本文以《商务馆学汉语词典》的词语辨析栏为主要研究对象，利用北京语言大学HSK动态作文语料库对该栏目中被辨析词的选择、被辨析词对/列的确定、栏目的位置分布等问题进行了探讨；利用北京大学CCL语料库对该栏目中例证的正确性和典型性进行了分析；以《新汉语水平考试HSK大纲》和《汉语水平词汇与汉字等级大纲》为标准，对被辨析词难度和词语辨析栏中释义、用法说明、例证的用词难度进行了考察；通过考察重复程度，对该栏目与词典其他部分的互动关系展开了讨论；最后对该栏目的整体情况做出了评价，并提出了一些编纂建议。主要研究结论如下：第一，部分被辨析词的典型性不足，一些被辨析词列缺乏完整性，在一定程度上影响了易混淆词辨析的有效性；建议将混用率作为选定被辨析词的参考数据，进一步发挥该栏目纠正和预防汉语学习者词语混用的作用。第二，词语辨析栏的分布不够科学，减小了该栏目被词典使用者参阅的几率；建议依据被辨析词的混用方向来确定该栏目的位置，保证该栏目的利用率。第三，词语辨析栏和部分注意栏在说明对象、说明内容上有所交叉；需要进一步明确专栏定位，在突出专栏特性的同时，也要保证专栏内部的一致性。第四，词语辨析栏中缺乏参见信息，使得该栏目联系相关词条、沟通正文和附录的功能受到了影响；建议完善参见系统，提高词典使用效率。第五，词语辨析栏释义、用法说明和例证中的部分用词难度过大，一些表述存在不统一的情况；应该对元语言的难度和数量进行严格控制，以保证词语辨析栏的可读性。第六，词语辨析栏和词典正文词条在释义内容和用法说明上少有重复，但例证之间重复的情况较为突出，不利于体现该栏目的必要性；应该避免与词条内容的直接重复，同时注意在丰富性、有序性、解释性和指导性上全面提高例证质量。

The main object of this research is the discrimination column in the Commercial Press Learner’s Dictionary of Contemporary Chinese. Firstly, by using the HSK dynamic composition corpus of BLCU, the problem of how to select the words in the column, how to arrange them into gropes, how to differentiate them, where to put the columns are discussed. Secondly, by using CCL corpus of PKU, the correctness and typicality of examples are analyzed. Thirdly, New Outline of HSK and Chinese Proficiency Vocabulary and Chinese Characters Grade Outline are token as the standard to analysis the words of interpretations, instructions, and examples. Then, by investigating the degree of repetition, its interactive relationship with the rest of the dictionary is discussed. Finally, the overall evaluation of the columns is given; some compiling suggestions for this column are put forward. The main research conclusions are as follows:First, the typicality of some differentiated words is insufficient; some words lists are lack of integrity, that the effectiveness of differentiating is limited. The confusing rate is proposed to be used as reference data in selecting and differentiating words of the column, in order to further play the function of correcting and preventing words confusion and misusing.Second, distribution of discrimination column is not scientific, which will reduce its possibility to be referred to by users. To analyze the direction of word misused will help determining the column position, and ensure the utilization of the column.Third, the discrimination column, and the attention column have overlapping in content and function, which means clearer positioning and column features need to be further plaied and internal consistency of each column also needs to be reinforced.Fourth, the lack of “see also” information makes the contact of text and appendices, and that of related terms are affected, therefore, perfecting the refer system to improve the efficiency of dictionary use is needed.Fifth, some words used in definitions, instructions and illustrations of discrimination column are difficult, and expressions are not unified, so that the quantity and the difficulty of metalanguage should be strictly controlled, to ensure that differentiated words are easy to understand.Sixth, discrimination column and text entry share some repetition of definitions and instructions; the repeat between examples is not conducive to reflect the necessity of the columns. Therefore, the directly repeat of the entry content should be avoided and also should pay attention to abundant, well-organized and explanatory examples and guidance to improve the quality of the column.