Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Words that we use in daily life are stored in memory and more specifically in what is referred to as the mental lexicon. The structure and organization of the human mental lexicon is complex, and multiple semantic models of how our vocabulary is organized have been proposed. A considerable amount of research has investigated these semantic models, either by focusing on one single model or by comparing two different models using particular semantic tasks. The latter approach has also been implemented in this study which compared two well-known semantic models each proposing a different structure of the human mental lexicon, i.e. a model based on word associations and a model based on text corpora. The association-based model states that two words are closely related when they are located close to each other in the semantic network (and thus share many associations), while the corpus-based model proclaims that two words are closely related when they share a similar word co-occurrence pattern in text corpora. The research hypothesis of this study is that the association model approximates the structure of the human mental lexicon closer than the corpus model. To test this hypothesis, two different experiments were performed, i.e. a hangman game and a false memory experiment. In the hangman game experiment, participants had to guess critical words based on hints sequentially presented one by one. The hints represented words that are related to the critical word, either according to the association model or to the corpus model. The false memory experiment employed the Deese-Roediger-McDermot (DRM) paradigm in which the participants’ rate of false recognition of critical words was compared following a study phase during which a list of related words was presented selected either based on the association model or on the corpus model. The hangman experiment revealed that participants needed a lower number of hints to guess the critical word when hints were selected based on the association model. The false memory experiment showed that a larger number of critical words were ’falsely’ (since not having been presented) recognized when the list words shown during the study phase were selected from the association model. In conclusion, the participants found the task easier (hangman experiment) and were more easily tricked into false recognitions (false memory experiment) when based on the association model as compared to the corpus model, providing support to our research hypothesis that the word association model is better in line with the structure of the human mental lexicon. However, it should be taken into consideration that this conclusion is limited to the nature of semantic tasks as assessed in this study.