Pragmatic Communication Strategies:
An Empirical Study of Compliment Responses
IWAI Chiaki Hiroshima City University
HIGUCHI Shinichi Graduate School, Hiroshima City University
IKEGAMI Masato Graduate School, Hiroshima City University

Back to Paperhome

1. Introduction
  Reviewing the narrowly researched areas in studies of communication strategies (CS), Kasper (1997), Iwai (2000), and Takatsuka (2001) stress the necessity of expanding the areas beyond referential communication problems, including pragmatic problems, in language use. In contrast to constitutive rules of lexicon and grammar, pragmatic rules are regulative (Thomas 1995, p. 109). Acquisition of such pragmatic rules of language use is presumably more difficult for second language (L2) learners, especially for learners of English as a foreign language (EFL), than acquisition of constitutive rules due to their unclear and normative nature.

Despite the difference in such rule systems, L2 learners' lack of pragmatic competence needs to be complemented, in actual language use, by strategic competence just as their inadequate grammatical competence does. In this sense, the authors of the present study consider it suitable to include in CS studies pragmatic problems that L2 learners encounter because such studies would induce valuable pedagogical implications for L2 teaching, especially in EFL contexts such as Japan, where actual use of the target language is severely restricted. In addition, Kasper and Rose (1999) point out that developmental changes in L2 learners' pragmatic competence have not been fully investigated and, thus, this is an urgent issue in interlanguage pragmatics.

The objectives of the present study are twofold. The first objective is to construct a data collection method to examine L2 learners' pragmatic competence from the perspective of CS. The second objective is to analyze the data on compliment responses which were collected with this method and to examine the results in terms of possible insights they may provide to apply the CS perspective to interpragmatic studies.

2. Review of Studies on Compliments and Compliment Responses
  Past pragmatic studies on compliments (CP) and compliment responses (CR) can be categorized roughly into the following three types, which also appear to cover the majority of other interlanguage and cross-cultural pragmatic studies:

1) Studies which have investigated CP and CR in a particular L1 speech community (L1 pragmatic perspective: e.g., Pomerantz 1978, Holmes 1988),
2) studies which have examined cross-cultural differences in CP and CR (Cross-cultural pragmatic perspective: e.g., Barnlund et al. 1985, Herbert 1990, Chen 1993, Nelson et al. 1996), and
3) studies which discuss L2 learners' deficits in pragmatic competence regarding CP and CR and suggest instructional methods (Interlanguage pragmatic perspective: e.g., Holmes & Brown 1987, Nelson et al. 1995).

These studies present insightful views concerning, for instance, what politeness principles govern native English speakers' CP and CR acts (Pomerantz 1978) and how CP and CR acts are realized similarly or differently in different cultures, e.g., between Americans and Japanese (Barnlund et al. 1985), Americans and Chinese (Chen 1993), and Americans and Arabic speakers in Syria (Nelson et al. 1996). However, they do not necessarily answer our fundamental questions regarding how pragmatic competence should be treated or taught in EFL contexts. To be more specific, it seems necessary to inquire into, at least, the following four issues related to EFL learners' pragmatic problems:

1) What pragmatic norms should be incorporated into EFL class activities (see Iwai & Rinnert 2001),
2) how EFL learners' pragmatic problems can be identified (including identification of pragmalinguistic vs. sociopragmatic problems in Thomas' (1983) dichotomy),
3) how EFL learners' pragmatic competence can be judged or measured, and
4) what pragmatic competence should be the goal of EFL instruction (i.e., latent vs. practical capability of language use).

Among these four issues, the present study concentrates on questions 2 and 3, developing concrete empirical data collection methodology and analysis in the following sections.

3. This Study

3.1 Respondents
  In order to investigate how EFL learners express CR strategically, a questionnaire survey was given to 135 Japanese college English learners in four classes at two different coeducational universities. Before proceeding to the main section of the questionnaire, the respondents were requested to answer background information questions, including gender, English proficiency, academic year, and the total length of their stay (referred to as the 'stay' factor hereafter) in an English speaking community or communities.1) Excluding responses from 5 students who left some background questions unanswered, the total number of respondents came to 130, the details of whose background are summarized in Table 1

3.2 Research questions
  The main research questions examined in this study are as follows:

1. Does pragmatic competence of Japanese college EFL learners to produce CR vary according to the differences in their gender, English proficiency, academic year, or stay experience?
2. If Japanese college EFL learners have problems in producing CR, are they primarily pragmatic problems or linguistic problems? How do they solve these problems strategically?

3.3 Questionnaire
  The main section of the questionnaire consisted of three speech act items (gratitude, compliment response, and request refusal) in a discourse completion test (DCT) format, which has been widely used in sociolinguistic and interpragmatic studies. The situational information for each of the DCT items was specified to facilitate the respondents' understanding, and the final DCT questions were determined on the basis of a pilot study. The directions and the situational information were all given, both orally and in a written form, in Japanese to facilitate respondents' understandings and not to provide them with any linguistic hints which could influence their answers to the DCT questions.

Among the three DCT items, this study focuses only on CR, and the following illustrates the complete question for the CR situation (translated from the original Japanese question):

Persons appearing in this situation:
・You   ・A Thai student with whom you became friends while studying abroad

Place: A cafeteria at a British university

Background: You are friends with a Thai student. But you don't know very much about Thai culture.

Situation: While you are drinking tea at a cafeteria, your Thai friend comes to you and compliments you on your jacket. But it is a cheap jacket you bought on sale, and you don't like it very much. How do you respond to your friend's compliment?

Thai student: Oh, what a nice jacket you are wearing! I like it!


The DCT items in this study were unique in terms of the following four points. First, the given situations were culturally neutralized; that is, situations with which the respondents were presumably unfamiliar were chosen so that they had to produce their responses strategically. The second and the most unique point is that the respondents were requested to answer 'what they could say' rather than 'what they would say', the latter of which has been the most common DCT questioning format in past pragmatic studies. In this study, however, the latter method was not adopted since competence data cannot be obtained from a single performative answer to a 'what one would say' question. To allow multiple answers for what they could say, six open spaces for responses were provided on the response sheet.

Third, the respondents were advised to use Japanese, locally or globally, in their responses if they could not express their intended concepts in English. This was done so as to trace the respondents' linguistic problems in answering questions. Despite this carefully worded direction given prior to answering the DCT, however, in practice their use of Japanese was limited to the solutions for local linguistic problems, mainly lexical problems.

Finally, retrospective multiple-choice questions (R-MC) were added to the DCT to investigate psychological processes that were reportedly involved in reaching the CR goal (Appendix 1). The answer sheet was folded in half, and the R-MC questions were hidden while the respondents answered the DCT questions. The respondents unfolded the sheet and answered the R-MC questions immediately after they had completed the DCT questions.

3.4 Data Analysis
  The data encoding procedure was based on a slightly modified version (Table 2) of the semantic formulas (SF) presented in the CR taxonomy by Herbert (1990). First, the three researchers of this study encoded a randomly selected 10% of the entire responses, and over 80% reliability was confirmed. After amending discrepancies among the three researchers, each one of them encoded all responses (346 responses in total, 2.66 responses per individual on average), and the final encoding outcomes were obtained after adjusting minor disagreements among the three raters.

In addition to SF encoding, the following quantitative data for semantic and syntactic features were obtained for each respondent: 1) Total number of non-overlapping semantic formulas (SF score), 2) total T-units, and 3) total error-free T-units.2) (See Appendix 2 for a sample of processed data.) These quantitative data were used to examine the influence of independent variables (gender, academic year, proficiency, and stay) on CR. The interaction of these main effects could have been tested; however, due to the severely imbalanced distribution of the respondents (see Table 1), it was decided to test the effect of each variable separately, as shown in the next section.

4. Results

4.1 Semantic Formulas and T-units
  To begin with, the SF scores were compared for each one of the four independent variables, and the mean scores and statistical results (tested by an independent t-test for gender and stay, and by a one-way ANOVA for proficiency and year) are shown in Table 3. In the same manner, comparisons of T-units and of error-free T-units were made, and their results are summarized in Tables 4 and 5, respectively.

These results indicate clearly that the gender and stay factors were related to SF scores, total T-units, and total error-free T-units. That is, female respondents wrote longer and grammatically more accurate responses with more semantic formulas. The same tendencies were also seen in the responses of the students who had been exposed to English speaking environments. The interaction of these two factors was not tested; however, these results may not be unrelated to the fact that more female respondents in this study had 'stay' experience than male respondents (see Table 1).

The most interesting finding in these three statistical comparisons is that the proficiency factor caused a significant difference neither in SF scores nor in total T-units. In contrast, grammaticality of the responses (error-free T-units) was significantly affected by the proficiency factor, and the results of a post hoc test (Fisher's LSD) revealed that responses in the low proficiency group consisted of significantly fewer error-free T-units than those in the other two groups (p<.01), but there was no difference between these latter two groups (n.s.). In addition, the variable of academic year was found to have no influence at all on the responses. For this reason, this factor is discarded from the next section.

4.2 Retrospective Multiple Choice Questions
  The results of analysis of the R-MC questions are displayed in Table 6. Overall, 452 answers were obtained, and it became clear that the respondents basically attempted to be polite (Item 1, 25.7%), translated Japanese concepts into English (Item 2, 36.7%), or wrote anything they could come up with (Item 5, 20.8%). At the same time, most of these respondents tended to have little confidence in their English responses (Item 3, 8.4%), but they rarely omitted what they wanted to express (Item 4, 8.4%).

These answers for the R-MC questions were compared using chi-square tests in relation to each one of the three factors (gender, proficiency, and stay); however, no significant difference was observed for any one of them. This finding is worth noting since it indicates that the respondents processed their responses in very similar ways although the derived outcomes varied in terms of their SF scores, T-units, and grammaticality according to the respondents' gender, proficiency, and overseas experience.

4.3 Response patterns
  As shown in Table 7, the majority of the CR responses were realized using three or fewer than three semantic formulas, e.g., 'appreciation token' + ('scale down') + ('comment history'), and the cumulative share for such responses was 94.8%.

For this reason, the analysis in this section was carried out on the first one, two, or three semantic segments of all the responses. The results of the analysis conducted in this way are summarized in Table 8, in which the (x) and (y) signs represent any semantic formulas that had or had not appeared preceding or following the central formula.

From this analysis, it was found that almost 60% of CR responses consisted of an appreciation token, e.g., 'thank you' or 'I'm glad you say so', which was followed or preceded by other semantic segments. The other common strategies to express CR were 1) to disagree with the complimenter (30.1%), 2) to challenge complimenter's sincerity (22.3%), 3) to scale down CR (20.2%), and 4) to comment on how the complimentee obtained the complimented item (12.7%). One more interesting pattern is that a sizable number of respondents interpreted the compliment as a request (11.3%), which is reportedly common in some cultures such as Syria (Nelson et al. 1996).

To grasp the influence of the three independent variables on these frequently used semantic formulas, group differences were examined using chi-square tests, and the results are shown in Table 9. Surprisingly, no significant difference was found in any one of them, except for the 'challenging sincerity' formula, e.g., 'Really?' and 'Do you think so?' Furthermore, comparisons of the three major CR categories (acceptance, non-acceptance, and acknowledgement) resulted in almost no significant difference, except for the stay factor on the 'non-acceptance' category, as shown at the right end of Table 9.

5. Discussion and Conclusion
  In contrast to the traditional DCT, the present study examined 'what Japanese college EFL learners could say' to respond to a compliment. Briefly summarizing the results, we can say that the learners who experienced overseas stay (mainly female respondents) can produce longer CR utterances with more semantic formulas, while proficiency alone affects neither the CR length nor the number of semantic formulas. These findings appear to indicate the difficulty of developing learners' pragmatic competence in an EFL context, although accuracy of language production can be improved by raising learners' proficiency.

Furthermore, the most common process involved in producing CR responses was to transfer Japanese concepts to English, but the respondents were unsure of the appropriateness of their responses in most cases (Table 6). When the semantic formulas used were examined one by one under the three major CR acts, the respondents were very similar regardless of their gender, stay, and proficiency features (Table 9).

From these findings, the answer to the first research question of this study is rather negative. That is, EFL learners' pragmatic competence is not differentiated largely by the factors of proficiency or academic year. In addition, the 'stay' experienced can say more than the non-experienced, but this does not necessarily mean that the experienced used different semantic patterns.

The answer for the second research question is still by no means clear. Judging from the R-MC answers, the majority of the respondents presumably either attempted to translate Japanese concepts into English or expressed what they could say in English, but the derived outcomes had many grammatical errors; thus, their problems, on the one hand, were mainly linguistic. On the other hand, many respondents were not sure if their responses were appropriate or not as English expressions, although, as evidenced by the R-MC answers, one fourth of the responses were constructed so as not to threaten the complimenter's face. In this sense, the respondents also appear to have experienced pragmatic difficulties.

Before concluding the discussion of this study, its methodological and theoretical shortcomings need to be mentioned. The first problem is concerned with the stay factor and, thus, is a matter of methodology. Even though the respondents were grouped into two categories according to their stay experience, the actual stay length of our respondents was fairly short (with about 80%of them having stayed less than two months). To investigate the influence of this factor further, it would be necessary to collect data from EFL learners who experienced a longer stay.

Second, a retrospective oral interview would be preferable to the R-MC method to grasp accurately how learners solved their pragmalinguistic problems. With the R-MC format of this study, the number of choices had to be restricted due to practical limitations, which made it possible to draw only a rough sketch of how the respondents processed the given tasks.3) In collecting data from a larger number of respondents, however, this questionnaire format still seems to be a promising way in terms of efficiency and practicality when it is combined with follow-up interviews with a subset of the respondents.

The third methodological shortcoming of the present study is that it dealt with only one fairly simple speech act in one situation. To verify the findings in this study, other speech act data need to be analyzed. The researchers of this study are, in fact, at present processing the responses for the other DCT items of this study.

Finally, one theoretical problem, which may be more serious than the methodological problems, should be mentioned. That is, we feel it necessary to question what it means to observe pragmatic problems from the perspective of communication strategies. The term 'strategy' has been commonly used in pragmatic studies, but it does not seem to include the concept of compensation, which has been the main concern for CS studies because they have focused on how L2 learners make up for their linguistic deficits to facilitate their L2 communication. Some research areas are overlapping between CS studies and pragmatic studies, but they do not seem to share exactly the same perspective on strategic language use. This theoretical problem cannot be solved in this study alone and needs to be discussed further in future studies.

Our deep gratitude goes to Carol Rinnert at Hiroshima City University, who helped us analyze grammatical problems of the DCT responses and revise this paper with many valuable suggestions.

1) The respondents reported the total length of the stay in months; however, for the actual analysis, they were categorized into two groups: those who had experienced stay (Stay group) and those who had not (No stay group). In addition, respondents reported their English proficiency by answering questions in a self-evaluation format, and each respondent's proficiency was determined by averaging an Eiken (administered by the Society of Testing English Proficiency) grade and a TOEIC level, both of which were elicited on a five point scale. The correlation coefficient (tested by a Pearson's correlation) of these two self-evaluated grades was significantly high (r=0.737, p<0.001). Roughly speaking, the advanced group students were beyond the 2nd grade level of Eiken, the intermediate group students between the 2nd grade level and the pre-2nd grade level, and the lower group students below the 3rd grade level. (->Back)

2) Any T-units in which a Japanese word or words were used was counted as one non-error-free T-unit. (->Back)

3) In designing the data collection method, more than 10 items were listed. However, judging from the practicality of implementing the questionnaire survey and the time respondents would need, the total number of items was reduced to 6. (->Back)


Barnlund, D. C., and Araki, S. 1985. International encounters: The management of compliments by Japanese and Americans. Journal of Cross-Cultural Psychology 16, 9-26.

Chen, R. 1993. Responding to compliments: A contrastive study of politeness strategies between American English and Chinese speakers. Journal of Pragmatics 20, 49-75.

Herbert, R. K. 1990. Sex-based differences in compliment behavior. Language in Society 19, 201-224.

Holmes, J. 1988. Compliments and compliment responses in New Zealand English. Anthropological Linguistics 28, 485-508.

Holmes, J., and Brown, D. F. 1987. Teachers and students learning about compliments. TESOL Quarterly 21/3, 523-546.

Iwai, C. 2000. Daini gengo shiyo-ni okeru communication horyaku [Communication strategies in the use of second languages]. Hiroshima, Keisuisha.

Iwai, C., and Rinnert, C. 2001. Cross-cultural comparison of strategic realization of pragmatic competence: Implications for learning world Englishes. Hiroshima Journal of International Studies 7, 155-179.

Kasper, G. 1997. Beyond reference. In G. Kasper and E. Kellerman (eds.), Communication strategies: Psycholinguistic and sociolinguistic perspectives. New York: Longman, pp. 345-360.

Kasper, G., and Rose, K. R. 1999. Pragmatics and SLA. Annual Review of Applied Linguistics 19, 81-104.

Nelson, G. L., El Bakary, W., and Al Batal, M. 1995. Egyptian and American compliments: Focus on second language learners. In S. M. Gass and J. Neu (eds.), Speech acts across cultures: Challenges to communication in a second language. New York: Mouton de Gruyter, pp. 109-128.

Nelson, G. L., Al-Batal M, and Echols, E. 1996. Arabic and English compliment responses: Potential for pragmatic failure. Applied Linguistics 7/4, 411-432.

Pomerantz, A. 1978. Compliment responses: Notes in the cooperation of multiple constraints. In J. Schenke (ed.), Studies in the organization of conversational interaction. New York: Academic Press, pp. 79-112.

Takatsuka, N. 2001. Goyooronteki mondai-o kaihi kaiketsu surutame-no communication horyaku: Goyooronteki horyaku-no bunruiho-no teian [Pragmatic communication strategies to solve pragmatic problems]. CASELE Research Bulletin No. 15, pp. 41-50.

Thomas, J. 1983. Cross-cultural pragmatic failures. Applied Linguistics 4/2, 92-112.

Thomas, J. 1995. Meaning in interaction: An introduction to pragmatics. London: Longman.

Appendix 1: Retrospective Multiple-choice questions (Multiple answers allowed)

1) 相手に失礼にならないよう配慮した。(I tried not to be impolite.)
2) 日本語をそのまま英語にした。(I translated what I thought in Japanese directly into English.)
3) 英語ではこのような表現をすると思った。(I thought this would be appropriate in this situation.)
4) 思った通り英語で言えないので、表現を省略した。(I omitted what I wanted to say due to my difficulty of expressing it in English.)
5) 何でもいいので、言えることを書いた。(I just wrote anything I could say.)
6) その他。(Others: Specify the reason.)

Appendix 2: A sample of processed data

Respondent #1 (Gender-female, Year-1, Proficiency-intermediate, Stay-yes)

/ Thank you. / But I think this jacket don't suit me. /
   1            10/Error 1

/ Thanks. / This jacket is very cheap. / It was good 買い物./
   1          8           5/Error 1

・/ Oh your jacket is good too. /

・/ Do you really like this? / Thank you. /
       9          1

SF score=5 (semantic formulas 1, 7, 8, 9, and 10)
Total T-units=8, Error-free T-units=6

Back to Paperhome