Grammatical constraints on language switching: Language control is not just executive control

Abstract

The current study investigated the roles of grammaticality and executive control on bilingual language selection by examining production speed and failures of language control, or intrusion errors (e.g., saying el instead of the), in young and aging bilinguals. Production of mixed-language connected speech was elicited by asking Spanish-English bilinguals to read aloud paragraphs that had mostly grammatical (conforming to naturally occurring constraints) or mostly ungrammatical (haphazard mixing) language switches, and low or high switching rate. Mixed-language speech was slower and less accurate when switch-rate was high, but especially (for speed) or only (for intrusion errors) if switches were also ungrammatical. Executive function ability (measured with a variety of tasks in young bilinguals in Experiment 1, and aging bilinguals in Experiment 2), slowed production and increased intrusion rate in a generalized fashion, but with little or no interaction with grammaticality. Aging effects appeared to reflect reduced monitoring ability (evidenced by a lower rate of self-corrected intrusions). These results demonstrate robust effects of grammatical encoding on language selection, and imply that executive control influences bilingual language production only after sentence planning and lexical selection.

Bilingual language intrusions and other speech errors in Alzheimer’s disease

“…Different predictions however, would follow for intrusion errors, which produced significantly reversed-dominance effects in previous studies such that bilinguals replaced dominant-language targets with non-dominant-language translations more often than the reverse (i.e., more often than they replaced non-dominant-language targets with dominant-language translations; Gollan et al, 2014; Gollan & Goldrick, 2016, in press). For example, an English-dominant bilingual would be more likely to replace the English word reason with its Spanish equivalent, razón , when reading aloud a sentence written mostly in Spanish (e.g., Es por esa reason que digo que la leyenda de La Llorona es verdad ) than he would be to replace the Spanish razón with reason , when reading the English equivalent sentence (i.e., It is because of that razón that I say that the legend of the Weeping Woman is true ).…”

“…For example, an English-dominant bilingual would be more likely to replace the English word reason with its Spanish equivalent, razón , when reading aloud a sentence written mostly in Spanish (e.g., Es por esa reason que digo que la leyenda de La Llorona es verdad ) than he would be to replace the Spanish razón with reason , when reading the English equivalent sentence (i.e., It is because of that razón that I say that the legend of the Weeping Woman is true ). Moreover, reversed-dominance effects though slightly smaller, were not significantly smaller in older than in young bilinguals, and neither young nor older bilinguals exhibited more intrusions in the non-dominant than the dominant language (Gollan & Goldrick, 2016). …”

“…In addition, bilinguals reading aloud mixed-language paragraphs sometimes produced intrusion errors in which they spontaneously produced translations of written target words in their speech (e.g., saying pero instead of but ). Aging bilinguals produced more intrusions than proficiency-matched young bilinguals in both the read aloud task (Gollan & Goldrick, 2016) and in verbal fluency tasks (Gollan, Sandoval, & Salmon, 2011). Additionally, bilinguals with AD produced more intrusion errors with disease progression in both picture-naming and word translation (Costa, et al, 2012).…”

“…In recent work we used a paragraph reading task (Kolers, 1966) to demonstrate that reading aloud elicits connected speech in a manner that engages the language production system and leads speakers to produce speech errors that resemble errors produced in spontaneous conversation (Gollan, Schotter, Gomez, Murillo, & Rayner, 2014; Gollan & Goldrick, 2016, in press). For example, bilinguals produced significantly more errors when reading aloud in their non-dominant language than when reading in their dominant language (e.g., producing kept instead of keep , or turn instead of turning ).…”

“…The finding of intact reversed dominance effects in aging bilinguals with clear impairments in independent measures of nonlinguistic executive control abilities could further imply that language control mechanisms are relatively modular and specialized (Gollan & Goldrick, in press). An alternative possibility is that relatively little control is applied within the linguistic system to achieve language selection – so that even though impaired in normal aging – sufficient control abilities remained intact in aging, which therefore does not elicit impaired dominance effects (Gollan & Goldrick, 2016). Thus, bilinguals with AD will provide a stronger test of the hypothesis that general mechanisms of executive control underlie reversed dominance effects; executive control may be sufficiently impaired in bilinguals with AD to reveal weaker or absent dominance reversal (relative to cognitively intact aging bilinguals).…”