普通话水平测试的语言概率研究
摘要
普通话水平测试(下简称．PSC)是推广普通话工作的重要组成部分，是使
推广普通话工作逐步走向科学化、规范化、制度化的重要举措。PSC在国家语委
《普通话水平测试大纲》(下简称《大纲》)的指导下，正在从无到有地展开，国
家、地方的语委和普通话培训测试中心在测试理论和方法上作了大量的研究工
作。PSC是一个新兴的较为复杂的语言测试系统工程，既需要宏观的政策导向和
理论指导，也离不开微观的调查研究和统计分析，《大纲》作为指导我国PSC的
总纲，它在测试理论、测试覆盖面等方面需要统计语言学相关数据的有力支持。
在PSC工作中，《大纲》要求及试卷的声、韵、调、音节等频率情况，是否符合
汉语声韵调音节等客观语言概率，试卷究竟应该达到怎样的覆盖率指标，至今仍
然是人们不断改进和探讨的重要问题。PSC作为一种语言测试，本身具有测试对
象的广泛性和测试实施的特殊性，只有通过统计语言学的理论和方法，才能得出
确切而非经验的各项量化指标，才能进行精确的测试，才能进行科学的、更大领
域的判断和预测，提高PSC的科学性、准确性。
如果缺少统计语言学数据的支持，PSC必然是经验性。本文拟在应用语言
学的基础上，采用统计语言学的科学理论和方法，通过实验分析，以量化的手段，
对：PSC进行研究，力求对覆盖率指标等问题作出较为客观、科学的回答，从而
指导规范卷的命制，使测试卷反映的客观数据与实际语言概率大体相近，从而使
之更具有客观性、代表性和规范性。
在本研究中，我们通过对《现代汉语频率词典》以及根据《大纲》命制的
试卷进行了大量科学、细致地统计，借鉴了《声学手册》以及王渝光教授《普通
话水平测试研究》中的一些科研成果，并对这些数据进行教育统计等方面的统计
分析和理论论证，对汉语客观语言概率与《大纲》要求、试卷声韵调音节频率及
试题覆盖率情况，进行了详细的对比，得出如下结论：《大纲》规定的测试要求
基本符合汉语客观语言概率情况；题库中试题的单词、词语部分以及朗读篇目的
大部分试题基本符合汉语客观语言概率；400字的朗读篇目可分成两部分，或为
2篇200字的短文，或为300字和100字两个篇目；从实际操作的可行性来看，
规范卷汉字的覆盖面应达到400个以上，音节覆盖面应达到230个以上；从试卷
的外观形式来看，单字、词语部分不应出现相同的字词，多音字应尽量避免在单
字部分出现，某些字词在朗读篇目中不应过分集中，两个任选话题应有较大的差
别，朗读材料不应对话题有提示作用，以保证试卷有足够广的覆盖面，保证在相
对较少的时间内考查到尽可能多的内容。
Putonghua Shuiping Ceshi(shortened as PSC below )is a main component of
popularizing Putonghua and an important measure to develop Putonghua into a scientific,
standardized and systematic stage. Directed under "An outline on Putonghua Shuiping
Ceshi" (shortened as "Outline" below ) of the National Language Committee ,PSC is
gradually being launched .The National and Local Language Com/nittee and the Center of
PSC have done much study work in the theory and methodology of test. PSC is a new
and more complicated systematic project of language test, it not only needs macroscopic
guidance of policy and the direction of theory, but the microscopic investigation and
statistic analysis. The Outline, as the total outline directing PSC, needs the powerful
support of concerning data of the Statistical Linguistics about the test theory and
probability, etc. In PSC, whether the demand of the Outline and the frequency of the
initial consonant, compound vowel, tone and syllable of the test paper conform to the
Chinese objective language probability and what probability the test paper should reach
are still a vital issue awaiting further exploration and development. As a matter of fact,
PSC, as a kind of language test, is characterized by widespread objects and special
implementation, in which only with the aid of the Statistic Linguistics in theory and
methodology, rather than from the experience, can accurate data and quota be acquired
and judged and evaluated with more scientism and accuracy in a more extensive field.
If lacking the data support of the Statistical Linguistics, PSC is inevitably empirical.
On the ground of the Applied Linguistics, this essay tries to study PSC by experiment and
analysis in the way of the scientific theory of the Statistical Language, to give a more
objective and scientific answer to the probability by means of quantity, to guide how to
work out the standard test, which assure that the objective data of the test questions are
nearly same as the Chinese objective probability and the test questions are more objective,
typical and regular.
In this thesis, we did much scientific and careful statistics on Modem Chinese
Frequency Dictionary and the test paper, adopted some achievement from Acoustics
Handbook and PSC Study, made some statistic analysis and theoretic reason. By the
comparison of the Chinese objective probability with the Outline's demand and the
probability of the test paper, we made the following conclusions: 1) the test demand of the
Outline nearly corresponds to the Chinese objective probability; 2) the test questions in
the item bank mostly correspond to the Chinese objective probability; 3) the reading
material of 400 words can be composed of: two articles of 200 words or two articles of
300 words and 100 words; 4) there should be more than 400 different words and 230
syllables in a test paper ; 5) in test paper, there should not be same word in the 1 st and
2nd section; the words with several pronunciations should be avoided in the 1st
section ;there should not be excessive same words in the 3rd section; the two optional test
questions in the 4th section should be much different ; the 3rd section should not
enlighten the 4th section.
推广普通话工作逐步走向科学化、规范化、制度化的重要举措。PSC在国家语委
《普通话水平测试大纲》(下简称《大纲》)的指导下，正在从无到有地展开，国
家、地方的语委和普通话培训测试中心在测试理论和方法上作了大量的研究工
作。PSC是一个新兴的较为复杂的语言测试系统工程，既需要宏观的政策导向和
理论指导，也离不开微观的调查研究和统计分析，《大纲》作为指导我国PSC的
总纲，它在测试理论、测试覆盖面等方面需要统计语言学相关数据的有力支持。
在PSC工作中，《大纲》要求及试卷的声、韵、调、音节等频率情况，是否符合
汉语声韵调音节等客观语言概率，试卷究竟应该达到怎样的覆盖率指标，至今仍
然是人们不断改进和探讨的重要问题。PSC作为一种语言测试，本身具有测试对
象的广泛性和测试实施的特殊性，只有通过统计语言学的理论和方法，才能得出
确切而非经验的各项量化指标，才能进行精确的测试，才能进行科学的、更大领
域的判断和预测，提高PSC的科学性、准确性。
如果缺少统计语言学数据的支持，PSC必然是经验性。本文拟在应用语言
学的基础上，采用统计语言学的科学理论和方法，通过实验分析，以量化的手段，
对：PSC进行研究，力求对覆盖率指标等问题作出较为客观、科学的回答，从而
指导规范卷的命制，使测试卷反映的客观数据与实际语言概率大体相近，从而使
之更具有客观性、代表性和规范性。
在本研究中，我们通过对《现代汉语频率词典》以及根据《大纲》命制的
试卷进行了大量科学、细致地统计，借鉴了《声学手册》以及王渝光教授《普通
话水平测试研究》中的一些科研成果，并对这些数据进行教育统计等方面的统计
分析和理论论证，对汉语客观语言概率与《大纲》要求、试卷声韵调音节频率及
试题覆盖率情况，进行了详细的对比，得出如下结论：《大纲》规定的测试要求
基本符合汉语客观语言概率情况；题库中试题的单词、词语部分以及朗读篇目的
大部分试题基本符合汉语客观语言概率；400字的朗读篇目可分成两部分，或为
2篇200字的短文，或为300字和100字两个篇目；从实际操作的可行性来看，
规范卷汉字的覆盖面应达到400个以上，音节覆盖面应达到230个以上；从试卷
的外观形式来看，单字、词语部分不应出现相同的字词，多音字应尽量避免在单
字部分出现，某些字词在朗读篇目中不应过分集中，两个任选话题应有较大的差
别，朗读材料不应对话题有提示作用，以保证试卷有足够广的覆盖面，保证在相
对较少的时间内考查到尽可能多的内容。
Putonghua Shuiping Ceshi(shortened as PSC below )is a main component of
popularizing Putonghua and an important measure to develop Putonghua into a scientific,
standardized and systematic stage. Directed under "An outline on Putonghua Shuiping
Ceshi" (shortened as "Outline" below ) of the National Language Committee ,PSC is
gradually being launched .The National and Local Language Com/nittee and the Center of
PSC have done much study work in the theory and methodology of test. PSC is a new
and more complicated systematic project of language test, it not only needs macroscopic
guidance of policy and the direction of theory, but the microscopic investigation and
statistic analysis. The Outline, as the total outline directing PSC, needs the powerful
support of concerning data of the Statistical Linguistics about the test theory and
probability, etc. In PSC, whether the demand of the Outline and the frequency of the
initial consonant, compound vowel, tone and syllable of the test paper conform to the
Chinese objective language probability and what probability the test paper should reach
are still a vital issue awaiting further exploration and development. As a matter of fact,
PSC, as a kind of language test, is characterized by widespread objects and special
implementation, in which only with the aid of the Statistic Linguistics in theory and
methodology, rather than from the experience, can accurate data and quota be acquired
and judged and evaluated with more scientism and accuracy in a more extensive field.
If lacking the data support of the Statistical Linguistics, PSC is inevitably empirical.
On the ground of the Applied Linguistics, this essay tries to study PSC by experiment and
analysis in the way of the scientific theory of the Statistical Language, to give a more
objective and scientific answer to the probability by means of quantity, to guide how to
work out the standard test, which assure that the objective data of the test questions are
nearly same as the Chinese objective probability and the test questions are more objective,
typical and regular.
In this thesis, we did much scientific and careful statistics on Modem Chinese
Frequency Dictionary and the test paper, adopted some achievement from Acoustics
Handbook and PSC Study, made some statistic analysis and theoretic reason. By the
comparison of the Chinese objective probability with the Outline's demand and the
probability of the test paper, we made the following conclusions: 1) the test demand of the
Outline nearly corresponds to the Chinese objective probability; 2) the test questions in
the item bank mostly correspond to the Chinese objective probability; 3) the reading
material of 400 words can be composed of: two articles of 200 words or two articles of
300 words and 100 words; 4) there should be more than 400 different words and 230
syllables in a test paper ; 5) in test paper, there should not be same word in the 1 st and
2nd section; the words with several pronunciations should be avoided in the 1st
section ;there should not be excessive same words in the 3rd section; the two optional test
questions in the 4th section should be much different ; the 3rd section should not
enlighten the 4th section.
引文
王渝光：《普通话水平测试题库系统》，《普通话水平测试研究》，语文出版社，1997
骆小所：《语言学概论》，云南人民出版社，
桂诗春：《语言测试：新技术与新理论》，外语教学与研究，1989
卢纹岱：《 SPSS for Windows 统计分析》，电子工业出版社，2000
冯志伟：《数理语言学》，知识出版社，1985
马大猷 、沈 山豪：《声学手册》，科学出版社，1983
宋欣桥：《普通话水平测试员手册》，商务印书馆，2000
王孝玲：《教育统计学》，华东师范大学出版社，1994
北京语言学院语言教学研究所：《现代汉语频率词典》，北京语言学院出版社，1986
外文参考文献：
Anthony Woods, Paul Fletcher, Arthur Hughes: Statistics in Language Studies,北京语言文化大学出版社，2000
Alan Davies: Principles of Language Testing, 经济科学出版社，1997
骆小所：《语言学概论》，云南人民出版社，
桂诗春：《语言测试：新技术与新理论》，外语教学与研究，1989
卢纹岱：《 SPSS for Windows 统计分析》，电子工业出版社，2000
冯志伟：《数理语言学》，知识出版社，1985
马大猷 、沈 山豪：《声学手册》，科学出版社，1983
宋欣桥：《普通话水平测试员手册》，商务印书馆，2000
王孝玲：《教育统计学》，华东师范大学出版社，1994
北京语言学院语言教学研究所：《现代汉语频率词典》，北京语言学院出版社，1986
外文参考文献：
Anthony Woods, Paul Fletcher, Arthur Hughes: Statistics in Language Studies,北京语言文化大学出版社，2000
Alan Davies: Principles of Language Testing, 经济科学出版社，1997