The development of long-term lexical representations through Hebb repetition learning

Abstract

The Hebb repetition effect and naturalistic word-form learning AbstractThis study clarifies the involvement of short-and long-term memory in novel word-form learning, using the Hebb repetition paradigm. In Experiment 1, participants recalled sequences of visually presented syllables (e.g., la-va-bu-sa-fa-ra-re-si-di), with one particular (Hebb) sequence repeated on every third trial. Crucially, these Hebb sequences contained three orthographic nonword neighbours of existing Dutch base-words (e.g., lavabu -lavabo [kitchen sink]). Twenty-four hours later, the same participants performed two auditory lexicalization tests involving the actual Dutch base-words (e.g., lavabo, safari, residu). Both tests yielded slower reaction times for these Dutch base-words compared with matched control words, which reflects lexical competition between the base-words and the Hebb sequences, therefore demonstrating lexical engagement of the Hebb sequences. In Experiment 2, we subsequently used the Hebb paradigm as an analogue of word-form learning, in order to investigate whether the creation of novel lexical memories requires sleep. Whereas earlier findings indicate that overnight sleep plays a crucial role in lexical consolidation, the current results show that Hebb learning of phonological sequences creates novel word-forms representations in the mental lexicon by the mere passage of time, with sleep playing no necessary role.

Linking memory and language: Evidence for a serial-order learning impairment in dyslexia

“…It extends the previous examination of HRL in adults with dyslexia (Szmalec et al, 2011) by including not only an initial Hebb-learning session with a much larger number of Hebb repetitions (up to 20 in the current study vs. 12 in Szmalec et al, 2012) but also relearning on the subsequent day and one month after initial learning. This allows us to Serial-order learning in dyslexia estimate the retention of the learned Hebb sequences over time.…”

“…With this aim, participants again learned Hebb sequences (e.g., lava-bu-sa-fa-ra-re-si-di), containing lexical competitors (e.g., lavabu, safara, residi) of Inherent to the use of the lexical competition approach is the requirement that Hebb sequences closely resemble known words represented in the mental lexicon. Importantly, the earlier studies using this lexical-competitor approach (Szmalec et al, 2012) have demonstrated that this procedure yields Hebb learning curves (for normal readers) comparable to standard verbal Hebb learning curves (Szmalec et al, 2009(Szmalec et al, , 2011(Szmalec et al, , 2012, suggesting that the learning of syllable sequences derived from existing words does not seem to rely on strong support from these words. This might be due to the fact that the Hebb procedure exposes the participant to individual syllables, presented one by one, while the gradual and implicit grouping of those syllables into pseudoword-forms is only the outcome of the Hebb-learning process.…”

“…HRL of a syllable sequence like "lo fo du" is therefore assumed to be functionally equivalent to acquiring the novel word-form "LOFODU", similar to the way in which children learn new words by picking up statistical regularities from the verbal input in their environment (e.g., Saffran et al, 1996). Experimental evidence for the hypothesis that HRL mimics naturalistic word-form acquisition was provided by Szmalec and colleagues (Szmalec, Duyck, Vandierendonck, Barberá-Mata, & Page, 2009;Szmalec, Page, & Duyck, 2012). In these experiments, that included only normal readers, participants typically had to recall nonsense sequences of nine visually presented consonant-vowel syllables (CVs), with each sequence grouped by short pauses into three three-CV groups (e.g., "fi ke da -sa mo pu -vo ti zu").…”

“…Previous studies in normal readers have shown measurable savings from earlier HRL in an unannounced test three months after learning (Page & Norris, 2008), supporting the claim that HRL is indeed long-term learning. In the case of verbal HRL, it is assumed that the learned sequence creates novel entries in the mental lexicon (Szmalec et al, 2009(Szmalec et al, , 2012; see above). Szmalec et al (2011) explicitly characterized their serial-order account as a 'learning account': the dyslexic disadvantage is assumed to exist at the stage of the long-term learning of serial-order information (rather than solely at the stage of short-term representation of this information, although data suggest such a short-term deficit toosee below).…”

“…In a Hebb learning block, sequences of nine consonant-vowel syllables (CVs) were presented visually for immediate serial recall. One particular sequence, the Hebb sequence, was "repeated" on every third trial (in a manner similar to Szmalec et al, 2011Szmalec et al, , 2012, and as described below). On the other trials, the filler trials, the order of the syllables was randomized.…”