Psycholinguistic Variables Involved in Concept Recall Following the Presentation of Sequential Features.

VIVAS, L.; GARCÍA, A.; VIVAS, J.; RICCI, L.; MACINTYRE, M.

Vancouver

Encuentro; 41º LACUS (Linguistic Association of Canada and the United States) Annual Meeting; 2014

LACUS

This paper seeks to analyze psycholinguistic variables involved in concept recall. Previous studies have explored this issue through naming-to-description tasks (e.g., Marques 2005). In this paradigm, it is used a novel design of this tasks where a set of features describing a concept is presented and participants are asked to discover the latter?s name. Typically, the features are selected by the experimenter and presented simultaneously. The present study has two key differences relative to this standard procedure: first, the attributes and their order of presentation were empirically derived from dialect-specific (rioplatense Spanish) semantic-feature-production norms; second, they were presented sequentially in accordance with the weight revealed by such norms. The goal was to assess the recall process following the same path that people use when describing a concept through features. Our sample was composed of 120 adults (18-40 years old). For each of the 40 trials, 12 features relevant to a concept were extracted from the semantic-feature-production norms (Vivas et al. 2011). We presented concepts with either high or low Age of Acquisition (AoA). For instance, we presented as low AoA concepts `bed´ and `banana´, and as high AoA concepts `cherry´ and `palm tree´. Subjects were asked to name the concepts as fast as possible. Stimuli were presented using RecSem software (García et al. 2014), which provides data on the participants? correct responses and false alarms. The effects of the characteristics of the concepts ?lexical frequency, AoA, and familiarity? and their features ?type, distinctiveness, and relevance? were analyzed in terms of two response variables: (a) the number of participants who recalled the concept name, and (b) the mode of the correct responses (i.e., the ranking of the feature triggering the correct response in most subjects). As an example of the first variable?s results for low AoA concepts 120 participants could effectively recall `bed´ and 119 `banana´, while for high AoA concepts 110 could recall `palm tree´ and 87 `cherry´. To illustrate the second variable?s results, the mode of the correct responses of `bed´ was 3, which included the features `used to sleep´, `rest´ and `made of wood´, and the mode of the concept `cherry´ was 5 and included the features `is a fruit´, `is red´, `is small´, `is sweet´ and `is tasty´. To analyze the first response variable, a classical linear model was adjusted. As the random component did not have a normal distribution (Shapiro Wilk?s p-value=.0003), a Generalized Linear Model was adjusted ?specifically, a Tweedie model (Tweedie 1984), because of the negative asymmetry of the data. The final model showed that both AoA and the presence of distinctive features had a significant effect on the number of participants who produced the concept?s name (R2=.61; p