| dc.description.abstract | Studies investigating bilingual word production find that a switch in language almost universally results in a substantial performance switch cost. This thesis explores some of the recent claims that language switch costs may be smaller or even absent in particular circumstances or for particular bilingual tasks.
It has recently been questioned whether language switch costs in bilingual production may stem from the widespread use of artificial language cues in these studies: cues such as colour patches, not transparently associated with a language nor familiar as a cue for using that language, might increase demands on intentional control and inflate switch costs. Chapter 2 addresses the question: Do more naturalistic (ecologically valid) cues reduce or eliminate the language switch cost? In a two-session study, we examined the efficacy of four auditory and four visual cues, presented at two cue-stimulus intervals, in signifying the language required, in French, German and Spanish bilinguals performing a picture naming task. We compared (separately for visual and auditory cues) performance with “naturalistic” language cues – words spoken in the target language (e.g., “English” or “hello”) and faces of friends or unknown people – with non-naturalistic cues – flags and fragments of tunes.
In bilingual reading, the language for selection is generally unambiguously indicated by orthographic input (“dog” is an English but not French word). For languages that share a writing system, word forms can be highly similar or identical across the two language (e.g., interlingual homographs: “four”= 4 in English/“oven” in French). Chapter 3 presents three experiments in which we examine what kind of language control is exerted when cross-language conflict arises from language-ambiguous words. In Experiment 1, French-English bilinguals performed a semantic categorisation task. The need/opportunity for control was manipulated by (a) interspersing (or not) interlingual homographs on 20% of trials, and (b) cueing (or not) the to-be-read language. Performance on language-specific (non-homograph) words was compared between-participants, when (1) accompanied by homographs and cues (2) language cued/no homographs, and (3) no cues/no homographs. In Experiments 2 & 3, the mechanisms underlying this language control were explored by (1) including a manipulation of preparation interval (150 or
3
1000ms), and (2) using EEG to identify ERPs associated with language preparation/control.
The experiments presented in Chapter 4 further explore the nature of top-down language selection in bilingual visual comprehension by asking whether top-down control operates over early, orthographic word processing. In Experiment 1, Cantonese-English bilinguals categorised a target word as living or non-living when presented with a distractor item from the opposing script. Categorisation of the distractor item was either congruent, incongruent, or neutral (non-word) with categorisation of the target. The available time to prepare to read in the target script was manipulated via a variable cue-stimulus interval (50, 800, or 1200ms) where the cue specified the language/script of the upcoming target item. Experiment 2 asked Mandarin-English bilinguals to perform the same task except that the two-word stimuli from Experiment 1 reflected a minority (one-third) of trials. On the remaining trials, participants were presented with the single target word for categorisation. Analysis of single word trials distinguished between any reductions in switch cost or congruency effect stemming from preparation aiding location of the target and/or reading the target for meaning. | en_GB |
