Objective::The aim of this study is to develop and evaluate a set of psychometrically equivalent disyllabic wordlist(SC-10)in Singapore Mandarin for clinical use.Study design::A preliminary set of 1000 words were obta...Objective::The aim of this study is to develop and evaluate a set of psychometrically equivalent disyllabic wordlist(SC-10)in Singapore Mandarin for clinical use.Study design::A preliminary set of 1000 words were obtained from a dictionary of frequently used words by Singapore students.Ten native judges rate the familiarity level of each word.This is followed by a face-to-face public survey to rank the shortlisted word set from most to least familiar.The final 108 disyllabic words were recorded by a native female talker.20 normal hearing subjects were used to obtain the percentage of correct word recognition at 24 intensity levels(-10 dB HL to 26 dB HL in 2 dB increment).Psychometric function slopes were calculated for each word.100 words were eventually chosen and assigned into ten 10-word lists based on a psychometric balancing protocol.Minor digital adjustments were made to the intensity of each wordlist to improve their auditory homogeneity.Results::The calculated regression slopes in the psychometric functions for the ten lists are between 8.0 and 9.8%/dB.Single factor ANOVA analysis showed no significant difference in both the mean intensity required to obtain 50%recognition score(f=0.109,df=9,p=0.999)and the slopes of the psychometric functions(f=0.078,df=9,p=0.999)between the ten word lists.List validation on 25 normal hearing participants(PTA=11.0 dB HL,SD=4.3)showed a mean speech recognition threshold(SRT)of 9.3 dB HL(SD=3.5)and regression slope of 8.395%/dB.Quadratic regression analysis showed a positive correlation(r^(2)=0.923)between presentation level and word recognition score(WRS).The difference between PTA and SRT of each subject all fall within the clinically acceptable difference of 10 dB HL.Test-retest reliability,carried out on 20 subjects at three levels(SRT-5,SRT,and SRT+5 dB),showed no significance difference between word recognition score when the same participant is tested again at the same intensity level using a different wordlist.Conclusion::All in all,it shows that the SC-10 speech materials are valid for clinical use for Mandarin speech audiometry in Singapore.展开更多
Word sense disambiguation is used in many natural language processing fields. One of the ways of disambiguation is the use of decision list algorithm which is a supervised method. Supervised methods are considered as ...Word sense disambiguation is used in many natural language processing fields. One of the ways of disambiguation is the use of decision list algorithm which is a supervised method. Supervised methods are considered as the most accurate machine learning algorithms but they are strongly influenced by knowledge acquisition bottleneck which means that their efficiency depends on the size of the tagged training set, in which their preparation is difficult, time-consuming and costly. The proposed method in this article improves the efficiency of this algorithm where there is a small tagged training set. This method uses a statistical method for collocation extraction from a big untagged corpus. Thus, the more important collocations which are the features used for creation of learning hypotheses will be identified. Weighting the features improves the efficiency and accuracy of a decision list algorithm which has been trained with a small training corpus.展开更多
目的探讨以汉语语音学理论为依据编制普通话言语测听材料(mandarin speech test materials,MSTMs)单音节小词表,从全新角度建立更适用于临床应用的评估材料。方法以语音学理论为基础,将声母、韵母进行归类缩减后拼合。按照拼合后的音节...目的探讨以汉语语音学理论为依据编制普通话言语测听材料(mandarin speech test materials,MSTMs)单音节小词表,从全新角度建立更适用于临床应用的评估材料。方法以语音学理论为基础,将声母、韵母进行归类缩减后拼合。按照拼合后的音节选择日常生活中常用字,编制完成单音节小词表。对编制材料进行数字化录音以及音频处理,形成CD测试材料,并对词表进行初步等价性评估。结果编制出体现语音学特征的汉语普通话言语测听小词表24张,每表20个单音节字。结论体现语音学特征的汉语普通话言语测听小词表可以明显缩短测试时间,但在临床推广前尚需等价性评估。展开更多
文摘Objective::The aim of this study is to develop and evaluate a set of psychometrically equivalent disyllabic wordlist(SC-10)in Singapore Mandarin for clinical use.Study design::A preliminary set of 1000 words were obtained from a dictionary of frequently used words by Singapore students.Ten native judges rate the familiarity level of each word.This is followed by a face-to-face public survey to rank the shortlisted word set from most to least familiar.The final 108 disyllabic words were recorded by a native female talker.20 normal hearing subjects were used to obtain the percentage of correct word recognition at 24 intensity levels(-10 dB HL to 26 dB HL in 2 dB increment).Psychometric function slopes were calculated for each word.100 words were eventually chosen and assigned into ten 10-word lists based on a psychometric balancing protocol.Minor digital adjustments were made to the intensity of each wordlist to improve their auditory homogeneity.Results::The calculated regression slopes in the psychometric functions for the ten lists are between 8.0 and 9.8%/dB.Single factor ANOVA analysis showed no significant difference in both the mean intensity required to obtain 50%recognition score(f=0.109,df=9,p=0.999)and the slopes of the psychometric functions(f=0.078,df=9,p=0.999)between the ten word lists.List validation on 25 normal hearing participants(PTA=11.0 dB HL,SD=4.3)showed a mean speech recognition threshold(SRT)of 9.3 dB HL(SD=3.5)and regression slope of 8.395%/dB.Quadratic regression analysis showed a positive correlation(r^(2)=0.923)between presentation level and word recognition score(WRS).The difference between PTA and SRT of each subject all fall within the clinically acceptable difference of 10 dB HL.Test-retest reliability,carried out on 20 subjects at three levels(SRT-5,SRT,and SRT+5 dB),showed no significance difference between word recognition score when the same participant is tested again at the same intensity level using a different wordlist.Conclusion::All in all,it shows that the SC-10 speech materials are valid for clinical use for Mandarin speech audiometry in Singapore.
文摘Word sense disambiguation is used in many natural language processing fields. One of the ways of disambiguation is the use of decision list algorithm which is a supervised method. Supervised methods are considered as the most accurate machine learning algorithms but they are strongly influenced by knowledge acquisition bottleneck which means that their efficiency depends on the size of the tagged training set, in which their preparation is difficult, time-consuming and costly. The proposed method in this article improves the efficiency of this algorithm where there is a small tagged training set. This method uses a statistical method for collocation extraction from a big untagged corpus. Thus, the more important collocations which are the features used for creation of learning hypotheses will be identified. Weighting the features improves the efficiency and accuracy of a decision list algorithm which has been trained with a small training corpus.
文摘目的探讨以汉语语音学理论为依据编制普通话言语测听材料(mandarin speech test materials,MSTMs)单音节小词表,从全新角度建立更适用于临床应用的评估材料。方法以语音学理论为基础,将声母、韵母进行归类缩减后拼合。按照拼合后的音节选择日常生活中常用字,编制完成单音节小词表。对编制材料进行数字化录音以及音频处理,形成CD测试材料,并对词表进行初步等价性评估。结果编制出体现语音学特征的汉语普通话言语测听小词表24张,每表20个单音节字。结论体现语音学特征的汉语普通话言语测听小词表可以明显缩短测试时间,但在临床推广前尚需等价性评估。
