluent. A second older adult was learning German at the time of testing.

³It should be noted that due to the distinct meaning of the IHs in each language there were commonly differences in word class. For example, the letter string loin is a noun in English, but is an adverb in French meaning ‘far’. As a result, the target words also varied in word class. It has been suggested that that there are differences in word-class processing, demonstrating different ERP effects in response to verbs and nouns (Federmeier, Segal, Lombrozo, & Kutas, 2000; Khader, Scherag, Streb, and Rösler, 2003). These studies only examined nouns and verbs, and did so using a sentence context; therefore, it is uncertain how word class may affect the present results. However, the majority of the IH (86%) were either nouns or verbs, the proportion of which did not vary systematically across conditions.

⁴We do not believe the absence of this effect was due to low statistical power; the observed power for this interaction was .66.

⁵Recently van de Meerendonk, Kolk, Vissers, and Chwilla (2008) have found that a late positive shift (i.e., P600) is related to conflict between expected and encountered linguistic events. By this account a large P600 is elicited when the conflict is strong enough to trigger reanalysis. In the present experiment, it is conceivable that some conditions may trigger this reanalysis; however, in the present experiment there is semantic conflict (i.e., unrelated target words) as well as language conflict (i.e., inconsistent conditions), therefore predictions involving the P600 are not straightforward. In spite of this, the analysis that we conducted did not provide evidence for a late positive effect.

⁶Alternatively, it is possible that rather than the inconsistent L2 language cue causing interference, there was facilitation of the processing of the IH in the presence of the consistent L1 language cue. There was not a neutral baseline against which to compare these alternative accounts; however, the electrophysiological findings would favour the former explanation. Given that the target was in L1 in both of these conditions, the observed effect can only be the result of the language context manipulation.

⁷One limitation to this interpretation is that it is based on the failure to find a significant N400 priming effect for the inconsistent conditions. It is possible that we have somewhat limited power due to the sample size. Although our sample size was modest, we note that it is consistent with other studies that have observed significant effects in older adults (e.g., Faustmann, Murdoch, Finnigan, & Copland, 2007; Neumann, Obler, Gomes, & Shafer, 2009).