Free Word Association Case Study

4.2 Productive Word Associations Task
The most widely-used tool for collecting information about the associative links between lexical items and the structure of language learners’ mental lexicon is a free Word Associations Task. Invented by Galton (1883), this technique was initially used in psychology and psychiatry as means to access people’s unconscious motivations and to expose mental disorders. Kent and Rosanoff (1910) conducted the first English-based large-scale study, which was carried out with 1000 “(supposedly) normal” subjects of both sexes and diverse educational and socio-economic backgrounds. The main purpose of the study was to determine the nature of the variation of different forms of “pathological reactions” from “normal

Nelson, McEvoy and Dennis (2000) advocate for the use of traditional single-response procedure, also known as a discrete task. The major argument for the choice of a discrete tasks is that they tend to reliably elicit word’s strongest associates. Additionally, the findings of their study suggested that in general second associative responses are only weakly related to the stimuli words. The chaining analysis (response-to response correlations) performed as part of the study, however, revealed that only between 8-20% of second associates could be influenced by the first response items (Nelson, McEvoy, & Dennis, 2000:892), providing correlational evidence that refutes the widely-held assumption that second associations are plagued with chaining. Assessing native-likeness of L2 learners’ associative networks, Schmitt (1998(b)) challenges the assumption that language learners will always respond with the strongest association first and further supports the use of continuous association approach by arguing that eliciting multiple responses to each of the stimulus words “would supply a more convincing illustration that the stimulus word is incorporated into a subject's lexicon in a way similar to a native speaker.” (Schmitt, 1998(b): 391). This thesis research methodology adopts the multiple-response paradigm. However, to combat response chaining and account for the large number of

This distinction is mainly based on word class. Paradigmatic associations are those in which stimulus word and response word belong to the same word class (e.g., noun– noun); syntagmatic associations have a word class other than the stimulus word (e.g., noun–verb). Often the substitutability of paradigmatic words is contrasted with syntagmatic words occurring together with stimulus words in discourse (Cronin, 2002). The paradigmatic-syntagmatic approach to categorising associations has received criticism since the 1970s. Petrey (1977) labels the categories paradigmatic and syntagmatic inadequate for describing changes with age in word associations, “because [they] can designate nothing but syntactic properties” (1977: 69). She prefers the terms episodic and semantic (see section 2.1.2). When given the word add, children gave the episodic responses flour (syntagmatic) and cook (paradigmatic); both are related in the same way to add (they come from the same situation) but would be misclassified as opposites by Syntagmatic-Paradigmatic terminology. She adds that, “syntagmatic responses properly so-called are highly problematic because either semantic or episodic storage could produce them” (1977: 66). Nissen and Henriksen (2006) criticise the syntagmatic-paradigmatic shift as they find a surprisingly strong native speaker preference for syntagmatic