Validating a mental lexicon derived from a large-scale word association project
Semantics is at the core of any language based activity in our society but arguably one of the most complex aspects to study. To gain a deeper knowledge in how word meaning is processed and stored in human memory, a number of lexico-semantic models have been proposed. These models have been proven to be instrumental in a number of lines of research, ranging from purely theoretical questions, such as atypical word processing, the structure and acquisition of the mental lexicon in children, to more pragmatic issues, including second language learning, text processing and expert system development.
In studying the meaning of words, several approaches have been undertaken up to this day. One of these uses featural decomposition of words as a means to capture the representation of their. Here, participants are asked to give a number of features of exemplars from different categories. These methods typically perform well to capture meaning for a small subset of the lexicon consisting of concrete nouns, but do not scale in a straightforward way to other words such as abstract words, verbs or adjectives. In the past two decades, however, other, more large-scale approaches have been developed, like linguistically inspired representations (e.g., Wordnet) and models based on word co-occurrences in text corpora (e.g., Latent Semantic Analysis or LSA; and Hyperspace Analogue to Language or HAL). These approaches have been successful in deriving how words in a large subset of the lexicon are related. In addition, the same models have recently also been used in predicting diverse properties, such as concreteness or valence.
During my PhD, I want to study the potential of a large-scale word association network as an alternative semantic basis. Word associations have been used for over decades because they grant insight into the structure and processes of the human mind but this large-scale network has only been available for a couple of years.
I want to study differences in semantic structure, both interindividual in a normal population and in psychiatric populations using network representations of the mental lexicon derived from word associations. One of the key properties of this proposal is that the emphasis will be on affect, in line with recent proposals that stress the important contribution affect makes to word meaning, especially in the case of abstract words. Semantic processing is often distorted in psychiatric patients, together with distorted affect, and this is expressed in disruptive affective word connotation in a number of psychiatric patients, like severely depressed patients, bipolar patients, and schizophrenics. For these reasons, I want to focus on how this distortion in affect is present in the language of these groups, together with non-affective irregularities in semantics (like is the case in Alzheimer's disease). To study this distortion in language, mostly verbal fluency tasks, priming tasks, and word association tasks are used. With the availability of this large-scale association network, new opportunities arise to study this.