2018 Friday Poster 6448

Friday, November 2, 2018 | Poster Session I, Metcalf Small | 3pm

You win some and you lose some: Processing and judgment of scalar implicatures within attrition contexts
D. Miller, J. Rothman

Spanish has two scalar quantifiers that roughly correspond to some in English; however, each distributes uniquely depending on context. Due to semantic features constraining its meaning, algunos elicits a quantity implicature and refers only to subsets, whereas unos, like English, can refer to subsets and whole sets alike (e.g., Gutierrez-Rexach, 2001). We tested 30 L1 Spanish-L2 English bilinguals who moved to the UK in adulthood and 30 monolingual speakers of peninsular Spanish to see what happens to semantic mappings of these scalar terms in an immersion setting of L2 English.

The first experiment measured judgments using a picture-sentence verification design while simultaneously recording ERP data. Participants saw a picture with either five characters performing an activity or with a subset of characters performing one activity and the rest another. Pictures were followed by contextually appropriate or inappropriate sentences displayed on a computer monitor, after which acceptability judgments were made. The monolinguals consistently rejected algunos in whole contexts but accepted unos differentially based on context. This latter result formed a logical/pragmatic division, which is that for some individuals unos was ‘possibly all’ and for others it was ‘only some’, suggesting that unos may implicate beyond its argued semantics. Contrastively, the bilinguals exhibited an over-acceptance of algunos in whole contexts but accepted unos without preference for one context over the other, displaying no logical/pragmatic division. The monolinguals also showed a sustained N400 for algunos in whole contexts and a flat N400 for unos in whole contexts, but the offline responses had no bearing on the overall ERP analysis. The bilinguals, in contrast, showed no N400 for algunos whole contexts. The unos comparisons did reveal a flat, but insignificant, N400-like effect.

The second experiment, a non-binary interpretation task, allowed participants to assess each quantifier by associating it with a number within a defined reference set. The predefined sets included from four to six referents to determine if increased set size modulated judgments (Degen & Tanenhaus, 2011, 2015). The non-binary data were coded into whole and partitive responses to be amenable to a binary regression. Monolinguals judged algunos as partitive and unos as partitive or whole. The same pragmatic and logical responders were consistent in their judgments across tasks. The bilinguals, however, over-accepted algunos in whole contexts, though this effect decreased as a function of increased set size, and they also accepted unos in either context. Thus, for the bilinguals algunos and unos came to be indistinguishable from each other, reflecting the potential manings of English some.

These data together suggest that L1 attrition affects implicit processing as well as explicit reasoning of scalar quantifiers, and that bilingual performance can be mitigated by methodological design. As the data suggest, L1 semantic features can be remapped to pattern after L2 feature bundles after prolonged exposure to a competing L2 and decreased L1 input. We offer a formal analysis for the present case based on the Feature Reassembly Hypothesis (Lardiere, 2009) which we suggest could be fruitful for attrition research more generally (Schmid & Köpke, 2017).