Questionnaires are often used to examine people’s attitudes，beliefs and behaviors in language learning and teaching.The data we get from questionnaire research can be especially insightful and satisfying when patterns emerge from a large number of respondents，when apparent differences or similarities are found among groups，or when relationships are ascertained among variables.As researchers，we feel empowered when we make recommendations for learning and teaching if the pattern we find is not only salient and strong，but also attested on a large scale.

However，by nature，survey research is exploratory and shallow.It often does not go beyond pattern finding or relationship mapping.The patterns themselves are long on description and short on explanation.And of course，with every pattern there are exceptions.

Most consumers of our research only care about our conclusions and pedagogical implications.As researchers，however，it is our responsibility to make sure that we have selected the right tool，and that the tool we have used to reach our conclusions is indeed trustworthy.

The questionnaire as a research instrument has been a concern for applied linguistics for some time.As early as 1990，Reid （1990） traced the processes in developing her learning styles measure and boldly revealed the ‘dirty laundry of ESL [English as a second language] survey research’.Luppescu and Day （1990） attempted to develop a questionnaire to canvass the attitudes and beliefs of Japanese learners and teachers towards the learning of English，but found it difficult to validate the student questionnaire.In fact，instead of arriving at any conclusions about student beliefs，they presented a ‘more important conclusion’，that ‘questionnaire data should not blindly be accepted or considered meaningful unless they have been properly validated.

Compared to other fields such as the social sciences，education，marketing，and medicine that extensively make use of questionnaires，applied linguists have paid only cursory attention to the questionnaire tool itself.Despite an occasional paper on questionnaire validation （e.g.Brown，1997） or issues involved in using the Likert scale （e.g.Busch，1993； Gu，Wen，& Wu，1995），questionnaire design and validation remained a topic rarely touched upon until the end of the 20th century.Even up until this day，Dörnyei and Taguchi （2010； first edition published in 2003）.Only a few papers （e.g.Petrić & Czárl，2003） showcase how a questionnaire can be validated.

With these precious efforts，novice researchers today have much clearer guidance in the process of questionnaire design and validation.However，applied linguists have barely explored the major issues involved in the analysis of questionnaire data.One example would be the analysis of Likert scale data.It is now time for the field to pay more attention not just to what a questionnaire study reveals，but also to how the questionnaire is designed，validated，and analyzed.

Questionnaire research is often seen as ‘quick and dirty’.While the administration of a questionnaire does seem quick，the development，validation，and analysis of the questionnaire are far from a quick and easy process.Another common practice that makes it feel easy is the adoption/adaptation of existing questionnaires.Again，this is not as straightforward as it seems.In commenting on Park’s （2014） recent validation of the FLCAS，Horwitz （2016） emphasized the point that the factor structure of the foreign language anxiety scale may vary from group to group，and that： future empirical efforts to understand the components of FLA [foreign language anxiety] should clearly delineate the specific learner population and learning context being examined to determine if，and if so，how the construct of FLA differs across learning populations and situations.

References

[1]Brown，J.D.Designing surveys for language programs.In D.T.Griffee & D.Nunan （Eds.），Classroom teachers and classroom research[J].Tokyo： JALT，1997： 109-121.

[2] Brown，J.D.Likert items and scales of measurement[J]？ SHIKEN： JALT Testing & Evaluation SIG Newsletter，2011，15：10-14.

[3]Gu，Y.，Wen，Q.，& Wu，D.How often is often： Reference ambiguities of the Likert-scale in language learning strategy research[J].Occasional Papers in English Language Teaching，ERIC Document Reproduction Service No.ED 391 358.1995，5：19-35.

作者简介

郭萍（1981.07—），女，汉族，籍贯：四川隆昌县，讲师，学历：本科；单位：成都艺术职业学院；研究方向：外国语语言与教学。