Let’s say you were asked to recall the word list, “cheese, steak, wine,”. Because you know all these words, your brain would likely process both the phonological (speech sound) form and the meaning of the word. But how would this unfold? Would you first activate the phonological representation of each and then you draw on semantic knowledge (meaning-based information)? Or would both types of information be activated at the same time?
Verbal short-term memory has often been viewed as operating in separate stores. That is, the phonological form is held in mind and rehearsed using what is called, ‘the phonological loop’. At the point of recall, if the phonological trace has been degraded, then semantic information stored in long-term memory is activated to help with reconstruction of the word, a process referred to as redintegration. According to this view, semantic knowledge held in long-term memory is activated and transferred to short-term memory at the point of recall, and only after phonological information is processed. Recently, this 2-step process has been called into question.
According to the language-based model, verbal short-term memory reflects the direct (and automatic) activation of your phonological, lexical, and semantic representations for each word. There is considerable overlap between verbal short-term memory and language processing from the moment of encounter, and not just at the point of recall. If an integrated network activates all linguistic knowledge related to a word on encounter, then semantic effects should be observed to influence one of the most well-described short-term memory findings: serial recall, the immediate recall of items lists in correct order.
Poirier et al. (2015) completed a series of studies manipulating the semantic relationship of words in the list to a particular target word. In Experiment 1, participants were presented with a 6-item word list wherein the first three items were or were not semantically related to the target word (item 5). Recall of the stimuli in Experiment 1 could have been influenced by a potential grouping strategy, that is, the participants could have grouped semantically related words in the condition where the words were related. Therefore, in Experiment 2, all items were from the same semantic category and were either studied in silence (2a) or under articulatory suppression (2b). Participants were asked to recall the word list in the order presented across all experiments.
Across the three studies, the target word migrated towards earlier positions at the point of recall when semantically related words were present, and this was not the result of a grouping strategy. The authors suggest that the migration occurred as a result of coding serial position (i.e., order information) as part of the semantic network that supports language representations. This provides a new perspective on the processes involved in verbal short-term memory: language processing in short-term memory is closely tied to semantic networks in long-term memory, and thereby, short-term and long-term memory may be more integrated than they are distinct. This may have important implications for word and language learning such as thinking about how phonological rehearsal and semantic processing should be integrated during the learning phase, rather than learned discretely, in order to optimize on retention.
