Another point that is worth mentioning is that this observation provides further evidence that subsyllabic structures matter in the computation of prosodic markedness in Japanese. In terms of moraic structure, HL and LH are both sequences of three moras, but there exists a markedness difference between these two structures. This is not a trivial finding, as Japanese has long been believed to be a moraic language (Katada 1990; Labrune 2012; Trubetzkoy 1939), but the finding of this paper points to a role of syllables in Japanese prosodic phonology (McCawley 1968; Ito 1990; Ito and Mester to appear; Kawahara 2012, 2016; Kubozono 1999, 2003; Starr and Shih 2014; Tanaka 2013). 3
5.2.3. Other constructions, other languages
Our study focused on a very speciic environment: the interpretation of elided subjects in Japanese and Mandarin Chinese. It is not the case that our choice is random: we chose this particular construction for reasons that are stated in sections 1 and 2. However, we also fully acknowledge that the generalizability of our conclusion is limited, because our case study is focused on a very speciic construction in two speciic languages. We therefore do not intend to generalize our indings to all cases of L2 acquisition. In particular, in order to more conidently argue for the role of the Subset Principle in L2 acquisition, it is certainly necessary to look at other constructions in other languages. We do hope, however, that we have offered a substantial case study, and it is only through accumulations of these case studies can we investigate the general nature of L2 acquisition.
(Batchelder, 1999; Griner, 2005; Vance, 1987, 1991), and that it is likely that Japanese speakers simply memorize all the inflected forms.
In the first nonce-word study of verbal conjugation patterns, Vance (1987) found that (only) 31 out of 50 participants showed epenthesis when conjugating a nonce verb /hok-u/. Other responses were /hokutta/ (16) or /hota/ (3). A follow-up study (Vance, 1991) also shows that native Japanese speakers do not unambiguously choose the correct past tense form for a /k/-final stem. A later study (Griner, 2005) showed even fewer “correct” epenthesis responses, actually about 10%. A consen- sus that is emerging from these studies is that verb conjugation patterns are not rule-governed. To quote Vance (1991, p. 156), “[this experimental result] is consistent with the claim that even morphologically regular Japanese verb forms are stored in the lexicon.”
model. We also included random intercepts for talker and for word. Table 2 provides a summary of the fixed factors in the baseline model; Table 3 summarizes the Entropy model. Both models were fit to 280,550 data points (see the method section).
We start with a description of the baseline model. The intercept represents an abstract reference category (/a/ before consonants that are voiced, glottal, non-palatal, sonorant). At 89.9 ms, the intercept is higher than the average vowel duration in the data (Table 1). Accordingly, most of the fixed factors are negative, functioning to lower predictions relative to the intercept. The effect of VOWEL is small but reliable. It is in a negative direction, reflecting the observation
LEE, Seunghun (ICU) & KAWAHARA, Shigeto (Keio)
How do we then explain the patterns in (2)? We maintain that the same principle that Shaw et al. (2014) argue for—keep P(message| signal, context) high (Hall et al. 2016)—is at work. Take the leftmost column in (1) as an example. Given the context of the [hui] family, the probability of [hong] picking out the right denotation (/hong+hui/) is 1; on the other hand, the probability of [hui] picking out the right denotation is 1/5 (or to generalize, 1/n where n is the number of family members). In terms of entropy, [hui] has 0 bit, whereas [hong] has 2.3 bits (-log 2 (0.2)). In truncation, it is crucial that the listener can recover the original form before
3) EMA is a useful tool to quantify the degrees of jaw dis- placement (see e.g. Erickson et al. 2012). In some cases, it is necessary to factor out the effect of vowel height to di- rectly see the prosodic effects, because vowel height also affects jaw displacement in addition to prosodic strength (see e.g. Kawahara et al. 2014, Menezes and Erickson 2013, Williams et al. 2013). The C/D model’s approach to prosody has focused on jaw displacement; F0 control of prosody is an aspect of the C/D model which is still “under construction.” (p.c. Osamu Fujimura, 2015).
In contrastively emphasized speech, however, an op- posite effect is found, with low jaw opening occurring with high F0. That is, emphasized syllables have more jaw opening than non-emphasized syllables, [2, 3, 4, 8, 9, 36, 37] and also, generally, higher F0 [26, 38-40] (alt- hough, emphasis can occur with low F0 as well [41-44]). High F0 with large jaw opening may present an articu- latory challenge, if large jaw opening by default results in lower F0 for anatomical reasons. Do speakers resort to a special articulatory maneuver to cope with this ap- parent conflict? This is one of the main questions ad- dressed in the current study.