This paper presents preliminary results from a production study of French close-mid and open-mid vowels /e/ and /ɛ/ by Japanese learners of French as a second language. Among French vowels, the close-mid and open-mid vowels are especially difficult for learners because their pronunciation depends on an interplay of factors such as phonological environment, syllable structure, and spelling. In this study, 8 Japanese learners took part in a production experiment. Based on the analysis of the formant data of their productions and a comparison with French native speakers, this study found that (i) the acoustic characteristics of the two target vowels when produced by native speakers and second language learners differ, (ii) learners do not vary enough the degree of aperture between the two vowels to make a distinction, and (iii) the accuracy of the pronunciation is correlated with the position and spelling of the vowel. Further studies are required to investigate the perception of the two vowels by Japanese second language learners and understand better the difficulties they encounter in phonetic implementation.

This study explores whether an L2 learner modifies the acoustic values in L2 speech even after they shadow non-target speech stimuli produced by a native speaker. In order to test this research question, we ran an experiment, in which Japanese English learners read aloud a set of target words in the pre-shadowing and post-shadowing sessions, and they shadowed a different set of words in the shadowing session. The target group participants shadowed English utterances, and the control group participants shadowed Japanese utterances. Our statistical results suggest that both the groups converged towards English speakers in terms of VOT values to a little extent, regardless of which type of speech stimuli they were exposed to during the shadowing session.

Avoidance of voiced geminates is normally attributed to difficulties in speech production and perception. However, few studies have examined which acoustic correlates are responsible for perception of the voicing contrast in geminates. This study examines whether seven acoustic correlates (duration of preceding vowel (V₁), closure duration, presence/absence of closure voicing, duration of release burst, duration of following vowel (V₂), F0 of V₂, and F1 of V₂), which are assumed to be perceptual cues to voicing in singletons, are also responsible for perception of the voicing contrast in geminates. A perceptual experiment revealed that although the seven cues are indeed significant in singletons, closure duration and duration of the release burst are not significant in geminates. Closure voicing is the strongest cue both in singletons and geminates; however, its relative importance was greater in geminates than in singletons. These results suggest that cue weighting on perception of the voicing contrast is different for singletons and geminates.

This study examined whether clipped compounds in Japanese tend to avoid containing consecutive heavy syllables across the morpheme boundary (*H#H). A survey using the database of existing clipped compounds (Moon 2021) showed that there are fewer forms with H#H than those with LL#LL, H#LL or LL#H. We then conducted an experiment using nonce compounds each of which consisted of a nonce word (N1) and a real word (N2). The results showed that not all forms with H#H were avoided as a clipped compound. We confirmed, however, that when N1 begins with a syllable that contains a long vowel or a geminate consonant, these elements were not adopted as a new clipped compound and instead the third mora following them was adopted, which is consistent with the argument in the past literature.

Simplex loanword truncation is one of the highly productive word formation processes in Japanese. This paper aims to discuss how the length of the output forms of Japanese simplex truncation can be predicted. Although several mechanisms have been proposed in the literature, issues have been identified with each of them. This paper presents a syllable-based account: words will be shortened by deleting either the last two syllables if they have a heavy syllable as their penultimate syllable or the last three syllables if the penultimate syllable is light. This approach can solve the problems arising from the mechanisms proposed in previous studies theoretically. A dictionary-based survey was conducted, the results of which demonstrated that the proposed mechanism can account for approximately the same number of words covered by previous analyses.

This paper examines a progressing accentual change observed in nagara-clause containing a light verb, which entails tonal neutralization. It reveals that the tonal specification of the word immediately preceding the verb exerts a profound effect on the occurrence of neutralization by examining the speech data in the Corpus of Spontaneous Japanese. The results of the quantitative analysis indicate that neutralization is promoted when the preceding word is unaccented, whereas it is suppressed when the preceding word is accented. The second half of the paper provides a theoretical discussion of the results, illustrating that two prosodic constraints, culminativity and OCP, serve as the significant factors determining whether the neutralization occurs.

Representational theories of syllable structure can be broadly classified into two types, (i) segmentcentered approaches and (ii) syllable-centered approaches. This paper provides a summary of the motivations behind the use of syllable constituent categories in each theory type, with concurrent discussions of how moraicity is encoded in representations. Furthermore, this paper demonstrates how the categories corresponding to moraicity are employed in the analysis of Japanese loanword accentuation. Through this study, I aim to arrive at a more justified syllabic representation which reflects the extent to which it eliminates redundancy in representations and explicitly reveals causal relationships in phonological phenomena.

In five-mora loanwords, the accentual pattern varies depending on the syllable structures. For instance, as for loanwords containing the LHH structure, only the traditional patterns appear, while in those containing the HHL structure, half of them do not show the traditional patterns but exhibit novel forward accentuation. This study focuses on such variability and, through an analysis of the accent and phonological structure of five-mora loanwords, demonstrates that the proportion of accentuation is influenced by three factors: the constraint of WSP, the word-final epenthetic vowel, and the internal structure of the penultimate heavy syllable. Furthermore, it is pointed out that the presence of a binary foot in the predicted position of forward accent is a prerequisite for the forward accentuation.

The Mongolian phonemic system exhibits some asymmetries; however, previous studies have not theoretically analyzed them. Using element theory (ET), this study points out and analyzes three asymmetries in the Mongolian consonant system. First, labials and velars are opposed to alveolars in their occurring condition and phonetic realization, suggesting that labials and velars have the common element |U|, but alveolars do not. Second, /g/ behaves like voiced sonorants, whereas /t/ is parallel to other voiceless stops, and /b/ has intermediate characteristics between them in phonetic realization and coda constraint. This is explainable by assuming that /t/ has the elements |ʔ| and |H|, whereas /b/ has only |H| and /g/ has neither. Third, liquids /ɮ/ vs. /r/ are asymmetric in phonetic characteristics, coda constraint, and phonotactics, implying that /ɮ/ has the element |H|, whereas /r/ does not. In conclusion, the asymmetries in Mongolian can be elucidated using ET; in addition, this analysis demonstrates the theoretical validity of ET.

The text reuse detection tool TRACER was used to detect text reuse between Shenoute's Canons, a collection of Coptic letters, and the Coptic Bible. TRACER found 8,414 cases of text reuse between Shenoute’s texts and the Coptic Bible texts. From these, the authors selected 13 quotations that are valuable for Biblical studies. These quotations did not contain phrases indicating that they are quotations from the Bible, and they had many differences from the source Biblical texts, so they had not been discovered in previous research. In this study, it was possible to mechanically discover such implicit quotations with many changes from the large amount of text reuse. At the end of this paper, an interesting example was examined in which two passages from the Psalms were blended and quoted, from among the valuable quotations newly found by TRACER.