LAS0010.1177/0023830919849081Language and SpeechYan et al 849081research-article2019 Language Original Article and Speech Language and Speech 1 –19 Priming the Representation of © The Author(s) 2019 Article reuse guidelines: Left-Dominant Sandhi Words: A sagepub.com/journals-permissions https://doi.org/10.1177/0023830919849081DOI: 10.1177/0023830919849081 Shanghai Dialect Case Study journals.sagepub.com/home/las Hanbo Yan School of Chinese Studies and Exchange, Shanghai International Studies University, China Yu-Fu Chien Department of Chinese Language and Literature, Fudan University, China Jie Zhang Department of Linguistics, University of Kansas, USA Abstract The paper aims to examine how the acoustic input (the surface form) and the abstract linguistic representation (the underlying representation) interact during spoken word recognition by investigating left-dominant tone sandhi, a tonal alternation in which the underlying tone of the first syllable spreads to the sandhi domain. We conducted two auditory-auditory priming lexical decision experiments on Shanghai left-dominant sandhi words with less-frequent and frequent Shanghai users, in which each disyllabic target was preceded by monosyllabic primes either sharing the same underlying tone, surface tone, or being unrelated to the tone of the first syllable of the sandhi targets. Results showed a surface priming effect but not an underlying priming effect in younger speakers who used Shanghai less frequently, but no surface or underlying priming effect in older speakers who used Shanghai more often. Moreover, the surface priming did not interact with speakers’ familiarity ratings to the sandhi targets. These patterns suggest that left- dominant Shanghai sandhi words may be represented in the sandhi form in the mental lexicon. The results are discussed in the context of how phonological opacity, productivity, the non- structure-preserving nature of tone spreading, and speakers’ semantic knowledge influence the representation and processing of tone sandhi words. Keywords Priming, spoken word recognition, tone sandhi, tone spreading, Shanghai Wu Corresponding author: Hanbo Yan, School of Chinese Studies and Exchange, Shanghai International Studies University, 550 West Dalian Road, Building 2, Room 418, Shanghai, China. Email: [email protected] 2 Language and Speech 00(0) 1 Introduction During spoken word recognition, speakers decode the acoustic signal and map it onto the stored representation to recognize words. However, factors such as talker characteristics, coarticulation, speaking rate, and phonological alternation often lead to a mismatch between the surface and stored representations (Weber & Scharenborg, 2012). Variability in speech has been challenging theories of speech perception and spoken word recognition in that invariance of acoustic cues for a certain phoneme or word seems impossible to find. Therefore, there is not a perfect match between the two representational levels. Although unsystematic variability could be detrimental to spoken word recognition, words undergoing predictable phonological processes may not be as dif- ficult to retrieve since speakers can make inferences regarding the stored representations based on phonological environments. For example, Gaskell and Marslen-Wilson (1996) showed that both phonologically appropriate variants, such as leam bacon, and their citation counterparts, such as lean bacon, facilitated participants’ lexical decision in a cross-modal priming experiment. Gaskell and Marslen-Wilson (1998) argued that such a priming effect is due to the fact that speakers utilize the rule of place assimilation to retrieve the underlying form of the variants during lexical access. Unlike segmental processes (Gow, 2001, 2003; LoCasto & Connine, 2002; Ogasawara, 2012; Snoeren, Seguí, & Hallé, 2008), the processing and representations of words undergoing phono- logical alternations at the suprasegmental level have received considerably less attention. Zhou and Marslen-Wilson (1997) proposed two representational views regarding how tone sandhi words are stored in the mental lexicon. The surface representational view states that tone sandhi words are represented based on the surface form, while the canonical, or underlying representational view considers the abstract linguistic representation to be what speakers accessed during spoken word recognition. The latter view is more consistent with the assumption of traditional generative pho- nology that the surface form of a tone sandhi word is assumed to be derived from an underlying representation using a tone sandhi rule (Chen, 1987, 2000; Shih, 1997). More recent psycholin- guistic studies, however, have demonstrated that both underlying and surface representations may contribute to the processing of tone sandhi words depending on the frequency of the sandhi word and the productivity of the sandhi process (e.g., the rate of sandhi application in novel words esti- mated from wug tests). Chien, Sereno, and Zhang (2016) used an auditory-auditory priming lexical decision experi- ment to examine the representation of Mandarin Tone 3 sandhi words (T3 + T3 → T2 + T3). They found a facilitation effect when monosyllabic primes and the first syllables of disyllabic Tone 3 sandhi targets were matched in the underlying representation. When they were matched on the surface, no effect was observed. Moreover, word frequency of Tone 3 sandhi words does not modu- late the effect of priming. According to their results, Mandarin Tone 3 sandhi words are represented in the underlying representation (/T3 + T3/). While Mandarin Tone 3 sandhi has been reported to apply without exception to novel words in a wug test and is hence extremely productive (Zhang & Lai, 2010),1 Taiwanese tone sandhi is not as productive, producing approximately 40% (512 → 55) to 80% (24 → 33) of expected sandhi outputs for novel words in a similar experiment (Zhang, Lai, & Sailor, 2011). The lower productiv- ity of Taiwanese tone sandhi may be due to its circular chain-shift nature (i.e., the tone undergoes regular changes according to this circle: 51 → 55 → 33 → 21 → 51; 24 → 33, whenever it appears in non-phrase-final positions regardless of the tone in the final position). Chien, Sereno, and Zhang (2017) conducted a follow-up study investigating the processing and representation of less produc- tive Taiwanese tone sandhi words. For the more productive 24 → 33, a strong facilitation effect was observed when monosyllabic primes and the first syllable of disyllabic sandhi targets shared Yan et al 3 Table 1. Shanghai disyllabic left-dominant sandhi for open or sonorant-closed syllables. σ1 σ2 T1 [53] T2 [24] T3 [13] T1 [53] 55-31 T2 [24] 33-44 T3 [13] 22-44 the same underlying representation, but a weaker facilitation effect was found when they were overlapped on the surface. For the less productive 51 → 55, a reverse pattern was obtained, show- ing a strong surface facilitation effect but little underlying contribution to priming. Moreover, both sandhis showed an interaction between familiarity and prime type in their priming effect, indicat- ing that familiarity may modulate the type of priming effect observed. Based on these priming results, Chien et al. (2016, 2017) suggested that the representation and processing of tone sandhi words are influenced by the nature of the sandhi pattern. Although there are differences between Mandarin and Taiwanese sandhi patterns in term of productivity, both keep the base tone of the final syllable in a sandhi domain, while the preceding syllables undergo sandhi. This is known as right-dominant sandhi (Yue-Hashimoto, 1987; Zhang, 2007). On the contrary, in Shanghai Wu, the base tone of the first syllable extends to the entire sandhi domain (Xu, Tang, & Qian, 1981; Zee & Maddieson, 1979). This is illustrated in Table 1, in which the first column represents the base tone on σ1, the first row represents the base tone on σ2, and the body of the table represents the sandhi forms of disyllable words. For example, when 53 in open or sonorant-closed syllables is combined with a syllable carrying any other tone, it undergoes the sandhi 53 + X → 55 +31 (“X” refers to any tone among T1 to T3), with 53 spreading to the whole disyllabic domain. This is known as a left-dominant sandhi pattern. Notice that the sandhi tones 55 and 31 do not exist in Shanghai monosyllables. Only the combination of them forms a legal disyllabic tonal output. According to Xu et al. (1981), modifier-noun ([M N]) compounds and a portion of the verb-noun ([V N]), verb-modifier, subject-predicate, and coordinate compounds undergo this left-dominant sandhi pattern. Zhang and Meng (2016) investigated the productivity of Shanghai tone sandhi in both disyllabic real and novel items and argued that the sandhi pattern is relatively productive. For [M N] words, although statistical comparisons indicated that the fundamental frequency (F0) curves of real and nonce words are usually significantly different from each other, with F0 curves of nonce words showing more tonal characteristics of the base tone than those of real words, the pitch contour over the disyllabic novel [M N] words generally represents the contour of the base tone of the first syl- lable. This suggests a relatively productive application of the rightward tonal extension sandhi. As a relatively productive sandhi pattern, Shanghai speakers may be able to derive the surface sandhi form from the underlying representation (base tone of the first syllable) through a compu- tational