Linguistic competence and cross-cultural negotiations : A Mexican-Norwegian example

In the present paper we look at the connection between level of linguistic competence and negotiation outcome in cross-cultural negotiations. The corpus consists of six Mexican-Norwegian dyads participating in a negotiation simulation. The language of communication was English, which means that none of the participants negotiated in their native language. The negotiation transcripts were submitted to a qualitative linguistic analysis and graded according to level of competence. These results were coupled to negotiation outcome, which was calculated according to a cost matrix. The findings showed no statistically significant connection between language competence and outcome.


General
According to Bazerman and Carrol (1987:248) "Negotiation is the process by which two or more interdependent parties who do not have identical preferences across decision alternatives make joint decisions".Its main dilemma is the creation and claiming of value as "both parties cooperate to reach individual goals while they simultaneously compete for divergent interests" (Putnam, 1989, cited in Chatman et al. 1991:142).Negotiation is thus a special kind of social interaction requiring its own set of rules and procedures which distinguishes it from other types of social interaction.fundamental constraints on international business negotiations: "political and legal pluralism, international monetary factors, the role of governments and bureaucracies, instability and sudden change, ideological diversity, and cultural differences" (522).The impact of cultural differences on the negotiation process has also been treated by e.g.Graham and Sano (1989) and Le Poole (1989), dealing with Japanese/American and European/American relationships respectively.In a Scandinavian context, the work of Fant (1989Fant ( , 1993) ) compares the negotiation behaviour of Scandinavians, Mexicans and Spaniards, without, however, including negotiations across nationality barriers.This is the focus of Adler and Graham (1989), who discuss "whether intra-cultural behavior accurately predicts cross-cultural behaviour" (1989:515).
Among research focusing on the linguistic aspect, we have the work of Bülow-Møller (1992) concerning Danish-Danish, American-American and Danish-American negotiations, and that of Trosborg (1989), which postulates a set of assumptions concerning the problems facing the non-native in cross-cultural negotiations.

Communication research on cross-cultural negotiation
Communication research on negotiations does by its very nature cut across a number of disciplines, and allows a wide number of approaches.In a cross-cultural context, one aspect which deserves to be mentioned is the difference between large and small language communities, i.e. between those who can expect to use their first language in a negotiation setting, and those who have to rely on a second or even third language.While in the American tradition communication scholars have worked alongside social psychologists and anthropologists, the main focus in Scandinavia has been on the linguistic aspect, dealing with the problems facing the non-native speaker in cross-cultural negotiations.American communication research on negotiation differs from the traditional research of social psychologists mainly by its emphasis on communication as more than a mediator between thought and decision-making (cf.Chatman et al., 1991), and by focusing on micro-aspects of speech.In Europe the tradition tends to approach negotiation within the framework of language for specific purposes in comparing two or more different languages, which provides a more "skill-oriented" perspective.
As a tool for analysing the negotiation process communication researchers have developed different coding schemes that seek to tap strategies and tactics.Through the use of these schemes researchers have been able to identify differences e.g. between integrative and distributive communication tactics. 2 Donohue (1984) criticized this research for not having been able to capture process aspects of communication in bargaining.In order to include the "connectedness of utterances" he proposed Sackett's Lag Sequential Analysis (Sackett, 1979).By defining "criterion events" and the following sequential steps (lags) LSA makes in possible to calculate the statistical probability that one event (e.g.asking for information) will be followed by another (e.g.providing information).Although this approach may have a high facevalidity, we are sceptical to such a positivistic way of dealing with complex interaction.The fact that the method was developed to study e.g.mental retardation and mother and child interaction indicates that it is not unproblematic to adapt it to the bargaining situation 3 .A recent negotiation study using LSA, Weingart et al. (op.cit.), yielded inconclusive results.(For a copy of the coding scheme used by Weingart, cf.Appendix 1).
Communication researchers have over the last thirty years identified various dimensions of importance to the bargaining situation.(For a comprehensive review, see Chatman, op.cit.).They have, however, been criticised for focusing too heavily on the negotiation process without linking it to outcome (e.g.(O´Reilly et al., 1987, cited in  2 Integrative communication tactics are verbal behaviour that presumably has a positive effect on problem-solving and on the creation of value in bargaining.Distributive communication tactics, on the other hand, are taken to refer to behaviour that does not seem to promote the creation of value, only the distribution of it, through persuasion and competitive moves. 3Sackett himself (op.cit.)stated that "Not all data are appropriate for sequential analysis.Whether a data set can be analyzed yielding outcomes that actually answer questions under study depends on methods of data sampling and behavioral code definitions."(p.629) "The null hypotheses concerning individual probabilities used in this chapter all assume that the sequential flow of behavor is random.Yet, it is a truism of behavior that almost everything is correlated with everything else, and that behaviors that occur close together in time will be more similar than those occurring far apart.These two "facts" mean that these types of simple null hypotheses will almost always be rejected.Thus, the lag analysis will always show something significant, as will any of the concurrent or pattern sequential methods, when a large number of behaviors re studied.What seems needed is a model-testing rather than a null-hypothesis-testing approach."(p.647).Chatman 1991)).This is also a major problem in cross-cultural research.Previous studies have identified differences in negotiation behaviour across cultures (e.g.Tung (1982), the early work of Graham (1985) and Fant (1989Fant ( , 1992Fant ( , 1993)) without, however, linking their findings to negotiation outcome.Negotiation is a goaloriented activity and the outcome is the measure of its success.It could thus be argued that cross-cultural negotiation research -like other research on negotiations -is mainly interesting in so far as it helps predict outcomes.
In a recent study Graham (forthcoming) could only find minor differences in verbal behaviours across fourteen cultures (including Mexico -a finding which was supported by Natlandsmyr and Rognes) although there were significant differences in negotiation outcomes.This suggests that present communication research on bargaining has not been able to develop tools that help predict outcomes neither in intra-cultural nor in cross-cultural negotiations.
One main problem facing researchers in the field is a lack of empirical corpus.Studies such as Fant (1993, op.cit.)operate with a sample of three negotiations per setting, none of which are crosscultural in the sense that negotiators from different countries meet across the negotiation table, and thus provide data on expected behaviour in a cross-cultural setting without testing the ensuing hypotheses.Graham (forthcoming) provides an impressive listing of 14 countries including differences in outcome and verbal behaviours.The negotiations studied are, however, intra-cultural.Adler & Graham (1989) compare intra-cultural negotiation with cross-cultural negotiations where the participants negotiate in the language of one of the participants, i.e.English.They also discuss the other linguistic options available in cross-cultural settings, including situations where both parties negotiate in a third language, i.e. a lingua franca.Their material, however, does not include such situations.On the basis of their findings, they conclude that negotiating in a language that is foreign to one or both parties leads to a high degree of behavioural adjustment (acculturation), which presumably distinguishes cross-cultural negotiations from intra-cultural negotiations.In this context, linguistic capability is mentioned as a critical individual characteristic.

Purpose of this paper
In the present paper we are interested in the effect on outcome of negotiating in a lingua franca, i.e. when neither party is able to negotiate in their native language.

The significance of linguistic ability
According to Graham (forthcoming), we find the following hierarchy of problems in international business negotiations4 : 1. Language 2. Nonverbal behaviours 3. Values

Thinking and decision-making processes
The problems grow more serious as one goes down the list, because they become more subtle.It is thus easier to pinpoint communication problems due to language problems than those stemming from different patterns of thinking.This should not, however, be taken to imply that the author regards the linguistic aspect as being of little significance.As an American he is a member of a dominant language group, and he laments the little attention paid to second or third language learning in his culture, not least because his experience has demonstrated the advantage which bilingualism lends to the international negotiator, from being able to follow internal discussion among the opposite team -which to monolingual Americans are a source of irritation -to having the information given twice when using interpreters.We may also add that target language is never taught in a social vacuum, but is invariably accompanied by knowledge of the target culture, social phraseology being a case in point.The loss of the monolingual person is thus not limited to the ignorance of the meaning of simple words.

Linguistic options in cross-cultural negotiations
When members of different language communities meet in crosscultural negotiations, they have to decide which language is to be used: 1.They may decide to use the language of one of the participants, which means that one of the parties will be unable to use his or her native language.2. They may use a language which is the first language of neither party, i.e. a lingua franca.3.They may use an interpreter.In choosing among these alternatives one will have to take into account the linguistic ability of the participants, and also the power balance, since speaking another person's language may seem to be making concessions to the other party.

Native vs non-native speaker
In cross-cultural negotiations it will generally be the case that at least one of the parties is obliged to negotiate in his or her second or even third language.It is reasonable to assume that in contexts involving a native and a non-native speaker, the non-native will be at a disadvantage when it comes to expressing a full range of meanings.Trosborg's (1989) paper includes a number of assumptions about problems which may confront the user of a foreign language in negotiations.Most of these problems can be directly related to lack of mastery of a foreign language, which makes it difficult for the nonnative to handle awkward situations, e.g. the "[t]endency for most learners to revert to simplified and often tactless language instinctively when they find themselves in a situation of conflict" (201).In addition to the general problem of avoiding too blunt statements (206-07), nonnatives also have problems in taking the initiative and closing a theme .The non-native may take comfort from the fact that experienced international negotiators can "show openness and tolerance, and the possibility of "softening" the effect of an ill-chosen word or expression through reformulation and/or clarifying questions often exists" (216).And according to Dupont (1988:12) "international negotiators know about these difficulties and act accordingly, showing some openness and tolerance on the matter, especially since it is always possible to soften the effect through clever reformulating or clarifi-cation".Bülow-Møller (1992) takes this a step further, stating that there is "ample anectotal evidence from business people that they consider that they have often been accommodated to a degree that make it almost an advantage to be the NNS (i.e.non-native speaker, our note), especially if extra time was needed for a decision, or a tactical misunderstanding had to be arranged in order to go back on a decision."(13) One would, however, assume that in a less amiable context, where both parties want to get the maximum profit out of a situation, the foreigner is placed at a serious disadvantage.
Basing ourselves on Trosborg (211ff), we will outline the following features characterising foreign speakers of a language, i.e.limited ability with regard to: 1) taking the initiative, e.g.making proposals, offers and counteroffers 2) closing a theme of conversation in a quick, polite and diplomatic way 3) building up a line of argumentation, including the presentation of counter-arguments and elements of persuasion 4) qualifying one's statements, e.g. in signalling acceptance and/or rejection 5) winning and losing arguments without loss of face 6) avoiding threats The features outlined above are chiefly intended to refer to "advanced learners of English" (211).However, the same problems are likely to face non-natives with a poorer command of the language.Dealing with less advanced speakers, we shall also expect to find 7) problems due to poor command of basic vocabulary and syntax 8) inability to use phatic devices such as e.g.social phraseology.

Negotiating in a lingua franca
Both Trosborg and Bülow-Møller deal with contexts involving a native and a non-native speaker.Frequently, however, both parties will have to negotiate in a language which is the first language of neither.This "third" language is used as a lingua franca, which in the majority of cases means English (Crystal 1985:180).Problems relating to communicating in a foreign language will thus affect both parties, and any asymmetry at the linguistic level will be due to different degrees of command of the said language.
Communication between a native and a non-native speaker contains an inherent asymmetry with respect to language command.Discussing English as a lingua franca presents a different situation, where one will find both interlocutors with different levels of second-language command and participants at the same level of proficiency.With the exception of cases where one or both parties have a near-native competence, the communication process will have to do without the subtleties and full range of linguistic devices which belong to the native speaker of a language.Secondly, the absence of a native to provide the "absolute" frame of reference for the language aspect of the communication process there is always the risk of a communication breakdown if the linguistic competence of both non-natives is too weak to clear up misunderstandings.Fant (1989:262-64) sets up a number of assumptions concerning "the over-all organization of dialogue in Hispanic-Scandinavian interaction".These assumptions do however not take into account the problems involved when one or both of the parties are communicating in their second language.The fact that the channel of communication is qualitatively different from a native-native situation has implications for e.g.turn-taking and holding the floor, since limited language ability will presumably make it more difficult to produce very long stretches of speech.Interruptions may have the function of clarifying a point, and interjections signalling attention may be necessary as a means of indicating comprehension.

Linguistic competence and expected outcome
In a negotiation process, language is an essential part of the process through which a conflict of interests is resolved.If we regard level of linguistic ability as important to the outcome of a negotiation situation, a lingua franca negotiation will present us with two different situations: one where both parties have a (near)equal command, and another where one of the parties has a higher command than the other.With respect to the relationship between the parties, one would expect an asymmetry of competence to give the most proficient speaker an advantage, whereas in situations of relatively equal command linguistic ability would feature less prominently.

Situation
The subjects were male students, all but a few at the undergraduate level.A total of 30 Mexicans and 30 Norwegians participated in the negotiations.12 intra-cultural negotiations per country and six crosscultural negotiations were carried out.The Mexican-Mexican and Mexican-Norwegian negotiations took place in Mexico.The subjects were randomly paired into dyads, and none of them had had any formal training in negotiating.All the negotiations were video recorded.
The subjects participated in a negotiation simulation where the integrative potential was defined by a cost matrix from which they could deduce their individual outcome.The matrix construction is such that integrative solutions can only be achieved by giving on one dimension and taking on another.By giving the participants differential preferences on two out of three dimensions we prepared the ground for the give and take process (logrolling) to take place.Similar studies have reported that it could be a problem that optimal solutions are too easily obtainable (e.g.Weingart et al., 1990).The options were thus given in terms of cost instead of profit in line with the findings of Bazerman et al. (Bazerman and Neale, 1985) 5 .The subjects were informed that their bosses had agreed on price and volumes in an earlier meeting, and that their task was to agree on delivery time, product variations (design) and terms of payment.In order to get a binding agreement they would have to agree on all three conditions.They were allowed 15 minutes (including time to ask questions) in private to read the background information, and 25 minutes to negotiate.The cost matrixes of the seller and buyer are given in Appendix 2.

Levels of linguistic competence
In order to discuss language competence as a contributing factor to the negotiation outcome we need a set of notions to establish degrees of proficiency.If we turn to McArthur (1992), we find reference to the following three levels: 1. Beginner."There is often little creative scope, frustration is common....Performance is usually poor at this stage and dominated by the mother tongue."(578) 2. Intermediate."Much of the learner's grasp of syntax is now established, though with gaps and shortcomings.Dependence on translation (spoken or mental) is less compulsive."(l.c.) 3. Advanced.Advanced learners "seek delicate discriminations of meaning, stylistic niceties, subleties of culture and discourse, and greater acquaintance with the language."While they may exhibit "idiosyncrasies and recurring errors, often due to carrying over features of the mother tongue into the target language", many learners "achieve a close approximation to the skills of the native speaker of the target language".(l.c.) These three levels were never intended to present watertight compartments, nor is class membership permanent -"individuals rise imperceptibly from one to the other, and may also slip back" (l.c.).However, since we are looking at recorded pieces of discourse the candidates' linguistic development is of no interest.
The frame of reference provided by the three levels outlined above will need some elaboration to take into account the kind of variation one would expect to find in contexts involving non-native speakers of a language.We shall therefore attach some tentative characteristics to the three levels outlined above.a.A beginner will have problems in speaking in full sentences.For this reason he will produce few complete sentences, and have a limited ability to form complex verb phrases, e.g. using modal auxiliaries.Restricted vocabulary will lead to repetitions, in dialogues also of phrases uttered by interlocutor.We may also expect false starts and hesitation markers, and occasional communication breakdowns or misunderstandings due to poor language command.No or few set idiomatic phrases.b.At the intermediate level we shall expect the ability to produce connected passages of text featuring sentences containing subordinate clauses.Since subordination is a characteristic feature of speech (cf.e.g.Biber 1988:107, 229), its presence alone will not determine level of competence, but needs to be coupled with the number of false starts and incomplete syntactic units.There will be ability to reformulate to get a point across.Morphosyntactic and other grammatical mistakes will occur, but not to the detriment of overall comprehension.There will be a certain command of set idiomatic phrases, e.g. the general formulae for social intercourse.Since this will be a fairly comprehensive group, we will find it convenient to divide the group into lower intermediate, intermediate and upper intermediate, depending on the degree of competence with respect to the criteria outlined for this level.c.The competence of the advanced learner will approach that of the native.We will expect the ability to develop a line of reasoning in comprehensive passages of speech without making grammatical mistakes.Modifiers will be used both at phrase and clause level.There will be few problems in phatic communication and other idiomatic language.Errors will be at the idiomatic and stylistic level, misunderstandings will be due to ignorance of cultural factors.

Language competence coupled with negotiated outcome
Six dyads represent negotiations between one Norwegian and one Mexican, three of them featuring a Norwegian buyer and a Mexican seller, and three where the roles are reversed.If we look at the individual representatives of each language community, we find all the Norwegians in the intermediate language competence category, three lower intermediate, one between lower intermediate and intermediate, and two of them classified as intermediate.The Mexican group includes two beginners, three lower intermediate and one intermediate.The linguistic competence of the Norwegians is thus generally higher than that of the Mexicans, and in none of the dyads did the linguistic competence of the Mexican exceed that of the Norwegian.We do, however, note that none of the participants have a competence that approaches that of a native speaker.In two dyads the Mexican achieved the highest score, in two the Norwegian did, and two ended in an equal score for both parties.If we couple linguistic competence with negotiated result we can set up the following table: 1480 points indicates a perfect integrative (Pareto optimal) joint solution, whereas 1140 is a purely distributive solution (see Appendix 2 for an explanation of the outcomes).
It will be clear from Table 1 that there is no one-to-one correspondence between level of linguistic competence and achieved result.In fact, the correlation between linguistic competence and individual gains (Pearson's r) is -.09.Please note, however, that conventional statistical tests may be inappropriate due to the limited size of the sample.
Average linguistic competence for each dyad was negatively correlated to joint gains (r = -.55).Again, please interpret the numbers with a great deal of caution.
If we turn to our research questions, we find the following results: Research question 1: How does linguistic competence affect individual gain?
As expected, the overall linguistic competence of the Norwegians exceeded that of the Mexicans.However, the differences are not statistically significant (analysis of variance yields p < .14).

Comments
We note that in two cases, NM-1 and NM-3 both parties achieve (near)identical results despite the fact that the linguistic competence of one of In both cases, the language ability of the Mexican is at times insufficient to express his meaning, leading to his sentences occasionally being completed by the Norwegian.We thus see that the most proficient speaker to a certain extent assumes responsibility for the progress of the negotiation process, whichperhaps -may prevent him from maximising his own interests.
The participants in the four remaining dyads are all at a lower intermediate or intermediate level.In the case of NM-2 the most proficient speaker achieves the highest score.In NM-4, NM-5 and NM-6 the language competence is at the same level: NM-4 gives the same score for both, while NM-5 and NM-6 gives the highest score to the Mexican and the Norwegian respectively.

Research question 2: How does linguistic competence affect joint gain?
There appears to be a negative correlation between the linguistic competence of the participants and the joint outcome of the negotiation.Thus, the highest joint score was achieved by NM-4, an evenly matched dyad at lower intermediate level, and by NM-1, which included one beginner and one at the (lower) intermediate level.The most proficient pair, NM-6, only reached fourth place, and the lowest joint score of all, that of NM-2, was achieved by two speakers at the (lower) intermediate level.It must, however, be noted that the small sample and the high degree of variation may give exaggerated importance to one single observation.

Discussion
In a comparative study Natlandsmyr & Rognes (forthcoming) found that Norwegian dyads achieved significantly higher joint gains than did Mexican dyads.An explorative analysis of the progression of offers showed that the Norwegians managed to create a high integrative surplus value after about 6 to 10 exchanges of offers.The Mexicans, however, only managed to redistribute a fixed sum and, on average, achieved low joint gains.
If we couple behavioural differences between Norwegians and Mexicans with linguistic competence we find that the highest linguistic competence combines with a predominantly integrative behaviour, while the lowest linguistic competence combines with predominantly distributive behaviour.Two possible consequences of this may be suggested: Poor linguistic competence may reinforce distributive behaviour, since lack of language command makes it safer to stick to one's guns.On the other hand, the most proficient -and cooperationoriented -speaker, may feel responsible for the negotiation to reach an outcome, and yield ground during the process.This would go some way towards explaining the results achieved by NM-1 and NM3.
Another point concerns the material on which the present discussion is based.While a laboratory situation like the present one should get rid of any myths that language competence and negotiation outcome are in a one-to-one relationship, it also contains a number of limitations.Thus the nature of the situation allows one of the parties to achieve a good score merely by repeating his position, e.g.NM-1 and NM-3.
This strategy would however be impossible in less "closed" situations.Many of the linguistic weaknesses facing the non-native will thus be more in evidence in cases like -building a relationship -creating trust in a firm's competence or product, e.g. when trying to establish oneself in a new market -gaining preference in a situation where the choice is between nearequal products Before concluding, we would like to comment on the limitations of the present study.First, our methodological approach is explorative.The small variations in linguistic competence reported here may be due to the tentative nature of the classification scheme.Secondly, the use of a simulation puts a question-mark against the study's external validity.As pointed out by our Hermes reviewer, Anna Trosborg, simulations are low-risk situations.In real life poor linguistic competence may increase tension and affect the negotiation process in a negative way.Another weakness is the expert vs novice problem, since professional negotiators are usually more self-assertive and aggressive than students.Finally, one should be wary of drawing too strong conclusions on the basis of a total sample of 12 subjects.One of the main challenges of future research will thus be to get access to a substantial amount of cross-cultural data.
We hope, however, to have demonstrated that more work needs to be done on the issues involved in lingua franca negotiations, since studies involving one native speaker represent a qualitatively different situation.When both interlocutors have a limited command of the language, there is nobody to provide the absolute linguistic frame of reference to clear up misunderstandings.As we have seen, this role will in some cases be taken up by the most proficient speaker.On the other hand, lingua franca negotiations may lead to interlocutors being less afraid of asking to clarify a point, since both parties are disadvantaged linguistically.
Appendix 2: Cost matrix buyer and seller Explanation of cost matrixes Buyers and sellers have opposite preferences.However, both have a gross profit for each product of 1140 points (in our case US dollars).
Costs associated with the chosen alternatives are to be deducted from the gross profit.Example: a solution based on the choice of C (delivery time) -D (product variations) -F (financing terms) yields 1140 -(140+130+140) = 730 points to the buyer and 1140 -(170 + 210 +210) = 550 points to the seller.Due to the construction of the cost matrixes, a perfect (Pareto optimal) integrative solution can be found at A -E -I, yielding 740 points to both buyer and seller.

Table 1
Linguistic competence and outcome