Bilingual Sentence Production. Effects of Language Structure, Bilingual Profile, and Cognitive Load

Mikael André Albrecht will defend his thesis Bilingual sentence production. Effects of Language Structure, Bilingual Profile, and Cognitive Load on 13 December 2023.  

Summary of the thesis:

Spoken sentence production relies on a series of complex processes that impose considerable cognitive demands. That is, speakers must successfully translate a pre-lexical, conceptual message into speech fast enough to maintain conversational fluency. Speakers generally do not plan entire sentences to the point of articulation, instead planning only an initial chunk of the sentence this thoroughly while the rest of the sentence is planned incrementally as speech is ongoing for earlier portions of the sentence. This is called incremental planning while the initial chunk is known as a planning scope (Kempen & Hoenkamp, 1987; Levelt, 1989). Previous research on planning scope has shown that the initial phrase of the sentence plays a central role in determining speakers’ preferred planning scope (e.g., Allum & Wheeldon, 2007, 2009; Martin et al., 2014;   Smith & Wheeldon, 1999,2004; Zhao et al., 2014). In support of this, research has consistently found that speakers initiate speech more slowly when the initial phrase is simple (as in 1a) compared to then when the initial phrase is complex (as in 1b, e.g., Allum & Wheeldon, 2007, 2009; Martin et al., 2014). This is consistent with a phrasal scope of planning as speakers would plan a larger structure when the initial phrase is complex thus leading to longer speech onset latencies. 

1a) [The cat] goes above the house. 

1b) [The cat and the house] go up. 

Speech production is inherently more complex for bilinguals than for monolinguals. This is because bilinguals must control which language they are producing speech in. Previous research has shown that bilingual language production is nonselective meaning that bilinguals do not switch off their languages when they are not needed. Instead, both of a bilingual’s languages remain active which can in turn cause competition between them (e.g., Costa, 2005; Costa et al., 2000, Guo et al., 2011, Kroll & Stewart, 1994). Despite this complexity, bilinguals can control their languages as errors where bilinguals select an unintended language are rare (e.g., Gollan et al., 2011). Most accounts of language control posit an inhibitory mechanism where bilinguals apply inhibition to the unintended language to facilitate speech in the intended language by reducing competition (e.g., Green, 1998). 

This thesis examines bilingual language production and planning scope and whether speech planning is affected by the cognitive demands of language switching as well as interlingual overlap in syntactic structure. 64 Norwegian-English bilinguals described scenes of pairs of moving pictures in both of their languages. The movement of the pictures prompted responses that began with a simple- or complex-initial phrase (e.g., 1a and 1b respectively). Structural overlap was manipulated using noun definiteness with definite structures (e.g., “katten” – “the cat”) being less similar in English and Norwegian than indefinite ones (e.g., “en katt” – “a cat”). Participants were told to switch between their languages throughout the experiment and they also provided both subjective and objective measures of language profile in the form of a questionnaire (LEAP-Q, Marian et al., 2007) and several tests of language proficiency and fluency. 

Speech onset latencies and eye-fixations were recorded for each participant. The results showed longer eye-fixations to the second noun (“house” in the above examples) when the initial phrase was complex which is consistent with a phrasal scope of planning. The speech onset latencies showed a similar pattern of results. The results further showed that the added cognitive load of required language switching reduced speakers’ lower-level planning. Syntactic structure, meanwhile, affected production but did so independently of switching and the results suggest that this effect is not attributable to cognitive load. Overall, the results do provide evidence for a preferred phrasal scope of planning in both L1 and L2 production in proficient bilinguals and that this effect remains robust for early structural planning even when cognitive load is increased. 

Find more information about the time and place for the doctoral defense.