Memory traces for tonal language words revealed by auditory event-related potentials

Abstract In tonal languages such as Mandarin Chinese, suprasegmental tones are used to signal word meaning besides consonantsand vowels. To reveal memory traces for tonal language words, we presented native Mandarin Chinese speakers with asequence of spoken syllables as standards and disyllables as deviants in a passive oddball paradigm.The second syllableof each disyllable carried critical tonal information that would define the disyllable either as a meaningful word or as ameaningless pseudoword. The words and pseudowords were acoustically and phonologically matched as well ascounterbalanced. The auditory event-related potential in response to words was more negatively deflected than that inresponse to pseudowords. This effect was most prominent 164 ms after the word recognition point. Our study indicatesan activation of memory traces for tonal language words. Descriptors: Memory trace , Tonal language , Disyllabic word Pseudoword Auditory ERPAbout 60 to 70% of the world’s known languages are tonal (Yip,2002). In addition to consonants and vowels, tonal languages usetones to differentiate word meaning (Pike, 1948). Fundamentalfrequency (f0) contour, perceived as pitch contour, provides theprimary cue for tonal perception (Abramson, 1978; Fok Chan,1974; Gandour, 1978). In Mandarin Chinese, which is the mostwidely spoken tonal language in the world, there are four lexicaltones. For example, the syllable [bai], with tone 1, here denoted as[bai1], means “split,” whereas [bai2], [bai3], and [bai4] mean“white,” “swing,” and “defeat,” respectively. Unlike consonantsandvowels,whichareconsideredsegmentsinlinguistics,tonesareconsidered suprasegmental, meaning that their acoustic attributioncan extend over more than one speech segment. In autosegmentalphonology, a tonal tier is proposed in parallel with a segmental tier(Goldsmith, 1979).The concepts of tones and segments in linguistics raise thequestion of whether they involve different neural and cognitivemechanisms. Behavioral, electrophysiological, functional neu-roimaging, and brain lesion studies have been carried out to con-trast tonal and segmental perception or production. Some studiessupport that the processing of tones and segments involves a dif-ferent neural and cognitive mechanism (Gandour et al., 2000; LawOLiangvLiuet al.,2006;Luoet al.,2006) whereas the other studies fail to support it (Hsieh, Gandour,Wong, & Hutchins, 2001; Packard, 1986; Schirmer, Tang, Penney,Gunter, & Chen, 2005). In nontonal languages, the language-experience-dependent long-term memory traces for words arereflected in the fact that the mismatch negativity (MMN) responsesto words are more pronounced than responses to acousticallymatched pseudowords (Pulvermuller et al., 2001; Shtyrov,Kimppa, Pulvermuller, & Kujala, 2011; Shtyrov, Osswald, P Shtyrov & Pulvermuller, 2002). We are inter-estedinwhethersuprasegmentaltoneisalsoimplicatedinthewordmemory traces.Electroencephalographic (EEG) MMN and its magnetoen-cephalographic (MEG) counterpart, MMNm, are automatic brainresponses to any discriminable change in a repetitive auditorystimulation (Naatanen, Tervaniemi, Sussman, Paavilainen, W Picton,Alain, Otten, Ritter, A Schro-ger, 1998). MMN is usually elicited using an oddball paradigm inwhich a frequent