December 2007
Volume 11, Number 3

Contents  |   TESL-EJ Top

"I'm not agree with you." ESL Learners' Expressions of Disagreement

Karen Kreutel
Chemnitz University of Technology, Germany


In recent years, increasing emphasis has been put on pragmatic competence as part of foreign language curricula. This article analyzes the devices used by learners of English as a Second Language (ESL) in order to perform the speech act of disagreement in their L2. Data from 27 ESL learners were gathered by means of Discourse Completion Tests and compared to baseline data from 27 native speakers of American English. It was found that non-native speakers use mitigational devices such as hedges or explanations less frequently than native speakers, but often resort to undesirable features such as the "blunt opposite" or message abandonment. Moreover, three additional features of target-like disagreement expression were identified, namely, suggestions, exclamations, and a "sandwich pattern of mitigation." The data suggest that high lexico-grammatical proficiency does not necessarily imply high pragmatic competence. The findings are discussed in light of Wolfson's Bulge Theory, and teaching implications are discussed.


With the advent of the communicative approach to language teaching, the concept of communicative competence gained increasing attention from both language teachers and language acquisition researchers. This change in perspective entails, among other factors, a widening of the notion of language proficiency to include not only the traditional areas of grammar, lexis and phonology, but also pragmatic competence, that is, the ability to use language according to the cultural norms of the target language society. More and more evidence demonstrates that a lack of pragmatic skills can lead to misunderstandings and communication breakdowns, despite otherwise high levels of L2 proficiency (Kasper & Blum-Kulka, 1993; Bardovi-Harlig, 2001).

For most learners of a foreign or second language, however, gaining pragmatic competence in the target language is challenging. This is due in part to the fact that "pragmatic competence cannot be clearly judged as correct or incorrect according to prescriptive rules" (Nakajima, 1997, p. 50), which may also be one of the reasons pragmatic issues receive relatively little attention in the language classroom. Yet, its importance for successful L2 use has been emphasized more frequently in recent years, and a growing number of researchers and instructors are arriving at the conclusion that "besides acquiring elements of the target language, students must be able to function within the total meaning system of that language" (LoCastro, 1986, p. 5).

Within the field of pragmatics, which analyzes "how to do things with words" (Austin, 1962), most human utterances are regarded as actions fulfilling certain functions, and are therefore commonly referred to as speech acts (Thomas, 1995). With regard to the English language, some speech acts have received abundant scholarly attention (for example, requests), while others have been more or less neglected by empirical research. Oftentimes, this is also reflected in the availability of materials for English language instruction. This article investigates the speech act of disagreement, which, compared to other speech acts, has been largely ignored in pragmatics research. The paper examines how this "strategy for expressing opinion" (Burdine, 2001, p. 198) is carried out by ESL learners at different proficiency levels and from a variety of cultural and linguistic backgrounds. Their performance is compared to baseline data obtained from native speakers, and implications for English language instruction are derived.

Literature Review

In general, utterances of disagreement are perceived as undesired or dispreferred reactions by native speakers of English (Pomerantz, 1984) and are likely to cause feelings of discomfort (García, 1989). In friendly conversations, which account for a large amount of conversational encounters, speakers show reluctance and hesitance to express disagreement (Beebe & Takahashi, 1989) and "assume that disagreeable acts will not be issued without good reasoning" (Jacobs & Jackson, 1981, p. 122). This observation of a general desire to agree (Kuo, 1994) is confirmed by the fact that in natural data, agreement occurs to a far greater extent than disagreement (Pearson, 1986).

Brown and Levinson's (1987) influential work on the connection between politeness and language use suggests a basic principle that may account for this asymmetry between agreement and disagreement, namely that of face. They define face as "the public self-image that every member wants to claim for himself... [and assume] that all competent adult members of a society have (and know each other to have)" it (1987, p. 61, parentheses in original). What is more, "face is something that is emotionally invested, and that can be lost, maintained or enhanced, and must be constantly attended to in interaction" (1987, p. 61). What can be derived from these assumptions is the existence of a basic human need to save one's own as well as any other person's face, which results in a tendency to avoid face-threatening acts (FTAs). The authors distinguish further between negative face – "the basic claim to territories, personal preserves, rights to non-distraction—i.e., to freedom of action and freedom form imposition" (1987, p. 61)—and positive face—"the positive consistent self-image or 'personality' (crucially including the desire that this self-image be appreciated and approved of) claimed by interactants" (1987, p. 61, parentheses in original). Accordingly, strategies to save negative face are referred to as negative politeness, whereas strategies meant to save positive face are called positive politeness.

Applying Brown & Levinson's (1987) definitions and assumptions to the specific case of disagreement, we find that this speech act is most likely to constitute a threat to the hearer's (or recipient's) positive face as disagreement usually questions the recipient's competence or even truthfulness and thus damages his or her self-image. For this reason, agreement will be preferred over the FTA of disagreement whenever possible.

These considerations are, however, by no means intended to reduce the importance of the speech act of disagreement. On the contrary, disagreements are an essential part of everyday life, and the literature reports numerous situations in which it is more important to defend one's standpoint than to save face (Kotthoff, 1993; Locher, 2004) or where disagreement is in fact the preferred reaction (Pomerantz, 1984; Kotthoff, 1993). The above-mentioned reflections on the face-threatening nature of disagreement illustrate, however, the delicate nature of this speech act, and therefore help us understand why strategies expressing disagreement are much more complex than those expressing agreement (LoCastro, 1986; Hayashi, 1996). In any case, speakers have to bridge the gap between their desire to express their opinion and the presumption that this will be an undesired action. As is shown in more detail below, native speakers resort to a number of politeness strategies to reduce the potential face-threat of their speech act.

The tendency to agree and save the hearer's positive face by "claim[ing] common ground" (Brown & Levinson, 1987, p. 103) is also reflected in Leech's (1983) Agreement Maxim, which he formulated in the frame of his Politeness Principle: "(a) Minimize disagreement between self and other, and (b) Maximize agreement between self and other" (Leech, 1983, p. 132). I will now turn to empirical findings on how native speakers (NS) of English (subconsciously) fulfill this maxim before I present the results of previous studies on non-native speakers' (NNS) expression of disagreement.

Apart from the finding that in friendly conversations not aimed at dispute, native speakers rarely use the performative I disagree (Pearson, 1986; Beebe & Takahashi, 1989; Burdine, 2001), their strategies for expression of disagreement are generally characterized by mitigation, that is, by means to reduce the directness of the disagreement and with it the strength of the FTA. In Brown and Levinson's (1987) terms, this mean that native speakers often choose to perform the FTA off-record, that is, indirectly, as opposed to the direct way referred to as on-record. Other terms for mitigation found in the literature are softeners (Pearson, 1986; Beebe & Takahashi, 1989) and redress (Brown & Levinson, 1987; Hayashi, 1996; Bell, 1998).

The most frequently observed strategy of disagreement mitigation used by NS is that of partial or token agreement, where the speaker starts out by "agree[ing] with the prior speaker's position" (Pomerantz, 1984, p. 72) before voicing disagreement. In most cases, token agreement takes on the yes, but… form (LoCastro, 1986) and shows that the speakers are "responding to the preference struc­ture of the discourse as well as to the specific prior assessment with which they are disagreeing" (Mulkay, 1985, p. 207). Native speakers' use of token agreement has been empirically corroborated by Pomerantz (1984), Pearson (1986), LoCastro (1986), Kothoff (1993), Kuo (1996), Burdine (2001) and Locher (2004), among others.

Other surface realizations of mitigation are the use of modal verbs (Locher, 2004) and hedges (LoCastro, 1986; Locher, 2004), also called uncertainty markers (Pearson, 1986), reluctancy markers (Kotthoff, 1993), disclaimers (Hayashi, 1996), prefaces (Kuo, 1994) or modality markers (García, 1989). According to Aijmer (1986, p. 6), "the hedge frees the speaker from the responsibility for the word and saves him the trouble of finding a 'better' word or phrase,… [thereby] "soften[ing] the impact of negative statements" (Tannen, 1993, p. 28). Frequently used hedges are well, just, I think and I don't know (Locher, 2004). In addition, hedging is often realized on the suprasegmental level by hesitating or pausing (Pomerantz, 1984; Kuo, 1994; Bell, 1998; Locher, 2004).

Closely related to the strategy of hesitation are requests for clarification and repetition of the prior speaker's words (Pomerantz, 1984; Bell, 1998; Burdine, 2001; Locher, 2004). These strategies help the person disagreeing to "buy time" and thus to soften the FTA by its delay.

Giving explanations for their disagreement is a further typical mitigation strategy used by native speakers (Pearson, 1986; García, 1989; Kuo, 1994; Burdine, 2001). What is more, these explanations are often personally or emotionally colored (Beebe & Takahashi, 1989; Nakajima, 1997; Bell, 1998; Locher, 2004). Other mitigating elaborations on disagreements found with native speakers are expressions of regret (LoCastro, 1986) and positive remarks such as compliments, gratitude or signals of cooperation (Beebe & Takahashi, 1989; García, 1989). We can thus conclude from the empirical evidence that native-like strategies of disagreement expression display a high degree of complexity.

Turning to the findings for non-native speakers' expressions of disagreement, we find in most cases a lack of the complexity observed with native speakers. ESL learners' expressions of disagreement were found to be linguistically simple (Bell, 1998) as well as formulaic, short and minimalist (Beebe & Takahashi, 1989; García, 1989; Nakajima, 1997). They are mostly characterized by the absence of the surface features established for native-like language use, which in turn results in an overall lack of mitigation. Therefore, NNSs tend to appear harsh, too direct, or even rude. Apart from the absence of mitigational devices, observed surface characteristics of ESL learners' expressions of disagreement include the bare exclamation no as well as the blunt statement of the opposite, both observed with Koreans by Bell (1998), further, the lack of personalization observed with Japanese by Nakajima (1997), the refusal to cooperate observed with Venezuelans by García (1989) and the on-record realization of the speech act by means of the performative I disagree observed with Japanese by Pearson (1986).

Another important finding reported in the literature is that, possibly due to a lack of linguistic resources (Bardovi-Harlig, 1999), NNS often decide not to express their disagreement at all, even if they would express it in their L1, as Pearson (1986) found in her interviews with 300 Japanese college freshmen. Beebe and Takahashi (1989) replicated this result with Japanese speakers and noticed, in addition, that participants fell back on a questioning strategy typical for their L1 in lieu of any disagreement behavior associated with the L2.

Taking all these findings together, it becomes evident that ESL learners often lack appropriate disagreement strategies, which makes their utterances appear impolite and rude, and which may even result in message abandonment. Observations like these have crucial implications for ESL instruction aimed at empowering students linguistically and at equipping them with all the necessary tools to master everyday communicative situations successfully. This study seeks to add to the existing knowledge in this field and to provide teachers with more detailed information for the successful instruction of communicative competence.

Research Questions

Based on the previous discussion, the following research questions were investigated:

RQ 1: How do NSs and NNSs vary in their use of the following features of disagreement expression reported in the literature?

a) Desirable features (assumed to be native-like):

(A) token agreement

(B) hedges

(C) requests for clarifications

(D) explanations

(E) expressions of regret

(F) positive remarks

b) Undesirable features (associated with NNS):

(G) message abandonment

(H) total lack of mitigation

(I) use of the performative I disagree

(J) use of the performative negation I don't agree [1]

(K) use of the bare exclamation no

(L) blunt statement of the opposite

As proficiency level is commonly assumed to be a potential indicator of pragmatic competence in a second language, the following research question was put forward:

RQ 2: Can we detect a relationship between proficiency level and use of desirable or undesirable features of disagreement expression of the non-native participants?

In order to add to the existing knowledge of native and non-native speakers' disagreement strategies, the following research question requiring a more qualitative analysis was added:

RQ 3: Which, if any, additional characteristics of native and non-native expressions of disagreement can be detected?



Twenty-seven adult native speakers (NS) of American English as well as 27 adult non-native speakers (NNS) participated in the study, totaling a sample size of 54 informants. In order to control for gender influences, both subsamples consisted of 18 female and 9 male informants. Moreover, in order to reduce intervening effects of social background, both groups consisted of college students matriculated at a large state institution in the Midwest region of the United States. The NNS subsample consisted of international students from a total of ten countries representing eight different L1s. They were assigned to ESL classes based on the ACTFL proficiency guidelines. Table 1 and Table 2 show the characteristics of the NNS subsample in terms of country of origin and ACTFL proficiency level:

Table 1. Distribution of NNS According to Country of Origin 

Country #
Japan 6
Turkey 6
Korea 5
Taiwan 3
Thailand 2
Mexico 1
Niger 1
Mali 1
Burkina Faso 1
Latvia 1

Table 2. Distribution of NNS According to Proficiency Level 

Proficiency Level #
Low Advanced 8
High Intermediate 5
Mid Intermediate 4
Low Intermediate 3
High Beginner 7

Data collection

NS as well as NNS language data were elicited by means of Discourse Completion Tests (DCTs). DCTs are questionnaires consisting of written descriptions of selected scenarios. At the end of each scenario description, a conversational turn is provided to which the participants are asked to supply verbatim responses they would give in this situation (Bardovi-Harlig & Hartford, 1993).

Ten scenarios were administered[2]. This number was considered high enough to yield rich data material (as opposed to most of the previous studies on disagreement, which administered only one or two situations) but at the same time low enough not to demotivate the participants. In order to reduce intervening effects from stimulus sequencing and from increasing topic sensitivity, the scenarios were administered as part of an omnibus questionnaire. The scenarios covered a variety of topics and types of situations to avoid intervening effects of topic selection. For each situation, the option "I wouldn't say anything" was provided. Following Nakajima (1997), the option to comment on the situation or on the answer was added to every situation. Moreover, as English was not the participants' native language, the wording of the questionnaire was kept rather simple in order to minimize data distortion caused by incomprehensible input.

All situations in the DCTs are of the type "friendly conversation," that is, situations not aiming at dispute in which both parties wish to maintain a relationship and therefore try to save face. This conversation type was chosen because it has been reported to account for the largest amount of conversational encounters in the target language community (Beebe & Takahashi, 1989), and thus appears to be of highest importance to students and instructors. Accordingly, the "hearer" in each situation is a native speaker belonging to the target culture of American English speakers.


Prior to the actual data collection, the questionnaire was pre-tested in two steps. The first pre-test was carried out with two native speakers in order to ensure that all the items were worded in a clear and unambiguous way. The second pre-test was carried out with two intermediate-level non-native speakers not included in the sample to see if the language was comprehensible for ESL learners. No comprehension problems occurred apart from two questions about unknown vocabulary (landlord and intersection). It was therefore decided that a native speaker would be available during questionnaire administration to provide assistance with vocabulary questions to maximize stimulus comprehension.

Data analysis

The participants' responses were analyzed in three steps. First, invalid responses were discarded and the total number of valid responses was determined. As the study aimed at eliciting actual speech data, that is, actual utterances, metalinguistic responses such as "I would tell him to…" or "I would explain that…" were excluded from the analysis.

In a second step, instances of message abandonment were counted. This number is directly related to RQ 1, but it is also necessary to establish the overall number of disagreement realizations made by the participants as it serves as the basis for the analysis of the disagreement strategies used. A response was considered message abandonment if the participant either chose explicitly to opt out by checking "I wouldn't say anything" or if the participant's statement did not express any disagreement whatsoever.

In step three, the actual disagreement realizations were analyzed in a three-fold way. First, the occurrences of the features in question were counted for NSs and NNSs and subsequently compared. Second, the nonparametric correlation coefficient Kendall's Tau, a measure appropriate for small sample sizes with ordinal-scale data, was established for the NNS data to analyze possible relationships between proficiency level and disagreement realizations. Third, the answers expressing disagreement were examined again in a qualitative fashion for possible answers to RQ 3. Any conspicuous characteristics not covered by the categories listed in RQ 1 were noted and analyzed for reoccurrence in order to detect further patterns of NS and/or NNS disagreement expressions.


Valid responses and overall expression of disagreement

Out of the 270 possible non-native responses (10 situations times 27 participants), four had to be discarded as invalid. Accordingly, the findings are based on a total of 266 NNS responses. As for the native speakers, all of the 270 responses were valid and used for data analysis.

Fifty-three of the 266 NNS responses (19.9 % or one fifth) displayed message abandonment either by silence or by a response that showed agreement instead of disagreement. Accordingly, NNS expressed their disagreement 213 times. The native speakers chose to abandon their message in 32 cases (11.9 %), resulting in 238 actual NS disagreement expressions. Table 3 summarizes these numbers.

Table 3. Valid Responses and Actual Disagreement Expression 

Valid Responses 266 (= 100 %) 270 (= 100 %)
Message Abandonment 53 (19.9 %) 32 (11.9 %)
Responses Expressing Disagreement 213 (80.1 %) 238 (88.1 %)

RQ 1: Distribution of disagreement features reported in the literature

Figure 1 contrasts the distribution of the disagreement features employed by native and non-native speakers, that is, the features found within the 213 NNS and 238 NS responses expressing disagreement. NNSs employed 306 features, whereas NSs employed 400. Solid bars refer to desirable features, whereas dotted bars indicate undesirable features. The bar height indicates the respective feature's contribution to the overall number of features employed (in percentages). It is important to note that message abandonment and total lack of mitigation are excluded from these statistics and will be discussed separately as they refer to a different numerical basis.

Figure 1. Distribution of Desirable and Undesirable Features

As the graph shows, the present study confirms the most general finding from previous studies, namely that NNSs employ desirable features less frequently than NSs (on average, 65 % of NNS features are desirable vs. 84 % for NSs) and display a higher frequency of undesirable features (35 % NNSs vs. 16 % NSs). In fact, the differences between NSs and NNSs show statistical significance on the 0.01 level (p = .002), which indicates that they did not occur randomly. We can thus confirm the finding that NNSs tend to lack mitigational devices and use impolite expressions instead.

Interestingly, however, the general distribution within the two categories of desirable and undesirable features is roughly the same for both NNSs and NSs. In both groups, hedges are the most frequently used desirable device, with explanations ranging second and requests for clarification ranging third. What is more, the ranking of non-desirable features is identical, with the blunt opposite being most and the performative I disagree being least frequently used, while the bare exclamation no ranges second and the performative I don't agree third. Accordingly, even though they overuse undesirable features and tend to lack mitigation devices, the non-native participants seemed to have acquired a general notion of the conversational features associated with disagreement expression.

Surprisingly, the data at hand deviate from previous findings with regard to token agreement, expressions of regret, and use of performatives. The first two are usually associated with native speakers or a native-like language command (Pomerantz, 1984; Mulkay, 1985), whereas the latter is said to be typical of non-native speakers (Pearson, 1986). The NS participants in this study, however, did not use token agreement often; in fact, they used it less frequently than the non-native respondents. Even though the data at hand do not allow for any deduction of reasons for this phenomenon, this result might be attributed to the written mode of data collection. Spontaneous role-plays or other oral activities may possibly have yielded a higher amount of NS token agreement.

Expressions of regret occurred several times with NNS, but not one native speaker made use of this feature. If we take a closer look at the concrete surface realizations of the NNS expressions of regret, however, we find that they consist of I'm sorry in all cases. I'm sorry is said to be generally overused by non-native speakers because it is acquired relatively early and used as a general means of avoiding confrontation by expressing humbleness and deference. Conversely, among native speakers, I'm sorry is usually associated with apologies, that is, the speaker acknowledges a mistake or failure on his or her part. The fact that in the present study this feature is not used whatsoever by the native speakers indicates that this expression of reverence may be inappropriate when it comes to disagreement, indicating that a differing opinion is not necessarily a failure the speaker needs to apologize for. Accordingly, ESL learners should be sensitized that this feature reduces the authority and power behind a statement and may lead to the disagreement not being taken seriously by the listener.

Turning to the results for the performatives, I disagree and I don't agree, we find that the NNSs did not use them as frequently as the current research literature suggested. In fact, I disagree occurred only once and I don't agree only three times in the sample, which cannot be interpreted as frequent usage. Interestingly, though, in two of the three occurrences the performative was not used as a verb, but rather as an adjective, resulting in the sequence "I'm not agree." The question arises whether this insecurity regarding the correct use of the performative expressions contributed to the low frequency of performative use in general.

Two features still remain to be discussed in the frame of RQ 1, namely message abandonment and total lack of mitigation. These characteristics need to be dealt with separately, as message abandonment refers to the overall number of valid answers, and total lack of mitigation is a category on a higher numerical level than the actual surface features employed. Figure 2 shows the NS and NNS percentages for direct comparison. Again, the data confirm that non-native speakers display undesirable features (message abandonment and lack of mitigation) more often than native speakers.

Figure 2. Comparison of Message Abandonment and Lack of Mitigation

With regard to message abandonment, the NNSs used this option almost twice as often as the NSs did, a difference that shows significance on the .05 level (p = .046). Nevertheless, the NS participants chose to opt out in roughly one out of nine cases, a number that cannot be neglected given that the current literature generally views message abandonment as undesirable. NS as well as NNS comments on the questionnaires reveal, however, that in some situations, message abandonment was chosen deliberately, depending on how well they knew the addressee or on whether the expected benefit was high enough to risk the cost of disagreeing and/or hurting someone's feelings. Examples of these comments are: "It depends on how I well I know Josh" (NS), "not worth making a scene" (NS), "He sounds very exciting and proud of himself. I can't say..." (NNS) or "If I say directly, my friend'll hurt his feelings" (NNS).

Striking qualitative differences between NSs and NNSs do occur, however, if one has a closer look at message abandonment as a function of the situation (see Appendix A). An examination of the data shows that NS message abandonment generally occurred in situations 1 (a friend is about to buy an ugly sweater), 2 (correcting a classmate on the number of US states) and 6 (a friend wants to cook a meal for his girlfriend's parents that is not very delicious). These are situations in which nothing is at stake for the speaker and disagreeing might hurt the recipient's feelings needlessly. This finding may appear strange at first; it does, however, reveal a certain pattern when considered in the light of an assumption about the speech behavior of native English-speaking Americans called Bulge Theory. This theory was formulated by Wolfson (1990) based on her own--as well as other researchers'--findings on the speech acts of complimenting, inviting, expressing gratitude, parting, and expressing disapproval. Repeatedly, she found qualitative differences between the speech behavior directed to "intimate, status unequals and strangers on the one hand, and to non-intimate, status-equal friends, co-workers, and acquaintances on the other" (Wolfson, 1990, p. 74). She goes on to explain:

I call this theory the bulge, because of the way frequencies of certain types of speech behavior plot out on a diagram, with the two extremes showing very similar pictures as opposed to the middle section, which displays a characteristic bulge.… We find again and again that that the two extremes of social distance–minimum and maximum–seem to call forth very similar behavior, while relationships which are more toward the center show marked differences. (1990, p. 74)

What we find in the bulge, then, are those speech act realizations that fall under the category of preferred reactions and/or are characterized by face-saving devices. Compliments, expressions of gratitude, hedging and other mitigation devices as well as turns that leave room for negation occurred most frequently in the middle of the continuum. Wolfson attributes this phenomenon to the fact that the relationships between status-equals have the potential for a reduction of the existing social distance since the status of the relationship is relatively uncertain and therefore open to redefinition:

Put in other terms, the more status and social distance are seen as fixed, the easier it is for speakers to know what to expect of one another. In a complex urban society in which speakers may belong to a variety of non-overlapping networks, relationships among speakers are often uncertain. On the other hand, these relationships are dynamic, and open to negotiation. There is freedom here but not security. The emergent and relatively uncertain nature of such relationships is reflected in the care people take to signal solidarity… and to avoid confrontation. (1990, p. 74)

Situations 1, 2, and 6 are likely to belong to the bulge, in contrast to, for instance, the situations involving student-teacher interactions. By abandoning their message of disagreement, the speakers decided to refrain from a dispreferred reaction, thus showing the care and the attempt at demonstrating solidarity mentioned above. Accordingly, message abandonment cannot necessarily be classified as undesirable feature as in some situations it shows more tact and face-saving consideration than any type of mitigation could.

With the NNSs' responses, on the other hand, we see that their message abandonment does not display any patterns like the one just described. Although situations 1, 2, and 6 receive high rates of message abandonment as well, NNSs also decided to "swallow" their disagreement in situations where much is at stake and may result in, for instance, additional cleaning duty (situation 4), paying the rent twice (situation 7), or getting a low grade (situation 10). In situations like these, message abandonment is highly dysfunctional, and it appears that NNSs run the risk of being taken advantage of. The comments of one NS informant illustrate this clearly: "I will speak up when it comes to something like money" (situation 7), and "Again, I have no problem speaking my mind when it comes to something this important" (situation 10). Accordingly, it is settings like these that ESL teachers need to prepare their students for, and equip them with the necessary linguistic tools as well as sufficient confidence to speak up and successfully defend their viewpoints.

Finally, the frequencies for unmitigated disagreement, the last characteristic examined in RQ 1, are also significantly higher for NNSs than for NSs (p = .018). Considering the instances of actual disagreement expression, that is, all valid responses excluding message abandonment, the native speakers displayed unmitigated disagreement in only 9.8 % of the cases, whereas this characteristic was found in 20.8 % of the non-native speakers' disagreement realizations. The fact that the two groups also differed systematically with respect to this feature indicates and stresses the need for focused and adequate instruction to ESL learners all the more.

Similar to what was said earlier about message abandonment, it is important to note here that native speakers do choose to use unmitigated disagreement in some instances. Interestingly, the situations in which NS decided not to use mitigation are identical to the ones in which they decide not to not abandon their message, that is, situations 4, 7, and 10. This finding, too, is in line with the assumptions of the Bulge Theory. In these situations, the status constellation is rather clearly defined. Consequently, the speakers do not need to soften or abandon their message for the sake of improving their social relationships and can therefore be more direct.

A second factor seems to play a role in the degree of mitigation used by native speakers, namely that of urgency. In general, it can be said that the less time is available and/or the more is at stake in a given situation, the more direct and blunt the answer will be. Situation 8 (telling the car driver where to turn) is interesting in this respect. Here, the degree of mitigation (or, alternatively, bluntness) seems directly related to the urgency of the situation. The answers indicate that the participants differed in their interpretation of how fast they were approaching the intersection. Apparently, the more time they assumed they had left, the more elaborate and mitigated their answers were. One of the participants even offered two response versions–an urgent, blunt one (comment: "[if it's coming up fast]") and a more relaxed one with more mitigation devices. Thus, it needs to be kept in mind that under certain circumstances, it is adequate to disagree in a blunt, unmitigated manner.

RQ 2: Variation according to proficiency level

The analysis of the NNS data revealed neither strong nor significant correlations between proficiency and the features discussed so far. Even after grouping the results according to the two categories "desirable" and "undesirable features," [3] no strong correlations could be detected. Hence, even though it seems logical to assume that a higher lexico-grammatical proficiency facilitates pragmatic proficiency (Bardovi-Harlig, 1999), it cannot be assumed that the former automatically gives rise to the latter. Rather, the data at hand confirm the assumption that explicit instruction may be necessary for the acquisition of pragmatic features in the target language and that "learners who are not instructed at all will have difficulty in acquiring appropriate language use patterns" (Kasper & Schmidt, 1996, p. 160).

RQ 3: Additional characteristics of NNS and NS disagreement expression

The qualitative analysis of the data yielded the following three additional, repeatedly-occurring features of disagreement expression. The first of these features is considered desirable, while the remaining two are viewed as undesirable. Each is discussed in more detail below.

Table 4. Additional Characteristics of NS and NNS Disagreement Expressions

Desirable Features Undesirable Features
  • suggestions
  • exclamations of indignation
  • lack of initial mitigation


The data material from the native as well as from the non-native participants contained a conspicuously high number of suggestions. Non-native speakers employed this strategy 49 times, while native speakers used it in 69 cases. Adding the number of suggestions to the overall amount of features used (that is, taking a total of 355 non-native and 469 native features into consideration), we find that suggestions account for a fairly high portion in each subsample, namely, 14.7 % for native speakers and 13.8 % for non-native speakers. In fact, in the NS subsample, suggestions range third after hedges and explanations. The question arises why suggestions have not received any attention in the literature on disagreement expression to date.

In the data at hand, suggestions were used to solve, conceal, or soften the disagreement. Therefore, suggestions either fully substituted or accompanied the disagreement, as illustrated by the following to examples:

Disagreement substitution:

Situation 6 (a friend wants to cook for his girlfriend's parents)—NNS:

"How about go to the restaurant? I know there is a good restaurant."

Situation 5 (museum or beach)—NS:

"It's a beautiful day and it's a great day to go to the beach—let's talk about it!"

Disagreement accompaniment (the suggestions are italicized):

Situation 9 (different opinions on a movie)—NNS:

"Really? I think the actors did a good job. But we can see another movie anytime."

Situation 7 (the landlord says the rent wasn't paid)—NS:

"Oh really? I thought I did. Let me call the bank to see if the check has cleared."

We consider suggestions a desirable feature as all of the suggestions employed by the participants pertain to Leech's Maxims of Tact and Generosity, according to which speakers appear polite when they "minimize benefit to self" and "maximize benefit to others" (Leech, 1983, p. 132). Without exception, all the suggestions brought forward in the sample took the addressee's perspective and interests into consideration and tried to offer a compromise that was in line with both the speaker's opinion (that is, the disagreement) and the hearer's interests, as the following examples (in addition to the ones already mentioned) demonstrate:

Situation 9 (a friend says she did not like the movie)—NNS:

"I am so sorry, maybe I liked the movie because I am rare person, but next time you will choose it."

Situation 6 (a friend wants to cook for his girlfriend's parents)—NS:

"Do you want any help? I am a really good cook and you know chances are that you will be nervous next week and mess up."

In conclusion, it becomes evident that suggestions have a mitigating function and can therefore be considered desirable face-saving strategies of disagreement expression.


Overall, exclamations did not occur very frequently in the data and were used by both native and non-native speakers. Consequently, at first glance the data does not indicate any need to address this feature specifically. Nonetheless, a closer look reveals quite substantial differences between the two subsamples on the qualitative level, namely that many of the NNS exclamations are of an extremely impolite and rude nature and/or are used in inappropriate contexts.

Exclamations were used in nine cases by non-native and in fourteen cases by native speakers. The following table summarizes the surface realizations, how many times each of them occurred in the sample and the situations in which they were used:

Table 5.  Exclamations Used by Native and Non-native Speakers

Sub-sample Surface Realization (no.) Situations (no.)
NNS "What?" (2) # 1—a friend is about to buy an ugly sweater

# 10—the teacher says a paper was not handed in on time

"Oh my god!" (2) # 2—correcting a classmate on the number of US states

# 10—the teacher says a paper was not handed in on time

"Impossible!" (1) # 10—the teacher says a paper was not handed in on time
"Gee!" (1) # 5—museum or beach during vacation with a friend
"Are you crazy?" (2) # 2—correcting a classmate on the number of US states

# 10—the teacher says a paper was not handed in on time

"What the hell!" (1) # 10—the teacher says a paper was not handed in on time
NS "What?" (10) # 1—a friend is about to buy an ugly sweater (2)

# 4—cleaning duty with the roommate (1)

# 10—the teacher says a paper was not handed in on time (7)

"Whoa!" (1) # 1—a friend is about to buy an ugly sweater
"Oh my god!" (1) # 10—the teacher says a paper was not handed in on time
"Bullshit!" (2) # 4—cleaning duty with the roommate (2)

The table illustrates the differences in context and in degree of harshness between native and non-native speakers. Only one of the NS exclamations is definitely rude ("Bullshit!"), but interestingly, it is only used with the roommate, a person that is socially close and usually not of higher power status than the speaker. What is more, one of the two subjects who used this rather strong expression provided the following comment: "to somebody I've been living with for a long time," suggesting that in order to use such impolite language, the relationship between the speakers has to be rather fixed. Similar comments were made in connection with the more moderate exclamation "What?" in situation 1: "b/c we're friends, I'd be very honest" and "The above answer is only if it is a very good friend. Otherwise I wouldn't say anything."

Interestingly, situation 10 yielded the highest number of NS exclamations. On the one hand, as stated earlier, the danger of scoring an 'F' (fail) for a requirement that the student had obviously fulfilled lends a certain urgency to the situation and therefore gives rise to a more direct language use. On the other hand, the relationship between a teacher and a student is usually located at the far end of the intimacy continuum and characterized by a preset hierarchy known to both speakers. Consequently, native speakers used the rather impolite feature of exclamations either with very close (situations 1 and 4) or very distant conversation partners (situation 10), which, again, corresponds with the assumptions of the Bulge Theory discussed earlier.

The non-native speakers, on the other hand, do not necessarily follow this pattern. In addition to employing exclamations with interlocutors situated at the ends of the continuum, they use these with speakers from the 'bulge,' that is, speakers they do not have a fixed social relationship with, when they correct (and embarrass) a classmate in front of their peers (situation 2) or devalue a friend's suggestion during vacation (situation 5).

The differences between the native and non-native participants become even more obvious when we look at the offensiveness of the exclamations themselves. Those used by non-native speakers are far more derogatory and therefore much more likely to cause a threat to the hearer's face, even if they were uttered with a mitigating tone of voice. What is more, the strongest exclamations are used in the face of the teacher, with whom the social distance is usually fairly high. As we have seen above, students may use relatively direct language when talking to their teacher, but they still have to show respect for the higher status of the addressee. The ESL learners might have wanted to express their surprise and indignation about being treated unfairly, but they did not do so in a way that will solve the problem at hand. Rather, it might upset the teacher and eventually damage the relationship with him or her severely and irreparably.

Lacking of Initial Mitigation

The last qualitative finding deals with the sequencing of mitigation within a response. It was found that many of the responses given by the NNS participants sounded harsh or rude, even though they did contain mitigation devices. A closer examination of these answers revealed that the impression of harshness resulted from the fact that the non-native speakers frequently employed mitigation only at the end of their utterances, but that they started off in a rather blunt and rude way. The two patterns that occurred with relatively high frequencies were:

a) NO + blunt opposite + mitigation

b) blunt opposite + mitigation,

with explanation being the most frequently employed mitigation device, as in:

a) Situation 4—(cleaning duty with roommate):

"No, you clean. I did last time."

b) Situation 5—(museum or beach during vacation with a friend)

"I want go to beach. It's beautiful wheather."

Comparing this finding with the responses from native speakers, we find that in almost all cases, NS sstart their utterance with one or several mitigation devices (for example, "Really? Why? I, actually, thought it was good."—Situation 9). Evidently, what was said about token agreement earlier seems to be valid for disagreement utterances in general: In order to successfully mitigate a disagreement expression, part of the mitigation has to occur prior to the actual disagreement, thus softening the dispreferred reaction in advance. This rule of "mitigation first" was violated by NSs only in situations which were urgent or threatening or where much was at stake for the speaker, as discussed above.

In addition, most of the NSs use mitigation not only at the beginning of their utterances, but also at the end, creating a certain sandwich pattern that "wraps" the dispreferred speech act into preferred reactions. The following examples illustrate this pattern (mitigation devices are italicized):

Situation 2 (correcting a classmate on the number of US states):
"Oh, I thought I read it was only 50. Maybe I'm wrong. I'll have to check again."
Situation 9 (different opinions on a movie):
"Oh, I kinda liked it. Don't know what that says about me though!"
Situation 10 (the teacher claims a paper was not handed in on time):
"There must be some mistake—I did hand it in. I, uh, have the computer disk right here and it shows the date the file was created and the last time it was used, which was when I printed my paper out."

Based on its frequent occurrence in the NS data, this feature is assumed to be quite effective. Also, because of its rather formulaic character, it is supposed to lend itself well to teaching and learning.

Teaching Implications

The main purpose of this article is to add to the general knowledge base of interlanguage pragmatics. Nonetheless, the direct comparison of native speakers and ESL learners also allows a number of practical implications for the language classroom.

First of all, the fact that the non-native speakers in the sample often chose to opt out in situations where they needed to speak up and clearly express their opinion shows that there is a demand for explicit instruction of the speech act of disagreement expression. It should be an essential element of any speaking curriculum as it is crucial for empowering ESL learners to master everyday situations successfully and to not be taken advantage of in real-life encounters outside of the secure classroom environment.

Second, it appears to be beneficial to familiarize students with the concept of the Bulge Theory. This will not only sensitize them for the diversity of social encounters, but also make them realize the need for a wide repertoire of disagreement expressions.

Third, the study revealed the need for successful and appropriate mitigation instruction. A closer look at the NS responses collected in this study reveals a number of recurring linguistic patterns for token agreement, hedges, and requests for clarification that should not be too difficult to teach and/or to learn and that cover the largest portion of devices used by native speakers. Table 6 summarizes these devices and indicates their frequency of occurrence in the data, hopefully encouraging instructors to turn their attention to these features. Although token agreement was not used very frequently, it was included here for two reasons: First, even if its importance could not be corroborated in the present study, it is cited as the most distinguishing feature of NS disagreement expressions in the current literature. Second, it is relatively easy to teach and to learn because of its rather formulaic character.

Table 6. NS Surface Realizations of Mitigated Disagreement

Mitigation Strategy Surface Realization Frequency of Occurrence
token agreement "I thought the same thing, but…"

"It's alright / okay, but…"

"I like your idea, but…"

"That sounds good / like a nice idea, but…"


161 instances)

"I think" / "I thought"



"I'm (pretty) sure / positive / certain"


57 out of 161 = 35.4 %

28 out of 161 = 17.4 %

23 out of 161 = 14.3 %

18 out of 161 = 11.2 %

16 out of 161 =   9.9 %

TOTAL:  88.2 %

requests for clarification

(total: 66 instances)


"Are you sure?"

29 out of 66 = 43.9 %

10 out of 66 = 15.2 %

TOTAL:  59.1 %

Explanations and positive remarks, on the other hand, require considerably more elaboration from the speaker. Yet, this does not mean that these features cannot be taught or learned effectively. Raising student awareness and providing ample opportunities to practice should help most learners significantly. In fact, teaching positive remarks and token agreement could be fruitfully combined, as token agreement by definition starts with a positive statement before the actual disagreement is uttered. With careful guidance and structured practice, students should soon be able to produce relevant and fitting remarks spontaneously--for both token agreement and explanations. After all, the high frequency of suggestions used by the ESL learners indicates that creative elaboration on a given situation is accomplishable even by lower-level students, and providing explanations is in itself a speech act necessary to fulfill a number of additional communicative purposes.

Finally, the non-native participants displayed a certain degree of familiarity with hedges, token agreement, and suggestions to soften or solve the disagreement. What they need to be explicitly taught, however, is to use these (and any other) mitigation devices at the beginning of their utterances, not only at the end, in order to soften the dispreferred speech act in advance. Non-native speakers will be perceived as much more polite and native-like if they try to apply this pattern, and will ultimately be more successful in communicating their intended messages.


Overall, the study confirmed previous empirical findings. Non-native speakers used desirable, mitigating features significantly less frequently than native speakers, which is particularly true for hedging, clarification requests, and positive remarks. In addition, NNSs showed a significantly higher usage of undesirable rude and impolite features, in particular the bare exclamation no, the blunt opposite, and utterances lacking mitigation completely. Moreover, non-native speakers opted out of the speech act of disagreement significantly more often than native speakers. While message abandonment was found to be appropriate in certain settings, it turned out that NNSs also refrained from uttering their viewpoints in situations where silence could have resulted in severe disadvantages for the speaker.

In contrast to previous investigations, however, this study detected relatively few instances of NS token agreement, and expressions of regret were not used at all by the NS informants. Moreover, it was found that the non-native informants did not display the high-frequency use of performatives reported in other studies.

Furthermore, the NNSs expressed regret solely by means of I'm sorry, a feature which is usually associated with an apology on the speaker's part. It implies failure and a morally weaker position and was thus not considered an adequate means for disagreement expression. We are suggesting that the status of expressions of regret as a desirable feature of disagreement expression may thus have to be re-evaluated.

No relationship between pragmatic skills and proficiency level could be found. This confirms the assumption that pragmatic competence is not automatically linked to proficiency in the grammatical and lexical spheres and thus stresses the importance of explicit pragmatic instruction.

Three additional features of disagreement expression were discovered in the present study, specifically suggestions, exclamations, and sequencing of mitigation. Both NSs and NNSs displayed a high number of face-saving, mitigating suggestions. In fact, suggestions partially or completely substituted disagreement utterances and turned out to be a vital element of polite American English disagreement expression. The question arises why such a significant component has not been reported about in the research literature to date.

Exclamations were used by both groups, but with considerable qualitative differences. Non-native speakers used harsher and more face-threatening exclamations; they also used them in conversations with status-higher and socially more distant people (in the data, teacher and landlord), that is, in encounters in which native speakers did not use them.

Finally, the data at hand revealed a general pattern for native-like mitigation sequencing. NNSs were found to use their mitigation devices mostly at the end of an utterance, creating a rude impression even though mitigation was employed. This stands in a stark contrast to the NS responses, which typically started with one or more mitigation devices before the actual disagreement was uttered. What is more, NSs tended to use mitigation at the beginning and at the end of their responses, displaying a sandwich pattern of mitigation.

Even though the findings presented here are mostly in line with previous research and seem fairly unambiguous, some methodological limitations of the study need to be noted. Most importantly, the use of Discourse Completion Tests has been repeatedly questioned as a source of authentic language material. Linguistic research is admittedly facing a genuine dilemma here: Although we are aware of the differences between oracy and literacy, we use the written mode to elicit speech act data that we assume to represent spontaneous reactions in the oral mode. It is therefore a vital argument that the study might have yielded different results if naturally occurring conversations or oral activities such as role-plays, video recordings or interviews with the informants had been used. Accordingly, research employing these means of data collection is highly encouraged.

Moreover, the study did not deliberately manipulate the stimuli, that is, the situations in the questionnaire, according to situational and social variables. The qualitative analysis, however, revealed a considerable influence of factors such as status and social distance on the subjects' answer choices, in particular with native speakers. Accordingly, it is assumed that a study design that systematically varies these stimuli from the start may yield more illuminating insights and lead to an even more profound picture of the speech act at hand.

Despite these methodological shortcomings, however, the study yielded important results and contributed to a better understanding of descriptive as well as pedagogical aspects of the speech act disagreement. Basic findings from previous research could be confirmed and expanded (NNS often lack the devices typical of NS realizations of disagreement, for example, mitigation or explanations and use less successful devices instead, for example, the blunt opposite or message abandonment), and some entirely new insights were gained (for example, use of suggestions or the sandwich pattern of mitigation). Overall, it is hoped that this study has been able to raise both researchers' and instructors' awareness of this area of English language acquisition and to provide practical and applicable input for the classroom. After all, teaching communicative competence should not stop at the threshold of unpopular and unfavorable speech acts; rather, it should aim at enabling students to function confidently and successfully in all types of situations they may possibly encounter.

About the Author

Karen Kreutel, formerly with Kent State University in Ohio, is now in the Department of English, Chemnitz University of Technology (Technische Universität Chemnitz), Chemnitz, Germany. Correspondence concerning this article should be addressed to Karen Kreutel, Department of English, Institut für Anglistik/Amerikanistik, TU Chemnitz, 09107 Chemnitz, Germany.


I would like to thank Jessie Carduner of Kent State University for her helpful comments on an earlier draft of this paper.


[1] Even though the negative performative I don't agree did not occur in the literature, it was considered a logical option and therefore added to the list of possible features.

[2] The questionnaire is reproduced in the Appendix.

[3] The variable 'expressions of regret' was excluded here, based on its debatable status discussed earlier.


Aijmer, K. (1986). Discourse variation and hedging. In J. Aarts, & W. Meijsm (Eds.), Corpus Linguistics II. New studies in the analysis and exploitation of computer corpora (pp. 1-18). Amsterdam: Rodopi.

Austin, J. L. (1962). How to do things with words. Oxford: Oxford University Press.

Bardovi-Harlig, K. (1999). Exploring the Interlanguage of Interlanguage Pragmatics: A Research Agenda for Acquisitional Pragmatics. Language Learning, 49.4, 677-713.

Bardovi-Harlig, K. (2001). Evaluating the empirical evidence. Grounds for instruction in pragmatics? In K. R. Rose, & G. Kasper (Eds.), Pragmatics in language teaching (pp. 13-32). New York: Cambridge University Press.

Bardovi-Harlig, K., & Hartford, B. S. (1993 April). Refining the DCT: Comparing open questionnaires and dialogue completion tasks. Paper presented at the 6th Annual Meeting of the International Conference on Pragmatics and Language Learning, Urbana, IL, USA.

Beebe, L. M., & Takahashi, T. (1989). Sociolinguistic variation in face-threatening speech acts. Chastisement and disagreement. In M. R. Eisenstein (Ed.), The dynamic interlanguage: Empirical studies in second language variation (pp. 199-218). New York: Plenum Press.

Bell, N. (1998). Politeness in the Speech of Korean ESL Learners. Working Papers in Educational Linguistics, 14.1, 25-47.

Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in language use. Cambridge: Cambridge University Press.

Burdine, S. (2001). The lexical phrase as pedagogical tool: Teaching disagreement strategies in ESL. In R. C. Simpson, & J. M. Swales (Eds.), Corpus Linguistics in North America: Selections from the 1999 Symposium (pp. 195-210). Ann Arbor: University of Michigan Press.

García, C. (1989). Disagreeing and requesting by Americans and Venezuelans. Linguistics and Education, 1.3, 299-322.

Hayashi, T. (1996). Politeness in conflict management: A conversation analysis of dispreferred message from a cognitive perspective. Journal of Pragmatics, 25.2, 227-55.

Jacobs, S., & Jackson, S. (1981). Argument as a natural category: The routine grounds for arguing in conversation. Western Journal of Speech Communication, 45.2, 118-32.

Kasper, G., & Blum-Kulka, S. (1993). Interlanguage pragmatics: An introduction. In G. Kasper & S. Blum-Kulka (Eds.), Interlanguage pragmatics (pp. 3-17). New York: Oxford University Press.

Kasper, G., & Schmidt, R. (1996). Developmental issues in interlanguage pragmatics. Studies in Second Language Acquisition, 18.2, 149-169.

Kotthoff, H. (1993). Disagreement and concession in disputes: On the context sensitivity of preference structures. Language in Society, 22.2, 193-216.

Kuo, S. (1994). Agreement and disagreement strategies in a radio conversation. Research on Language and Social Interaction, 27.2, 95-121.

Leech, G. N. (1983). Principles of pragmatics. New York: Longman.

LoCastro, V. (1986 November). Yes, I agree with you, but…: Agreement and disagreement in Japanese and American English. Paper presented at the Japan Association Language Teachers' International Conference on Language Teaching and Learning, Hamamatsu, Japan.

Locher, M. A. (2004). Power and politeness in action. Disagreements in oral communication. Berlin: Mouton de Gruyter.

Mulkay, M. (1985). Agreement and disagreement in conversations and letters. Text, 5.3, 201-28.

Nakajima, Y. (1997). Politeness strategies in the workplace: Which experiences help Japanese businessmen acquire American English native-like strategies? Working Papers in Educational Linguistics, 13.1, 49-69.

Pearson, E. (1986). Agreement/disagreement: An example of results of discourse analysis applied to the oral english classroom. International Review of Applied Linguistics, 74.1, 47-61.

Pomerantz, A. (1984). Agreeing and disagreeing with assessments: some features of preferred/dispreferred turn shapes. In J. M. Atkinson, & J. Heritage (Eds.), Structures of social action. Studies in conversation analysis (pp. 57-101). Cambridge: Cambridge University Press.

Tannen, D. (1993). "What's in a frame? Surface evidence for underlying expectations." In D. Tannen (Ed.), Framing in Discourse (pp. 14-56). Oxford: Oxford University Press.

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

Wolfson, N. (1990). The bulge: A theory of speech behavior and social distance. Penn Working Papers in Educational Linguistics, 2.1, 55-83.


Comment/view comments on this article. There are currently comments.

© Copyright rests with authors. Please cite TESL-EJ appropriately.

Editor's Note: The HTML version contains no page numbers. Please use the PDF version of this article for citations.