Research paperEvent-related potentials in an associative word pair learning paradigm
Introduction
Investigation in the electrophysiology of semantic processing began with a famous study of Kutas and Hillyard (1980) who showed that a negative deflection peaking around 400 ms post-stimulus strongly distinguished between the last words in sentences that were semantically congruent versus incongruent with the context of the sentence (e.g., “I drink my coffee with cream and sugar” versus “I drink my coffee with cream and dog”). This deflection, labeled N400, was shown to be a part of the normal brain response to words and other meaningful (or potentially meaningful) stimuli but it is typically attenuated when the word is strongly prepared by its context (Kutas & Federmeier, 2011).
In addition to the above-depicted sentence paradigm with congruent versus incongruent endings, similar results were obtained in experiments with word pairs (Holcomb, 1988). In both paradigms, the response to a word is regarded as a function of the background, or the context, on which the word appears (Weingarten et al., 2016). In the word pair paradigm, the context for the second word in a pair (referred to as “target”) is built by the first word (“prime”). In the sentence paradigm, the last word is primed by the context of the sentence. N400 amplitude is typically larger for targets that are incongruent with their primes than for congruent targets and inversely related to the degree of semantic expectancy. The similar paradigm can be extended to an even larger context of the whole conversation (e.g., van Berkum, Hagoort, & Brown, 1999): N400 amplitude is increased in response to targets incongruent with the idea of the conversation, even if they are completely congruent with the immediately preceding sentence. Generally, the modulation of the N400 amplitude by the preceding context is referred to as the semantic priming effect (Brualla, Romero, Serrano, & Valdizán, 1998).
In addition to the semantic relationship, an association between two words can emerge due to their spatial and/or temporal co-occurrence in the real world and/or in a language (McRae, Khalkhali, & Hare, 2012). N400 effect is sensitive to how often two concepts co-occurred within an individual's past experience. An example of such co-occurring concepts is the pair rock and roll (Lancia, 2007; Buchanan, Holmes, Teasley, & Hutchison, 2013). One word can remind us of the other one because of their main association with the music style, although there is no apparent semantic relationship. Moreover, there is evidence of a graded effect of association strength: the N400 to the second word in moderately associated word pairs (e.g., protection – cover) is larger than in strongly associated pairs (e.g., atom – bomb) but smaller than in the pairs without association (e.g., parade – slice) (Ortu, Allan, & Donaldson, 2013).
While numerous studies have investigated the effects of associations learnt during the previous life, only a few works were devoted to associative relationships built immediately in the course of the experiment. In the acquisition phase of the experiment of Balass, Nelson, and Perfetti (2010), participants learned very rare English words. Specifically, the learned either to associate spelling of the word with its pronunciation (OP condition), or spelling with meaning (OM condition), or pronunciation with meaning (PM condition). In the test phase word pairs were presented, whose first word was either one of the just learnt rare words, or a new rare word, or a highly familiar word. The N400 to the second word was larger to new rare words and to words previously learnt in the OP condition that to highly familiar words and to rare words learnt in the OM or PM condition. Thus, the results indicated that the N400 is specifically affected by the semantics of newly learnt words rather than their orthography and pronunciation.
The data of another semantic learning experiment (Borovsky, Kutas, & Elman, 2010) lead to the same conclusion. In the leaning phase of the experiment, participants were presented real words or pseudowords in sentences strongly or weakly restraining their meaning (examples for a pseudoword being: He tried to put the pieces of the broken plate back together with MARF, and She walked across the room to Mike's desk to return his MARF, for strong and weak context, respectively). In the test phase participants had to decide whether a sentence had an appropriate (e.g., they used a MARF) or inappropriate (e.g., they greeted a MARF) meaning. The most interesting result was obtained in response to the verb (“used” or “greeted”) in the test sentence: for real words regardless of the context in the learning phase, and for pseudowords presented with strong context (but not for pseudowords presented with weak context), the N400 was significantly larger to inappropriate than appropriate verbs.
Sometimes, the N400 in response to verbal stimuli is followed by a late positive component (LPC) – a long-latency, positive-going shift, usually with a parietal maximum. Although semantic learning can occasionally happen with a single presentation (see Borovsky et al., 2010), mostly words are presented repeatedly to learn. The main effect of repetition of words in lists is the marked enhancement of the LPC (e.g., Rugg, 1985; Rugg, Doyle, & Holdstock, 1994; Rugg, Pearl, Walker, Roberts, & Holdstock, 1994) – either along with a reduction of the N400 or without it (Van Petten, Kutas, Kluender, Mitchiner, & McIsaac, 1991). As compared with the semantic N400 effect described above, the similar N400 effect elicited by word repetition peaks later and lasts longer (e.g., Besson, Kutas, & Van Petten, 1992).
Whereas the N400 priming effect contrasting semantically related versus semantically unrelated word pairs has been carefully investigated in numerous experiments (e.g., Holcomb, 1993; Kiang, Patriciu, Roy, Christensen, & Zipursky, 2013; Kiefer, 2005), we could not find any study that contrasted two groups of equally unrelated word pairs: learnt versus unlearnt pairs. Since learning of non-associated word pairs involves their repeated presentation, a repetition effect can be expected. However, in the current case two repetition effects can be dissociated: (1) single word level repetition effect that may cause similar effects of N400 reduction with an increase of the LPC, and (2) the repetition of the association within word pairs, that has not been studied by previous research.
In two experiments we tested whether unintentional learning of semantically not associated word pairs would result in N400 attenuation similar to that in semantically associated pairs (Balass et al., 2010; Holcomb, 1993). We compared ERPs to unrelated word pairs that have been recently learnt (“old” pairs) and to other unrelated word pairs that were presented for the first time (“new pairs”). In the first experiment, we expected that the N400 to the second word in a pair to be smaller in learnt word pairs than in new word pairs because consistent repetition of semantically unrelated word pairs may make them acting like related word pairs. The effect is expected to be stronger than the repetition effect on the first words, thus indicating that associated pair repetition involves some processes beyond simple word repetition. Further, we aimed to clarify whether the N400 effect is to be explained by increased familiarity of repeatedly presented words or is based on building new associations between the members of a pair. For this sake, in the second experiment we introduced so-called old-new where the first word was already learnt in another association, and the second word was new (unlearnt). We hypothesized that, if the new association between words in pairs has an effect independent of that of repetition, then the N400 to the second word in the old-new pairs would have larger amplitude as compared to the new-new pairs.
Section snippets
Participants
Twenty-two subjects participated in this study. The group comprised thirteen females and nine males with a mean age of 25.23 (range: 19–33, standard deviation: 12.45) recruited from the University of Tuebingen student population. None of them reported visual, auditory or neurological deficits. According to the Edinburgh Handedness Scale (Oldfield, 1971), seventeen participants were right-handed, four left-handed and one participant was ambidextrous. They were either native German speakers or
Performance
In the first experiment, median reaction time (RT) was slower after new word pairs as compared to old pairs in the acquisition and test phases (F (2, 40) = 21.26, p < .001, η2 = 0.55), with means (averaged over individual medians) ± standard error (SE) being 937 ± 63.3 ms, 867 ± 62.7 ms, and 1120 ± 75.1 ms, for old (acquisition), old (test), and new pairs, resp. Pairwise comparisons confirmed the significance of the differences: old (acquisition) vs new, p < .001; old (test) vs new, p < .001;
Discussion
In the following, we shall first summarize the obtained results. Then, we shall discuss whether, and in what extent, the standard repetition effect could explain these results. After this, we shall discuss other factors necessary for explanation. Finally, we shall present some speculations about possible associative mechanisms and discuss limitations of our study.
Conclusions
We found that the acquisition of new word association in semantically unrelated word pairs results in a change of N400 amplitude very much similar to that observed in semantically related word pairs. The mere effect of word repetition, though present, does not explain all obtained results. Consistent co-occurrence of words elicited additional effects more specifically related to associative learning. The relative role of automatic processes (particularly, the Automatic Spreading Activation) and
Author statement
Maryam Farshad: Formal analysis, Writing – original draft Preparation, Yuri G. Pavlov: Conceptualization, Formal analysis, Investigation, Methodology, Projectadministration, Validation, Visualization, Writing – review & editing, Boris Kotchoubey: Conceptualization, Formal analysis, Methodology, Project administration, Supervision, Funding acquisition, Writing – review & editing.
Acknowledgments
The study was supported by the German Research Society (Deutsche Forschungsgemeinschaft), Grant KO-1753/13.
References (58)
- et al.
Word learning: An ERP investigation of word experience effects on recognition and word processing
Contemporary Educational Psychology
(2010) - et al.
Learning to use words: Event related potentials index single-shot contextual word learning
Cognition
(2010) - et al.
Auditory event-related potentials to semantic priming during sleep
Electroencephalography and Clinical Neurophysiology
(1998) - et al.
Mediated priming in the lexical decision task: Evidence from event-related potentials and reaction time
Journal of Memory and Language
(2000) The electrophysiology of incidental and intentional retrieval: ERP old/new effect in lexical decision and recognition memory
Neuropsychologia
(1999)- et al.
Automatic vs. controlled processes in semantic priming differentiation by event-related potentials
International Journal of Psychophysiology
(2002) Automatic and attentional processing: An event-related brain potential analysis of semantic priming*1
Brain and Language
(1988)- et al.
Test–retest reliability and stability of N400 effects in a word-pair semantic priming paradigm
Clinical Neurophysiology
(2013) The N400 is modulated by unconsciously perceived masked words: Further evidence for an automatic spreading activation account of N400 priming effects
Cognitive Brain Research
(2002)Event-related potentials, cognition, and behavior: A biological approach
Neuroscience & Biobehavioral Reviews
(2006)