Early studies of the bilingual lexicon

Since the 1960s, much research has addressed the issue of how the two languages of a bilingual are cognitively represented. This early research, mostly during the 1960s and 1970s, posed the question as one versus two separate stores. Not all, but a large part of these studies focused specifically on the organization of the lexicon and these will be concerned here. Some researchers suggested entirely separate lexicons (Kolers, 1963). Much of this research was influenced by the three types of bilingualism originally proposed by Weinreich (1953). In what he labeled a coordinate structure, both meaning and form are completely language specific. In a compound structure, meaning is shared and form is language specific. In a subordinate structure, the form representation in the weaker language is attached to the form in the dominant language, and not directly to the meaning component. With time, the coordinate structure has received less support, because it is considered unlikely that a word meaning is so divergent in two languages that they require completely separate representations. This leaves us with three components; one for shared conceptual representation and two for language dependent form representations. A large body of the subsequent research addressed the issue of how these three components relate to each other. Potter, So, von Eckhardt and Feldman (1984) investigated how form and meaning of a second language are connected and reintroduced two of Weinreich (1953) models under new names. In one model, L2 forms are associatively linked to the first language (L1) form representation. They call this the word association hypothesis (similar to Weinreich’s subordinate structure). In a second proposal, second language (L2) word forms are linked directly to the conceptual level of representation, regardless of level of proficiency. This was called the concept-mediation hypothesis (similar to Weinreich’s compound structure). Potter et al. (ibid.) added a developmental perspective by assuming that lower levels of L2 proficiency involves processing by means of word association links, whereas higher levels of L2 proficiency second language word forms are processed in terms of conceptmediation. Kroll and Sholl (1992; see also Kroll & Stewart, 1994) introduced a model which included both types of processing as co-existing within the same speaker. In their model, the revised hierarchical model, the links