The Role of Serial Recognition in Explicating Short-Term Memory

The interrogation of verbal short-term memory processes sometimes makes use of the serial recognition task, which involves two sequential presentations of a list of items. These lists may be identical or differ by the transposition of two adjacent items, thus requiring a same/different judgement. When utilised, this task has primarily been used as a companion to the immediate serial recall task, since many of the task requirements are similar with the exception of requiring a reproduction of each list item (as in the case of recall). Together they have been used to assess contributions of long-term language knowledge to immediate short-term memory. However, literature assessing the operations involved in serial recognition alone is sparse and inconsistent. There is discord over whether or not the task involves a contribution from long-term language knowledge; if it primarily makes use of the order of items and not their identity; or if the task involves the use of only perceptual-gestural mechanisms without reference to the processing of language representations at all. Additionally, some literature suggests that set size may influence the degree to which the task utilises long-term language variables, while other studies suggest the modality of task presentation is a contributing factor. As a consequence, there is no current comprehensive framework that can accommodate this array of findings without significant modification. The present thesis sets out to assess the role of long-term language in serial recognition by testing the role of set size, task difficulty (in terms of list length), and the lexico-semantic variable (concreteness and word frequency) which is manipulated across 5 experiments. Further, it assesses the possibility that measurements typically used in serial recognition are not sensitive to discriminate between levels of lexico-semantic variables in the task. Therefore, reaction time as well as the assessment of transposition location (comparing the positions in which items are switched, as well as overall list differences) are considered. The results across all 5 experiments fit best within a language-based explanation of verbal short-term memory, and in particular the Activated Network model, which suggests item identity and order information interact with each other in such a way that order information, or tasks that primarily focus on order information (such as serial recognition), are affected by item identity information and their long-term language representations. A number of questions arise from the outcomes of these experiments, in particular the influence of articulatory suppression on word frequency in serial recognition and the implications of a replicable word frequency effect in serial recognition for a number of theoretical accounts of verbal short-term memory. How these findings can inform future research is discussed.