This study aims to explore the acoustic features of Japanese geminate stop articulated by the native Cantonese speakers, and to reveal the condition of interference from the aspect of the entering consonant of Cantonese. Through doing a recording experiment, while comparing the Japanese geminate stop pronounced by native Japanese speakers with the Japanese geminate stop pronounced by native Cantonese speakers and native northern Chinese dialect speakers, the author analyzes the interference of the entering consonant in Japanese geminate stop of native Cantonese speakers based on the experimental results from aspects of the duration time of vowels and consonants.

The experimental results reveal that the acoustic features of Japanese geminate stop articulated by native Cantonese speakers are not reflected in the consonants but in the vowels. While dealing with the pattern of preceding vowels and subsequent vowels of Japanese geminate stop, the native Cantonese speakers are affected by the entering consonant of Cantonese. Therefore, they tend to use a pattern very similar to Cantonese. Especially for the subsequent vowels, the native Cantonese speakers get accustomed to lengthening them.

The purpose of this research is to (1) measure event-related potential (ERP) obtained when a sentence including a non-contradictory reading Nagara clause with a non-grammatical nominative noun phrase (NP) is used as a stimulus and (2) investigate the implications of the measured ERP on the distribution restriction of the nominative NP in the Nagara clause. Since the P600 component, which is said to cause a syntactic deviation, appeared in the results of the experimentations, the analysis that the distribution restriction of nominative NP in a non-contradictory Nagara clause is a syntactic restriction is supported. However, the P600 component appeared only in the group in which the corresponding grammatical Nagara clause sentences were shown as stimulus sentences and not in the group in which the corresponding grammatical stimulus sentence was not presented. Regarding the amount of potential, only the corresponding stimulus sentences could obtain a significant difference. From these results, it was confirmed that cases presenting grammatical pairs of stimulus sentences were more likely to grasp syntactic deviations of non-grammatical stimulus sentences, and that the P600 effect could be obtained positively.

This study investigates features of rhoticised vowels in several Irish accents of English. It identifies the distribution of rhoticised vowels in each Irish accent. As the rhoticity lowers the frequencies of the third formant, the transition of the third formants in the rime (a vowel + /r/) duration is examined. This paper reveals that the vowels which are rhoticised vary from accent to accent. The NURSE vowel is rhoticised by the subjects from Dublin, Tyrone, Mayo, Cork, and Waterford, while the START vowel is rhoticised only by those from Mayo, Cork, and Waterford. Overall, central vowels, such as the NURSE and START vowels, are more likely to be rhoticised than other vowels.

It is commonly assumed that /l/ is consistently realised as the clear (palatalised) variant in Welsh and Irish accents of English, while the dark (velarised) variant is realised in Scottish accents. The clear /l/ acoustically lowers the frequencies of the first formant (F1) and raises those of the second formant (F2), while the dark /l/ is the other way around. This study examines allophonic features of /l/ in these three accents of English by measuring the frequencies of F1 and F2 of /l/ from the original recordings. The results of F2 confirm the above-mentioned differences among the accents (clear /l/ in Welsh and Irish accents and dark /l/ in Scottish accents), while those of F1 do not. Furthermore, this paper demonstrates that onset /l/ is slightly, but consistently, clearer (lower F1 frequencies and higher F2 frequencies) than coda /l/, although in these three accents clear /l/ and dark /l/ are not in complementary distribution as in Received Pronunciation (RP).