The Acoustic Perception Of The Approximant Initials In Mandarin

At present,the acoustic perception of vowel and tone in Mandarin have been basically completed,but for the consonant,it is limited to the voiceless plosives.Due to the approximant is an important part of the consonant,systematic and comprehensive study of the acoustic perception on approximant is of great significance for enriching and perfecting the acoustic perception of consonant.In order to explore the acoustic perception on approximant initial,the paper adopts the method of identification test and discrimination test based on previous studies,and conduct the unmodified natural speech acoustic experiment and the synthetic speech acoustic experiment to find out the characteristics of the acoustic perception on approximant initials.Iin addition,the nasality of the approximant initials was measured to associate with the experimental results.As indicated,in the natural speech acoustic experiment,the accuracy is high,indicating that the subjects can basically distinguish the approximant initials in mandarin,but there are still slight confusions.In the single-word experiment,in general,the accuracy is lower when the vowels are Qichihu?Hekouhu,oir when the tone is the second tone;In the double-word experiment,the accuracy when the contrast word is in the back is higher than that when the contrast word 1s in the front.This is mainly due to the fact that the preceding reference word increase the reaction time for the participants,so that they can distinguish the words they hear well.In the synthetic speech acoustic experiment,the main acoustic difference of the group/m/-/n/is located on the transition between the initial and the subsequent vowel(the average of the acoustic boundary F2 is 992 Hz);To the group/l/-/r/,both the initial and the transition of the subsequent vowel can be used as acoustic differences,but in terms of the results,the transition(the average of the acoustic boundary F2 is 1184 Hz?F3 is 2345 Hz)is more important;To the group/n/-/1/,neither the initial nor the transition of the vowel(the average of the acoustic boundary F1 is 1587 Hz?F3 is 2970 Hz)can be the perfect acoustic difference.It may due to the bandwidth?duration and intensity which also cause the acoustic difference;To the group/m/-/1/,the acoustic boundary can be basically obtained by modifying the fomiant of the initials(the average of acoustic boundary of/ma/-/la/F1 is 342Hz,the average of acoustic boundary of/mi/-/li/F3 is 2403Hz);To the group/m/-/r/,the acoustic boundary can be basically obtained by modifying the formant of the initials(the average of the acoustic boundaiy F2 is 869 Hz);To the group/n/-/r/,the acoustic boundary can be basically obtained by modifying the formant of the initials(the average of the acoustic boundary F2 is 1214 Hz).In regard to the influencing factors of the experiment,the influencing factor of gender was examined in the natural speech acoustic experiment,and the influencing factors of display order and gender were examined in the synthetic speech acoustic experiment.Neither in the natural speech nor in the synthetic speech,these two influencing factors can not significantly affect the experimental results by the SPSS independent sample T-test.In the synthetic speech acoustic experiment,only the group/m/-/n/(because of the limited data,some groups cannot use SPSS to carry out the correlation analysis)makes significant differences in the syllable of different vowels.Other groups have no significant differences in the syllable of different vowels or tones.Related to the nasality of the approximant,we found that in the condition of vowel,the higher the nasality of the approximant,the more difficult it is to distinguish the speech for the subject;In the condition of tone,to the group of/l/-/r/?/m/-/l/?/n/-/r/,the nasality of the approximant is not closey related to the experimental results either in the natural speech or in the synthetic speech.For the group of/m/-/n/?/m/-/r/,the higher the nasality of the approximant,the more difficult it is to perceive the changes of synthesized speech for the subject.For the group of/n/-/l/,the higher the nasality of the approximant,the lower the accuracy of the natural speech.The main reason is that the excessive nasalization will make the key information lost in the speech,thereby increasing the difficulty of resolution.