Phonological Short-Term Memory and New Word Learning: Evidence from Paired-Associate and Hebb Repetition Paradigms

This thesis investigated the role of phonological short-term memory (PSTM) in
the long-term learning of new phonological word-forms.. Previous studies using the
paired-associate paradigm have suggested that the learning of unfamiliar material is
mediated by PSTM (e.g. Papagno & Vallar, 1992). The first'aim was. to replicate a!1d
extend this previous work. The second aim was to determine whether the Hebb
repetition paradigm could provide an alternative method with which to investigate the
role of PSTM in new word-form learning. Seven experiments w~re conducted to
explore these aims.
Experiment 1 obtained phonological similarity effects for words and nonwords in
an immediate serial recall task, confirming that. the chosen manipulation of
phonological similarity was adequate. Experiments 2 and' 3 . adopted the
paired-associate task and replicated Papagno and Vallar' (1992), thus extending their
results to English participants and materials. Phonological similarity was shown to
selectively disrupt the learning of nonword pairs. In contrast, some evidence was found
to suggest that phonological similarity fails to affect the learning' of word pairs.
However, Experiment 3 showed that the detrimental effect of phonological similarity
was restricted to an intermediate phase of learning. These findings suggest that PSTM
mediates the learning of unfamiliar material, although the role of PSTM may change
during the course of learning; Experiments 4 and 6 adopted the Hebb repetition task
and generated patterns of results consistent with Papagno and Vallar (1992) and
Experiments 2 and 3. Phonological similarity disrupted the learning of nonword
s~quences, but not the learning of word sequences. These findings suggest that PSTM
mediates sequence learning for unfamiliar material, thereby providing initia~ evidence
that the Hebb repetition paradigm may be a possible analogue of new word-form
learning. The role qf PSTM in nonword sequence learning could not be reliably
assessed in Experiments 5 and 7 due to the absence of reliable Hebb Effects. Analyses
of between-trial learning and forgetting rates using a Markov model revealed that
phonological similarity had a negative impact on forgetting rates for nonwords in both
paired-assoCiate and. Hebb repetItion paradigms, suggesting that phonological
representations of nonwords are particularly fragile. Finally, it is proposed that the
paired-associate paradigm represents a closer analogue of new word-form learning than
the Hebb repetition paradigm as it makes use of existing lexical-semantic information.