Research has shown that the acquisition of second-language (SL) grammar and pragmatics differs for foreign language (FL) and SL contexts (Bardovi-Harlig & Dörnyei, Reference Bardovi-Harlig and Dörnyei1998; Kasper & Rose, Reference Kasper and Rose2002; Kasper & Schmidt, Reference Kasper and Schmidt1996; Schauer, Reference Schauer2006). However, a clear-cut distinction between the influence of FL and SL environments on the acquisition of grammar and pragmatics is difficult to draw because their impacts are complex and multilayered. For instance, some recent studies that have investigated the issue of pragmatic and grammatical acquisition have found evidence in support of the hypothesis that SL environments foster awareness of pragmatic appropriateness, whereas FL environments focus on grammatical accuracy (e.g., Bardovi-Harlig & Dörnyei; Schauer). However, evidence has also been reported that English FL (EFL) speakers showed a higher sensitivity to pragmatic errors than their English SL (ESL) counterparts (Niezgoda & Röver, Reference Niezgoda, Röver, Rose and Kasper2001).

These somewhat controversial findings in the existing research prompt further investigations about how SL and FL environments influence learners’ acquisition of specific areas of grammar and pragmatics. The present study aimed to examine the influence of SL and FL environments on the employment of tense backshifting in English indirect reported speech. The backshift phenomenon was selected as an area of the English language that provides an interesting cross section between grammar and pragmatics because the choice of whether to backshift is not simply governed by the presence of a past tense verb in the main clause but also by extralinguistic factors such as the time lapse between the actual and reported speech, the validity of the utterance at the time of the report (Bland, Reference Bland2008; Celce-Murcia & Larsen-Freeman, Reference Celce-Murcia and Larsen-Freeman1999; Goodell, Reference Goodell1987; Graver, Reference Graver1971; Standwell, Reference Standwell1985; Thewlis, Reference Thewlis2007), and the intention and attitude of the reporter toward what was said (Declerck & Tanaka, Reference Declerck and Tanaka1996; Riddle, Reference Riddle1986; Stein, Reference Stein1982), among others.

Therefore, the theoretical framework that underlies this study was derived in view of two main aspects of theory and research: (a) the body of literature about the influence of the learning environment—ESL versus EFL—on the acquisition of grammar and pragmatics and (b) the phenomenon of tense backshifting in English indirect reported speech as an insufficiently studied cross section between grammar and pragmatics. Therefore, before presenting the study and its findings, its theoretical and empirical background will be briefly outlined.

STUDIES ABOUT THE INFLUENCE OF THE LEARNING ENVIRONMENT ON GRAMMATICAL AND PRAGMATIC PROFICIENCY

Three fairly recent studies (i.e., Bardovi-Harlig & Dörnyei, Reference Bardovi-Harlig and Dörnyei1998; Niezgoda & Röver, Reference Niezgoda, Röver, Rose and Kasper2001; Schauer, Reference Schauer2006) set out to examine the influence of the acquisition environment—ESL versus EFL—on learners’ grammatical and pragmatic competence. These studies are related not only by their common research agenda but also by the use of the same research tool, originally developed by Bardovi-Harlig and Dörnyei. In fact, the other two studies were replications of Bardovi-Harlig and Dörnyei’s seminal work. Their instrument involved three sets of video-taped speech scenarios: The first set contained embedded pragmatic problems, the second embedded grammatical errors, and the third set served as distractors with no errors. The participants, which included ESL and EFL learners, were asked to judge the appropriateness of the statements (e.g., speech acts of apologies, refusals, requests, and suggestions) and to determine the severity of the perceived errors on a Likert scale.

There were also differences between and among these studies related to participants’ characteristics and modifications in the instrument as well as to the ESL and EFL contexts. Bardovi-Harlig and Dörnyei’s (Reference Bardovi-Harlig and Dörnyei1998) study involved the following samples: 370 Hungarian EFL learners in Hungary, 173 ESL learners in the United States, 25 Hungarian teachers of English, 28 American teachers of English, and 112 Italian elementary school teachers enrolled in an English training program. The inclusion of the teacher samples aimed to examine if potential differences between the ESL and EFL student samples would also be observed for their teacher counterparts.

The statistical analyses of the data showed that both groups of EFL learners consistently noticed and rated grammatical errors as more serious than pragmatic ones. In comparison, the ESL sample revealed a much higher sensitivity to pragmatic problems by accurately identifying and giving them higher priority than grammatical issues. Although to a lesser degree, the trend observed within the student samples was also valid for the teacher samples, as the EFL teachers detected grammatical errors with greater precision and gave them higher importance than pragmatic ones. In contrast, the ESL teachers were more aware of pragmatic violations and considered them of higher importance than grammatical errors. Regarding the effect of length of residence in the United States, the results revealed that the more time a participant had spent in the United States, the higher sensitivity he or she showed to pragmatic inaccuracies.

Bardovi-Harlig and Dörnyei’s (Reference Bardovi-Harlig and Dörnyei1998) findings were in great part supported by the results of Niezgoda and Röver’s (Reference Niezgoda, Röver, Rose and Kasper2001) study set in different ESL and EFL contexts. The ESL sample consisted of international students in Hawai‘i and the EFL sample included university students from an intensive teacher education program in the Czech Republic. Congruent with Bardovi-Harlig and Dörnyei’s conclusions, the Czech participants noticed more grammatical errors and perceived those as more serious than the ESL participants in Hawai‘i. This finding is rather surprising, however, in that the Czech participants also demonstrated higher awareness of pragmatic inaccuracies than the ESL participants. As mentioned in Schauer (Reference Schauer2006), one possible explanation for these observed differences between the findings of Bardovi-Harlig and Dörnyei and of Niezgoda and Röver may be related to the fact that the ESL and EFL samples in the latter study were not precisely matched in terms of their proficiency in English. The ESL sample consisted of international students in Hawai‘i, primarily of Japanese origin, with different backgrounds and levels of English proficiency, whereas the Czech participants were teachers in training with a high level of grammatical and metalinguistic awareness of the structure of English.

The most recent replication of Bardovi-Harlig and Dörnyei’s (Reference Bardovi-Harlig and Dörnyei1998) study, as conducted by Schauer (Reference Schauer2006), aimed to triangulate the participants’ severity ratings with their respective explanations through retrospective interviews. The purpose was to find whether, when participants marked a statement as erroneous, they were actually aware of what the error was (i.e., of a grammatical or pragmatic nature). The participants included two groups of German learners of English, 17 EFL university students in an intensive 3-year translation study program, and 16 ESL students from a British University enrolled in different majors. Additionally, 20 British English speakers, enrolled in undergraduate and graduate courses, served as a control group. Like Bardovi-Harlig and Dörnyei, Schauer also examined the relationship between length of residence spent in the target language country and SL speakers’ pragmatic awareness. To examine this relationship, the ESL participants’ ability to detect pragmatic infelicities was compared at two different time periods through a dependent t test: 1 month after their arrival in England and at the end of their stay (i.e., after 9 months). The results entirely supported those reported in Bardovi-Harlig and Dörnyei in that the EFL participants recognized significantly more grammatical errors than pragmatic problems. This group also assigned higher importance to grammatical errors than to pragmatic ones. In contrast, the ESL sample more closely resembled the native-speaker (NS) responses, which identified more pragmatic errors and also judged them as more important than grammatical errors. The effect of the SL environment yielded similar results to those reported in Bardovi-Harlig and Dörnyei. For instance, by the end of their stay in the United Kingdom, the ESL participants’ ratings of the seriousness of pragmatic errors had increased significantly, to the extent that they outperformed the ratings of NSs.

The effect of length of residence on pragmatic competence has also been examined by a number of other studies. These have concluded that SL speakers who have spent longer periods of time in the SL environment approximated NS appropriateness in the employment of speech acts such as refusals (Félix-Brasdefer, Reference Félix-Brasdefer2004) and requests and apologies (Blum-Kulka & Olshtain, Reference Blum-Kulka and Olshtain1986; Olshtain & Blum-Kulka, Reference Olshtain, Blum-Kulka, Gass and Madden1985), among others. Additionally, length of residence was found to be a stronger factor in pragmatic development than level of proficiency (Félix-Brasdefer).

In light of the findings of Bardovi-Harlig and Dörnyei (Reference Bardovi-Harlig and Dörnyei1998) and Schauer (Reference Schauer2006), it seems that there is strong evidence in support of the claim that SL contexts foster enhanced awareness of pragmatic norms, whereas FL contexts put emphasis on grammatical accuracy. Moreover, in SL environments, pragmatic awareness has been found to be positively correlated with length of residence (Bardovi-Harlig & Dörnyei; Félix-Brasdefer, Reference Félix-Brasdefer2004; Schauer). In contrast, as shown by Niezgoda and Röver’s (Reference Niezgoda, Röver, Rose and Kasper2001) study, there is also some evidence that suggests that pragmatic norms are not necessarily learned only in an ESL environment; EFL settings can also foster pragmatic awareness alongside grammatical competence.

In sum, the common theme of both the current study as well as other contemporary studies revolves around the need for an integrated approach to SL grammar and pragmatics in research and in teaching. The importance of teaching language forms in light of their pragmatic potential has been emphasized by a number of researchers (Akai, Reference Akai2007; Bardovi-Harlig, Reference Bardovi-Harlig1999, Reference Bardovi-Harlig, Rose and Kasper2001; Bardovi-Harlig & Griffin, Reference Bardovi-Harlig and Griffin2005; Bardovi-Harlig & Hartford, Reference Bardovi-Harlig and Hartford1996; Kasper, Reference Kasper, Bardovi-Harlig and Hartford1997; Koike & Pearson, Reference Koike and Pearson2005; Rose, Reference Rose2005; Soler, Reference Soler2005; Takahashi, Reference Takahashi2005). Kasper (Reference Kasper1996, Reference Kasper, Rose and Kasper2001) also reminded language teachers of the need for relevant practice in reinforcing certain pragmatic norms. Otherwise, SL learners develop accuracy in the use of syntactic structures and lexical items but remain unaware of the pragmatic principles that underlie the choice of one word or form over another.

The present study was performed to supplement existing research about the influence of the learning environment on pragmatic and grammatical awareness. In the context of this study, the construct of grammatical and pragmatic proficiency was operationalized through the use or nonuse of tense backshifting in English indirect reported speech; the most pertinent theoretical and pedagogical issues related to it will be briefly covered.

TENSE BACKSHIFTING IN ENGLISH REPORTED SPEECH

Reporting what other people have said is performed frequently in oral and written discourse and thus constitutes an important aspect of a speaker’s communicative skills. Because world languages follow different syntactic and pragmatic norms in constructing indirect reported speech (Celce-Murcia & Larsen-Freeman, Reference Celce-Murcia and Larsen-Freeman1999), this area of language use offers an interesting topic for SLA research. In the context of the English language, the employment of indirect reported speech deserves particular attention because it often involves deictic changes in verb tenses, modal auxiliaries, personal pronouns, adverbials, and word order. As these changes are triggered by pragmatic and semantic factors (Bland, Reference Bland2008; Declerck, Reference Declerck1990, Reference Declerck1999; Goodell, Reference Goodell1987; Riddle, Reference Riddle1986; Thewlis, Reference Thewlis2007), the acquisition of English indirect speech requires both grammatical and pragmatic competence.

A central issue in the construction of English indirect reported speech is the phenomenon of tense backshifting. The term tense backshifting was first introduced by Comrie (Reference Comrie1985, Reference Comrie1986) to refer to the phenomenon of sequence of tenses as previously described by Jespersen (Reference Jespersen1924) and his followers (Hornstein, Reference Hornstein1990; Quirk, Greenbaum, Leech, & Svartvik, Reference Quirk, Greenbaum, Leech and Svartvik1985).Footnote ¹ Comrie defined backshifting as a change in the original tense of the direct speech utterance triggered by a past tense reporting verb in the main clause, where present tenses are changed to their past equivalents, past tenses to their past perfect equivalents, future tenses to their future-in-the-past equivalents, and so forth. Examples (1) and (2) illustrate the backshifting of future simple to future in the past.Footnote ²

1. (1) Mark to you (direct speech): I willbecome a lawyer and I willmake a lot of money.

2. (2) You to Mark (indirect speech): You told us that you wouldbecome a lawyer and wouldmake a lot of money.

It is interesting to note that pedagogical grammars and linguistic theory offer opposing views about the backshift phenomenon. Several English grammar books and textbooks (e.g., Bland, Reference Bland2008; Celce-Murcia & Larsen-Freeman, Reference Celce-Murcia and Larsen-Freeman1998; Haines & Stewart, Reference Haines and Stewart2004; Leech & Svartvik, Reference Leech and Svartvik1975; Norris, Reference Norris2008; Quirk et al., Reference Quirk, Greenbaum, Leech and Svartvik1985; Thewlis, Reference Thewlis2007; Vince & Sunderland, Reference Vince and Sunderland2003) follow Comrie’s definition of tense backshifting as an obligatory structural transformation triggered by a reporting verb in the past tense. They also list two exceptions to the rule.

The first exception to the rule concerns the validity of the utterance at the time of the report: Almost all grammar books and textbooks (e.g., Bland, Reference Bland2008; Celce-Murcia & Larsen-Freeman, Reference Celce-Murcia and Larsen-Freeman1998; Haines & Stewart, Reference Haines and Stewart2004; Leech & Svartvik, Reference Leech and Svartvik1975; Norris, Reference Norris2008; Quirk et al., Reference Quirk, Greenbaum, Leech and Svartvik1985; Thewlis, Reference Thewlis2007; Vince & Sunderland, Reference Vince and Sunderland2003) call attention to the fact that the backshift rule does not usually apply to statements that still hold true at the time of the report. There are two main subcategories: (a) statements that express universal truths (e.g., Our geography teacher saidthat the Earth makesone circle around its axis in 24 hours) or generally accepted values and beliefs (e.g., The man said that murder is a high-degree felony) and (b) immediate reporting. The exception in (b) is particularly emphasized by Goodell (Reference Goodell1987) and in more recent grammar sources (e.g., Bland, Reference Bland2008; Thewlis, Reference Thewlis2007; Vince & Sunderland, Reference Vince and Sunderland2003) and postulates that in reports that happen shortly after the direct speech, the backshift rule is often canceled because of the high likelihood of the original utterance to still be valid.

The second exception to the rule concerns the structure and meaning of the grammatical tense: Native speakers’ tendency of not backshifting past tenses to their past perfect equivalents has been noted by a number of authors (e.g., Bland, Reference Bland2008; Celce-Murcia & Larsen-Freeman, Reference Celce-Murcia and Larsen-Freeman1998; Standwell, Reference Standwell1985; Swan, Reference Swan1980; Thewlis, Reference Thewlis2007; Thomson & Martinet, Reference Thomson and Martinet1980). Two main reasons for avoiding the past perfect have been given: First, backshifting to past perfect may lead to a change in meaning and thus cause confusion about the actual meaning of the message. In the following examples Mary said that she wantedto finish the book last week. versus Mary said that she had wantedto finish the book last week, the second sentence carries a different meaning than the first because it implies that Mary had most probably not finished the book. Second, a shift to past perfect makes the original utterance sound “heavy and clumsy” (Standwell, Reference Standwell1985, p. 150).

Contrary to pedagogical grammar, linguistic theory (e.g., Declerck, Reference Declerck1990, Reference Declerck1999; Riddle, Reference Riddle1978, Reference Riddle1986; Salkie & Reed, Reference Salkie and Reed1997) has proposed arguments against the backshift rule, claiming that if the verb tense in reported speech is backshifted, this happens for pragmatic, semantic, structural, or contextual reasons, not because of a “formal syntactic rule that operates mechanically” (Declerck & Tanaka, Reference Declerck and Tanaka1996, p. 283). For example, Riddle (Reference Riddle1978) considered the sequence of tenses phenomenon a pragmatic tool for conveying the reporter’s rendition of the original utterance as believable or not or the reporter’s involvement with the subject or events of the original utterance. A similar view is expressed by Stein (as cited in Celce-Murcia & Larsen-Freeman, Reference Celce-Murcia and Larsen-Freeman1999), according to whom backshifting tenses conveys impartiality on behalf of the reporter, whereas not backshifting tenses implies that the reporter considers what was said to be true. In this sense, any utterance can be projected as valid by the reporter if the tense of the original is kept unchanged (The mother claimed that her son isinnocent vs. The mother claimed that her son wasinnocent).

It is interesting to note that despite the obvious controversy between linguistic and pedagogical views about tense backshifting, there is a lack of empirical research (e.g., Celce-Murcia & Larsen-Freeman, Reference Celce-Murcia and Larsen-Freeman1999) about the reasons, contexts, and purposes for which backshifting occurs in the speech of NSs. There is also a lack of empirical research (at least to the authors’ knowledge) about the status of backshifting in SLA and the influence of the learning environment—EFL or ESL contexts—on the way backshifting is acquired. In short, the observed deficiency in related empirical research served as the rationale for the present study, which sought to explore how tense backshifting is employed by native, SL, and FL speakers of English and whether there are differences in its use due to the environment of learning English, mainly comparing ESL versus EFL contexts.

ANALYSES AND RESULTS

The data were analyzed through Statistical Package for the Social Sciences (SPSS) for Windows, Version 14.0 (2005). The results are presented in view of the three research questions that guided the current study.

Differences Among EFL, ESL, and NS Participants

To address the first research question, which aimed to compare ESL and EFL speakers’ use of backshifting in relation to NSs, descriptive statistics and a one-way multivariate analysis of variance (MANOVA) were employed.Footnote ⁵ The three verb tenses—simple present, simple past, and future simple—in the two contexts of immediate and delayed report constituted the six dependent variables, whereas the four groups of participants served as the independent variable with four levels.

The MANOVA analysis was preceded by Box’s M-test of equality of covariance matrixes, which revealed that the assumption of equal covariance matrixes was observed at α = .05, Box’s M = 85.231, F(63, 49129.098) = 1.25, p = .085. The omnibus MANOVA, performed at α = .05, revealed that group membership was significantly associated with the overall use or nonuse of backshifting, λ = .101, F(18, 399.29) = 27.78, p < .001, partial η² = .5. To better understand the significant MANOVA effects, the MANOVA analysis was followed by univariate group comparisons on the six dependent variables—namely, the three grammatical tenses in the immediate and delayed report contexts. Prior to this analysis, Levene’s test statistics verified that the assumption of equal variances was observed at α = .05 for five of the dependent variables and at α =. 01 for the future simple delayed report.Footnote ⁶

Because six univariate comparisons were involved, the initial α level of .05 was lowered to .008 (Bonferroni adjustment) to control for the familywise error rate. The results revealed that group membership was significantly associated with how backshifting was employed in all three scenarios in the immediate report context and two of the scenarios in the delayed report context (i.e., simple present and simple past). The only situation in which groups did not show significant differences was for future simple forms in the delayed report context. Univariate results are provided in Table 1.

Table 1.

Tests of between-subject effects

Note

The asterisk *denotes a statistically significant result at α = .008.

The independent variable group included four participant samples, so it was necessary to follow the univariate analyses with a post hoc multiple comparison test to establish specific group differences. The Bonferroni pairwise comparisons performed for this purpose showed that the significant differences were between the EFL groups on the one hand and the ESL and NS groups on the other hand (see Table 2 for group mean scores); that is, both EFL groups employed backshifting significantly more than the NSs and the ESL groups in the same five comparisons that were found significant through the univariate comparisons (p < .001 for all five comparisons). The only speech scenario that did not show significant differences among any of the groups was the one for future simple forms in the delayed report context, NSs-ESL, p = .183; NSs-Bulgarian, p = 1.00; NSs-Bosnian, p = 1.00; Bulgarian-ESL, p = .243; Bulgarian-Bosnian, p = 1.00; ESL-Bosnian, p = .219.

Table 2.

Descriptive statistics for each group’s mean use of tense backshifting

The ESL participants’ use of backshifting was not significantly different from that of the NSs on any of the six speech scenarios; the calculated p values were much higher than the α of .05, future simple delayed report, p = .183; all remaining comparisons, p = .100. In turn, there was no significant difference between the two EFL groups on any of the six dependent variables, as the p values were much higher than the α of .05, simple present immediate, p = 1.00; simple present delayed, p = 1.00; past simple immediate, p = .174; past simple delayed, p = .298; future simple immediate, p = .155; future simple delayed, p = 1.00.

Reasons for Backshifting in Relation to Grammatical Tense

Overall, the statistical analyses showed that the ESL participants used backshifting to a significantly lesser degree than the EFL groups and in a manner similar to the NS control group.

Because the differences and similarities between the groups can be better illustrated in view of the semantics of each of the three tenses in the two time frames of immediate and delayed reports, this section outlines the observed patterns within each of the three grammatical tenses. Entwined with these results are also results for the second research question, which sought to identify the reasons for the use or nonuse of backshifting. This part of the data was analyzed by calculating the percentage of occurrence of each response within each group. Because some participants had selected more than one option, the calculated percentages within groups may exceed the 100% response rate. For the sake of brevity, only the most frequently selected reasons are presented and discussed here.

The simple present scenarios incorporated verbs that denoted ongoing actions and states. The purpose was to find out how the four groups of participants would employ backshifting in the context of immediate and delayed reports. As seen in Table 2 and Figure 1, none of the participant groups distinguished much between the two contexts of immediate and delayed reports with regard to ongoing actions and states. However, this pattern was different for the ESL and EFL groups, as shown by the percentages of participants who never employed backshifting (descriptive statistics can be found in Table 2). The majority of ESL and NS participants chose not to backshift at all in either the immediate or delayed report contexts, whereas almost none of the EFL participants employed zero backshifting. These patterns of differences were shown to be significant by the MANOVA.

Figure 1.

Mean backshift scores per group for simple present tense.

These significant differences were further explained by the participants’ reasons to backshift the simple present or not. For example, the most prevalent reason why the EFL participants chose to backshift simple present forms to simple past, in both the immediate and delayed report contexts, was their conscious observance of the sequence of tenses rule. The results indicated that the majority of the Bulgarian participants chose this reason in both contexts (73% in immediate report and 72% in the delayed report). In a similar manner, 76% of the Bosnian participants chose this reason in the immediate report context and 70% selected this option in the delayed report. In contrast, the NSs and the ESL participants chose not to backshift to simple past mainly because they were trying to convey certainty or belief that what was said in the original utterance was still true and valid (NSs: immediate report, 55%, delayed report, 54%; ESL: immediate report, 51%, delayed report, 49%). Additionally, approximately 50% of the NSs and the ESL group indicated that they wanted to retain the meaning of the original utterance. It seems that, in both cases, the NS and the ESL participants were influenced by the semantics of the concrete verbs as well as pragmatic factors rooted in common knowledge about the denoted activities. Specific ongoing activities such as writing a thesis and taking guitar lessons would normally not be completed over a period of 1 month.

The mean scores per group for the simple past tense are given in Figure 2. The simple past scenarios were used to examine if the postulation that NSs usually do not backshift past tenses to their past perfect equivalents (e.g., Celce-Murcia & Larsen-Freeman, 1999; Standwell, Reference Standwell1985; Thewlis, Reference Thewlis2007) would be supported by the data and how the ESL and EFL samples would compare with the NS group. The results provided support for this postulation in that the majority of NSs applied zero backshifting in both the immediate and delayed report contexts. The ESL group closely resembled the NSs, in that more than 70% of these participants did not employ backshifting at all. In contrast, the EFL participants acted in a significantly different way: The majority of them preferred to backshift to past perfect in both contexts.

Figure 2.

Mean scores per group for simple past tense.

The described trends were congruent with the participants’ explanations for the principles that guided their choices; that is, the ESL and NS groups were mainly concerned about retaining the meaning of the actual speech (NSs in both reports: 66%; ESL group in both reports: 60%) and keeping the natural flow of the report (approximately 50% of both groups). In turn, the majority of EFL participants indicated adherence to the sequence of tenses rule in both the immediate and delayed report contexts (Bulgarian: 88%; Bosnian: 84%).

The speech scenarios for the future simple tense included two instances of will in the meaning of a definite future plan and two instances of going to in the meaning of future intentions (Firsten and Killian, Reference Firsten and Killian2002). The future tense scenarios aimed to create opportunities for the reporters to show their viewpoint of the original message as believable or not. The expectation was that, over and above the time lapse between the actual and reported speech, the participants would opt to maintain the original tense if they wanted to convey the message as believable and would backshift to future in the past if their intention was to show disbelief by taking a distance from the report. The MANOVA results showed different trends for the immediate and delayed report contexts; that is, in the immediate report context, the EFL groups backshifted significantly more than the NS and ESL participants, the majority of whom preferred to maintain the original tense. In the delayed report condition, the four groups behaved similarly, opting to backshift to future simple in the past most of the time, as shown in Figure 3.

Figure 3.

Mean scores per group for future simple tense.

For the immediate report condition, the NSs and the ESL participants explained their decision to preserve the original tense by two main motives: to maintain the meaning (NSs: 69%, ESL: 60%) or to convey certainty (NSs: 39%, ESL: 45%). Because in the specific context of the speech scenario for future simple tense, these two principles could mean the same thing (i.e., that the reporters were trying to preserve the original meaning because they wanted to imply certainty in what they were told), it can be speculated that the majority of the NSs and the ESL group avoided the future in the past to convey the original message as believable. In contrast, the EFL participants overlooked the pragmatics of the situation and applied the backshift rule, following the sequence of tenses principle as pointed out by 95% of the Bulgarian and 92% of the Bosnian participants.

For the delayed report context, although all four groups chose to backshift, the motives underlying this choice were different. Within the EFL samples, 95% of the Bulgarian and 92% of the Bosnian participants indicated observance of the sequence of tenses rule, whereas none checked the option “convey doubt.” In contrast, 69% of the NSs and 67% of the ESL participants indicated that the main reason for backshifting to future in the past was the time lapse between the actual and the reported speech; only 17% of NSs and 15% of ESL participants intended to convey doubt through tense backshifting.

Length of Residence and Backshifting

The third research question concerned the ESL sample only. It was formulated on the basis of previous research that has found that the time of exposure to authentic SL input is positively correlated with learners’ authentic language use (e.g., Bardovi-Harlig & Dörnyei, Reference Bardovi-Harlig and Dörnyei1998; Félix-Brasdefer, Reference Félix-Brasdefer2004; Schauer, Reference Schauer2006). In the context of the present study, this question sought to find out if there was a relationship between the ESL participants’ length of residence in the United States and the way in which they employed backshifting. The expectation was that those participants who had stayed the longest in the United States would be the closest to the NSs’ performance by employing backshifting selectively and thus to a lesser degree.

To address this question, a Pearson’s r correlation analysis was performed for the two variables: ESL participants’ length of residence in the United States and their total backshifting scores on the five dependent variables for which the significant differences were observed. The future simple in the delayed report context, in which all four groups preferred to backshift the original tense, was not included. The results showed a significant negative correlation between the two variables (r = –.723, p < .001). In other words, the more time an ESL participant had spent in the United States, the less he or she employed backshifting, thus more closely resembling the NSs. The value of the R ² of .52 signified that nearly 52% of the variance in backshifting could be attributed to the effect of length of residence. Additionally, the scatterplot and the regression line suggest that the effect of length of residence is more systematic after the second year of an ESL speaker’s stay. Figure 4 presents these results.

Figure 4.

Correlation between length of residence and backshift score for the ESL group. The number of dots in the scatterplot is less than the number of ESL participants (N = 35) because some dots stand for more than one case.

DISCUSSION AND CONCLUSIONS

This study examined how tense backshifting is employed in English indirect reported speech by ESL and EFL speakers in comparison with NSs to identify whether the choice to backshift is governed by the sequence of tenses rule or by pragmatic and semantic factors. If tense backshifting in English is considered a grammatical means for conveying the reporter’s viewpoint regarding the original utterance—not simply a grammatical transformation triggered by a past tense reporting verb—then the findings of this study provide support for those of previous research that have found that EFL environments promote grammatical accuracy, whereas ESL contexts foster sensitivity to pragmatic norms (Bardovi-Harlig & Dörnyei, Reference Bardovi-Harlig and Dörnyei1998; Félix-Brasdefer, Reference Félix-Brasdefer2004; Schauer, Reference Schauer2006). In this case, the EFL and ESL participants employed backshifting in a different way, in that the EFL groups employed it significantly more and automatically, without awareness of the possibility to express certainty, doubt, or impartiality or to maintain the natural flow of the speech through avoidance of grammatical structures like the past perfect tense. In turn, their ESL counterparts’ use of backshifting resembled that of the NSs: Both of these groups were more selective in backshifting tenses guided by semantic and pragmatic factors.

In view of the findings of previous research (Bardovi-Harlig & Dörnyei, Reference Bardovi-Harlig and Dörnyei1998; Blum-Kulka & Olshtain, Reference Blum-Kulka and Olshtain1986; Félix-Brasdefer, Reference Félix-Brasdefer2004; Olshtain & Blum-Kulka, Reference Olshtain, Blum-Kulka, Gass and Madden1985; Schauer, Reference Schauer2006), the observed significant differences between the EFL participants on the one hand and their ESL counterparts on the other hand can be attributed to the influence of the ESL environment with its numerous opportunities for direct contact with an authentic SL input.Footnote ⁷ Support for this conclusion was also provided by the correlation analysis between the ESL participants’ length of stay in the United States and the way they employed the backshift principle. The results supported the expectation that the more time an ESL speaker had spent in the United States, the more selectively he or she would employ backshifting. The significant negative correlation showed that those ESL participants who had spent more than 2 years in the United States flouted the backshift rule the most, thus closely resembling the performance of the NSs. In the cases in which they preferred to retain the original tense of the direct speech, the ESL participants compromised the backshift rule for semantic and pragmatic reasons. For instance, much like the NSs, the ESL participants avoided grammatical structures like the past perfect in order to sound natural and retain the meaning of the original utterance. Their preference not to backshift the original tenses was guided by a conscious attempt to preserve the inherent meaning of the original utterance or to show their belief in the facts presented by the speaker. In fact, the NS and ESL participants generally preferred not to backshift the original tenses in both the immediate and delayed report contexts, as the only exception was observed in the future simple delayed report. Analyses of the results for each of the three tenses suggested that the reason that NSs and ESL participants preferred not to employ backshifting was related to the inherent semantics of the specific grammatical tense or the meanings of the verbs used in the speech scenarios.

In the case of the simple present tense, the majority of the NS and ESL participants preferred not to backshift to the simple past in both the immediate and delayed report contexts, consciously or unconsciously guided by the ongoing nature of the denoted activities and states, which, even after a period of 1 month, could have been considered valid. In contrast, the semantics of the simple present verbs had little influence over the choices made by the EFL groups, the majority of whom demonstrated adherence to the sequence of tenses rule over and above pragmatic and semantic factors.

A similar trend was observed for the simple past, whereby the NSs and the ESL participants preferred not to backshift to the past perfect in either of the two contexts, following their intuition for naturalness of style. In view of related pedagogical grammar literature (e.g., Bland, Reference Bland2008; Celce-Murcia & Larsen-Freeman, Reference Celce-Murcia and Larsen-Freeman1999; Haines & Stewart, Reference Haines and Stewart2004; Leech & Svartvik, Reference Leech and Svartvik1975; Norris, Reference Norris2008; Quirk et al., Reference Quirk, Greenbaum, Leech and Svartvik1985; Thewlis, Reference Thewlis2007; Vince & Sunderland, Reference Vince and Sunderland2003), these results entirely supported the postulation that NSs tend not to backshift past tenses to past perfect due to a change in meaning usually conveyed through the perfective aspect, or for reasons of style. This exception to the backshift rule, however, remained unnoticed by the majority of the EFL participants, who applied the sequence of tenses rule with no or little consideration for the natural flow of speech.

Because future plans or intentions are more susceptible to personal interpretations, the future simple scenarios, more than those for the other two tenses, provided opportunities for the participants to express belief or disbelief in the original utterance beyond the time-lapse variable. Based on the NS and ESL data, the nonuse of backshifting was observed only in the immediate report context with the purpose of conveying the original utterances as believable. In turn, backshifting was employed in the delayed report context, and the NS and ESL participants explained this fact by their observance of the time lapse rather than by an intention to show disbelief. These results were rather contradictory to the patterns observed in the scenarios for the simple present and simple past, for which the NS and ESL participants’ choice not to backshift transcended time lapse. One explanation could be that the participants’ choice to backshift was subconsciously governed by an intention to convey disbelief, but they had not rationalized it enough and instead opted for time lapse as a possible explanation. Overall, only a few NS and ESL participants seemed to be aware of the potential to convey a message of doubt or impartiality by backshifting the original tense. In contrast, the EFL participants’ overreliance on the sequence of tenses rule was interesting because tense backshifting is not a feature of indirect speech in either Bulgarian or Bosnian. It further suggested that the reason for the EFL participants’ overuse of backshifting could be the result of the special emphasis that was put on it during their English classes (S. Bilbija, personal communication, May 14, 2009; V. Koynakova, personal communication, May 18, 2009).