The development of conjunction use in advanced L 2 speech

The article discusses the results of a longitudinal study of how the use of conjunctions, as an aspect of spoken discourse competence of 13 selected advanced students of English, developed throughout their 3-year English as a foreign language (EFL) tertiary education. The analysis was carried out in relation to a number of variables, including 2 reference levels, one representing English native discourse and the other observed in teacher talk in actual EFL classes, language type exposure, as registered by the participants of the study on a weekly basis, and teaching procedures. The study investigated possible factors determining the development of 3 aspects of conjunction use: (a) formal conjunctions, (b) specific conjunctions, that is, those conjunctions that are both characteristic of natural English discourse and are underrepresented in L2 discourse, and (c) conjunction diversity. The results point to a restricting effect of teacher talk on the development of specific conjunction use and conjunction diversity. These 2 aspects of conjunction use enjoyed only a slight rise, approaching the teacher reference level. On the other hand, formal conjunctions use did increase radically throughout the study, exceeding the native reference level. In this case teacher talk played a reinforcing role at most. As indicated in a correlational analysis, although there was a clear tendency of the participants’ development of conjunction use towards the native reference level, exposure to authentic English may not have been facilitative of the development of this discourse aspect. An interesting observation was made with reference to the effect of formal instruction on the development of conjunction use: Although the subjects did receive intensive training in conjunction use in the 1st semester of their EFL course, it was not until the 2nd year that their levels of formal conjunction use in

It was more than two decades ago that Poland's education underwent a radical change in its English as a foreign language (EFL) practices.Since then teaching English has come a long way from overly formalized instruction, through methodological ad-libbing, to general approaches which are aimed at developing communicative competence in the first place and seem to administer to the specific needs of students, no matter what their age or proficiency level.This metamorphosis has produced various results: Some students have only learned to pidginize English, while others have learned to use it accurately and fluently for professional purposes.It could seem then that the implementation of these new methodologies has indeed helped produce L2 learners that will eventually demonstrate advanced levels of English, rich lexical repertoires and will, above all, construct a natural, native-like discourse.However, it is day-to-day observation that often undermines this belief.Many advanced EFL learners' output, especially in the spoken domain of communication, is rife with awkward utterances, unnatural wording or artificial responses in one-onone communicative encounters.
Spoken discourse is by no means a meaningless interactional tug of war.Nor is it just a mechanical, raw transfer of information from speakers to their recipients or a disorderly exchange of turns.Communication is, or rather should be, a spontaneous allocation of power and an unpredictable, yet logical flow of ideas.To master this competence is quite an undertaking for L2 learners.Just as in L1, L2 discourse construction requires that the learner demonstrate specific knowledge of linguistic devices, understand L2 cultural codes and be able to combine these elements into an individual utterance, unique for the discourse maker, yet still not exceeding the bounds of the social communicative rigor.
There are a number of questions related to L2 discourse construction which certainly beg answers in modern applied linguistics.Some of them are whether advanced L2 learners have the capabilities to construct a natural discourse or what position discourse competence development takes in teaching English.Moreover, it would be worth investigating whether EFL teachers realize the significance of discourse competence and, if so, whether they actually develop it in their classrooms.An example of such a study can be an attempt to establish how advanced students' discourse develops in the long term and what factors might stimulate or impede the process.This article attempts to address these questions in relation to a narrow patch of spoken discourse construction -conjunction use.It provides some theoretical background on discourse construction in relation to conjunction use, as well as the place of communicative competence in teaching of EFL.The main part of the article presents the study of how 13 advanced EFL students' conjunction use developed over a period of 3 years, including factors which might have affected this process.

Discourse and Communicative Competence
Of many models of communicative competence that attempt to single out all constitutive components, two have received the widest recognition.In Bachman's (1990) model, communicative competence is defined as language competence broken down into organizational competence and pragmatic competence.The model by Canale (1983) posits that there are four components that make up communicative competence.Two of them, that is, grammatical competence and discourse competence, reflect the use of the linguistic system itself.The other two, that is, sociolinguistic competence and strategic competence, reflect the functional aspects of communication.
It seems that no matter which model is considered as better reflecting the actual communicative mechanisms, the researchers' attention has been diverted from the grammatical aspects of communication onto the functionalist values of language production by portraying communication as "a synthesis of knowledge of basic grammatical principles, knowledge of how language is used in social contexts to perform communicative functions, and knowledge of how utterances and communicative functions can be combined according to the principles of discourse" (Canale & Swain, 1988, p. 73), the aspects of communication which are of particular interest to this paper.

Discourse Construction
The definitions of discourse are aplenty.It can be viewed simply as "a ling u i s t i c u n it t h a t c o m p r i s e s m o r e t h an o n e s e n t e nc e " ( F r o m k i n, R o dm a n , & Hyams, 2003, p. 581) or as language production built of a minimum of two stretches of speech (Kurcz, 2005, p. 161).Correct as these definitions seem, they refer only to the textuality of language production, which is indeed a significant discourse domain, yet often fails to determine the authenticity of one's discourse.It seems that to really comprehend the phenomenon of discourse construction, a further, perhaps more challenging, multi-dimensional linguistic inquiry must be undertaken, that which goes beyond the sentence itself (McCarthy, 2001, p. 96).
Individual discourse is realized in "the resources which people deploy in relating to one another -keeping separate from one another, cooperating, competing, dominating -and in seeking to change the ways in which they relate to one another" (Fairclough, 2003, p. 88).Here, discourse is seen as constructed on a psycholinguistic level, with individual choices undertaken to establish or maintain social relations.As claimed by Blommaert (2005, p. 29), these choices include semantic relations, as realized in, for example, wording or metaphor; grammar, materializing in, for example, transitivity; modality or cohesion, achieved through the use of, for example, conjunction or schemata; and text structure, for example, episode marking or turn-taking systems.

Conjunction Use as Part of Discourse Competence
Cohesion and textuality of discourse are realized through the use of various grammatical devices, including reference, or ellipsis, or the device of a particular interest to this paper, namely, conjunction.It should be noted that in discourse analysis conjunctions are not restricted to connectives on the level of syntax; this category is extended onto any devices that connect two sentences and help complete the transition from one thought to another.Thus, the category covers the use of the common and or however, but also complex conjunctions such as as a result or yet then (Quirk, Greenbaum, Leech, & Svartvik, 1985).
The role of conjunction in discourse construction is realized in the dimension other than that of reference or ellipsis.As claimed by Halliday and Hasan (1976), "conjunctive relations are not tied to any particular sequence in the expression" (p.227), which suggests that their role as cohesive devices is limited to their organic value in discourse (Halliday & Hasan, 1989, p. 81).Yet, no matter what discoursal role is attributed to conjunction, it does contribute to the texture of spoken and written discourse.As noted by Halliday and Matthiessen (2004), conjunction "provides the resources for marking logico-semantic rerelationships" (p.538) of longer stretches of speech or longer spans of paragraphs.Salkie (1995, p. 76) distinguishes four types of conjunction: (a) addition connectives (e.g., and), (b) opposition connectives (e.g., yet), (c) cause connectives (e.g., therefore), and (d) time connectives (e.g., then).Halliday and Hasan (1976) classify conjunctive cohesion into additive, adversative, causal, as well as the forth domain divided into temporal and continuative.This, however, as well as other conjunctive domains such as Halliday's (2004, p. 541) elaboration, extension, and enhancement, or internal/external conjunctive dimension, exceed the frame of the ongoing discussion.
It should be realized that although conjunction use helps achieve the logicality of discourse, cohesion and coherence do not fully determine the compre-hensibility of communication.It is not just the use of conjunctions that determines appropriate discourse construction.It is the natural use of these devices that does the job.It should therefore be underscored that although a deficit in natural discourse devices may not pose a threat to the textuality of speech, it will no doubt result in constructing discourse with a lower degree of naturalness.And this inauthentic conjunction use is indeed likely to make it more difficult for the other participant of the communicative encounter, particularly an L2 native speaker, to process the spoken output of their interlocutor.

The Authenticity of Classroom Discourse
Exposing learners to communication patterns typical of natural discourse and providing them with relevant practice opportunities may improve construction of discourse.A language classroom, in Poland a naturally dominant educational setting, attempts to create these conditions, whether explicitly or implicitly.A classroom, however, has its apparent institutional limitations and, therefore, its discourse is likely to deviate more or less from natural conventions.Nevertheless, there are claims that classroom communication, and in particular its "modified input and negotiated interaction are no antonyms of genuine communication" (Majer, 2003, p. 14).
What fuels this pedagogical optimism might be the failure of classroom research in the last three decades to suggest remedial measures to authenticate classroom communication.As a result, some theoreticians (cf.Majer, 2003;van Lier, 1996) have sought to challenge the old pedagogical dogma and claim that classroom discourse "constitutes one of many discourse domains" and, therefore, is "authentic in formal learning environments" (Majer, 2003, p. 14).Van Lier goes a step further first rhetorically asking how learners are going to transfer knowledge acquired in the classroom if classroom communication is unnatural (van Lier, 1984, p. 160) and then making a somewhat surprising claim that if teachers "spoke to their students differently, now as if they were addressing a neighbour, now a car mechanic, and so on, they would be using language inauthentically" (van Lier, 1996, p. 130).With all due respect to these theoreticians, it is difficult to escape the thought that this very defense of classroom discourse authenticity may in fact be an instance of label shifting: If something seems not susceptible to change, it can simply be renamed.Thus, what used to be inauthentic/unnatural becomes authentic/natural.And the objective of language instruction is indeed to help learners communicate in natural settings other than a foreign language classroom, that is, in casual social contact with a neighbor or a car mechanic.It cannot be excluded that classroom communication is one of natural discourse domains, materializing in what can be referred to as institutional discourse (Seedhouse, 2004), but Majer's and van Lier's propositions seem to exceed the logic of applied linguistics as, in fact, they amount to centralizing the peripheral and marginalizing the central.

Method
The main portion of this paper is a longitudinal study of how the use of conjunctions, as an aspect of discourse competence of selected advanced learners of English, developed over a period of 3 years and what factors might have affected this process.The study, which is part of a larger project investigating the development of various L2 discourse devices, for example, modality (Jaroszek, 2011), analyzes conjunctions singled out in the survey study (carried out in the year preceding the commencement of this research) and implements the procedures modified after their verification in the pilot study.The specific methods are described in the following section.

Participants
The participants initially included 18 students of English at an English language teacher training college selected from three groups of freshmen.The number of students was a conscious choice, as it was anticipated that some of the students might, for various reasons, quit their education, thus naturally becoming excluded from the study.Eventually, 13 students' conjunction use development was analyzed.There was an even number of students representing high English proficiency and those representing a low proficiency level selected from each group.The selection criterion was entrance examination results.The participants were selected on the basis of document analysis after entrance examinations in July and September 2004.Both spoken and written test results were analyzed.All the selected students gave consent to their participation in the study, had the magnitude of their required commitment in the course of the study explained to them, and were instructed on the procedures of data collection.They were, however, not informed as to the objective of the research, since it would have most likely affected their language performance, thus distorting the results.

Procedures
The study commenced in October 2004 and was completed in May 2007, spanning a total of 3 academic years of the subjects' college education.The development of the subjects' spoken conjunction use was measured peri-odically with the use of the tools described below.In addition, a number of instruments were used in an attempt to determine what factors affected this process.This section stipulates the data collection procedures.
Student diary.The aim of the diary was to identify what type of English the subjects were exposed to over the period of 3 years.The students were obligated to fill in a weekly diary form which was designed to record the type of their L2 exposure.In the first part the subjects were to specify the amount of time they spent in contact with a given type of English.The second part of the diary included the types of classroom interaction in college courses throughout the week.The diary clearly stated that the students were to specify the proportions of the interaction types as used in the classes with respect to student talking time.When absent from college, the students were to fill in the first part of the diary only.The subjects were instructed on how to interpret the terms used in the diary form.The diaries were collected on a weekly basis.Since some subjects occasionally happened to fail to hand in their forms, the results needed to be statistically calculated.
To retain the representative proportions for L2 exposure types measurement, the following equation was used: ExT = TN x (35/Nq), where ExT represents the proportionate L2 exposure, TN represents a total of exposure hrs as reported in the returned questionnaires, Nq represents the number of returned questionnaires, and 35 represents the constant number of weeks in 1 year of L2 exposure.

Student interviews (English).
The development of the participants' conjunction use was measured longitudinally over a period of 3 years.Their conjunction use was measured in spoken performance samples collected on seven occasions throughout the study: in November 2004, February 2005, June 2005, October 2005, June 2006, October 2006and May 2007.For each recording, the participants took part in two approximately 10-min discussions in groups of three.One discussion was designed to trigger the subjects' informal output, the other the formal one.The samples were tapescribed and examined for the use of conjunctions.

Student interviews (Polish).
In an effort to verify a possible L1 transfer in the use of conjunctions, student interviews were conducted in Polish in May 2007.This was designed to help identify the participants' L1 conjunction use and contrast it with the observed L2 performance.Its form was similar to that of the English interviews.
Native speaker interview.In May 2007, the spoken production of a native speaker of English was recorded following the same procedures which were used for regular student interviews.She participated in two approximately 10min discussions in a group of three (the remaining two students were nonnative speakers of English).She was a student at the same college as the research participants, hence she served as a reliable reference point in the study.The aim of this interview was to help compare the subjects' L2 conjunction use with that of their peer.It is realized that interviewing one person only is by no means representative, yet it does offer some reference for further analysis.

Teacher talk analysis.
As revealed in the pilot study, much of the reported classroom interaction involved a lock-step procedure.It can be concluded that it is also teacher talk that might have been one of the major factors affecting the students' conjunction use development.It seemed reasonable then to analyze the conjunctions applied by the teachers of the research subjects throughout a 3-year college program.Each teacher's one 45-min lesson unit was tapescribed and analyzed.This helped investigate possible relationships between teacher discourse and the students' conjunction use development.
Reference subjects.The teachers, whose discourse was subject to analysis, were fully qualified professionals with extensive experience and expertise in teaching English-oriented subjects to university students.A total of 12 teachers included four men and eight women, six with PhD and six with MA degrees.The age range was from 30 to 52, with the average of 41.The teachers were not notified of the exact time of recording, hence the high reliability of teacher talk samples.The English native college student was a 24-year-old female studying at the same college on a regular basis.She was a relatively extroverted type, extremely diligent and self-motivated.
It should be noted that intensity levels presented in this discussion are in the form of the following ratio: DDR = n/L, where DDR represents the discourse device ratio, n represents the number of occurrences found, and L represents the length of language output, as manifested in the number of transcribed text characters.
The ratio calculation helps sustain the proportions of speech stretches and the number of devices used.The length of speech, therefore, had no effect on the calculation result of conjunction use intensity.A similar procedure was used in the calculation of other intensity discourse types, unless otherwise stated.
An attempt will be undertaken to relate the student level of specific discourse device use to the teacher level, which will be an average calculation of the teachers' language output in actual classes (referred to as teacher reference), and to the native speaker's level, calculated from the language output of a native speaker female student recorded in the same communicative setting, referred to as native reference.To examine the reliability of the native reference level, two other samples of native speakers' language output are referred to.They are not taken as reference points, though.
Classroom procedures analysis.To verify the subjects' weekly diary reports and to examine teaching procedures for the use of techniques developing discourse competence, classroom observation was conducted.Since this research investigates spoken production, only speaking classes were observed twice a year.It helped identify the classroom procedures and the teachers' possible attempts to trigger the students' use of discourse devices.During the classroom observations, activities that promote the development of discourse competence, in a direct or indirect manner, were timed.This was expected to help determine the actual position of discourse-related instruction in EFL classroom practices.
In addition, teaching materials used in EFL courses taken by the participants were collected over a period of 3 years.They were examined by the researcher for the existence and use of activities that could help develop specific components of communicative competence.The intensity of discourse competence promotion will be specified on a scale from 0 to 10, where 0 represent no promotion.

Results
This section attempts to present and discuss the development of the overall use of conjunctions and to single out and analyze the development of those conjunctions the specificity of which determines the naturalness of L2 discourse, with reference to both the teacher reference level and native reference level, as well as types of language exposure.

Overall Results
The analysis of the use of conjunctions shows that the intensity levels of their employment can be radically different from recording to recording in the case of the same individual.This suggests that the degree to which the speaker uses conjunctions to link stretches of their speech may depend on the length of the stretches or on individual choices.The occasional deviations from both teacher reference (0.0081) and native reference (0.0089) are no indication of conjunctive deficit.
Although not much deviating from both the teacher and native reference levels, the students' overall conjunction intensity ratio did undergo changes from the first measurement at 0.010978 to the final measurement at 0.008571.Notwithstanding this somewhat insignificant alteration, the overall development trend provided by Figure 1 shows that the intensity of conjunction use by the students was 'corrected' to the reference levels.This again indicates that the exposure of L2 learners to specific input types does have a decisive effect on the development of discourse competence.

Figure 1 Overall conjunction use development
It should also be noted that the final level of conjunction use among the students appears to have remained independent of their L1 conjunction use intensity (0.01102201).This is an interesting finding since the L1 conjunction intensity level is almost identical with the L2 conjunction level at the initial measurement.It seems that the exposure to large quantities of L2 input balanced the L1 influence in this respect.

Specific Conjunctions
Since the use of conjunctions was dominated by the common and, but or so (0.056, 0.0015, 0.011 in native discourse respectively), which are found in large quantities both in L2 English discourse and natural English, the analysis of these three could distort the results and might not adequately reflect the possible development of conjunction intensity.Therefore, an attempt was made to single out those conjunctions which were both characteristic of natural English discourse and underrepresented in L2 discourse as examined in the survey study.These were and so, but still, and still, and then, and but then.They will be referred to in this discussion as specific conjunctions.The results of the study show that two subjects distorted the picture of possible development.Both Student 1 (S1) and S2 demonstrated a high level of specific conjunction use in the initial measurement, the remaining subjects having the same ratio three or four times lower or even at utter zero.To calculate a possible development trend, S1 and S2 were rejected.
As shown in Figure 2, after the rejection of S1 and S2, the development of the use of specific conjunctions was significant, from a jarringly low 0.000058 to 0.000298.Although in Recording 5 there was a sharp breakdown in the use of specific conjunctions, the overall trend was steady and significant, although the final level of student specific conjunction use did not reach the teacher reference level, let alone the native reference level.
What also increased with respect to the use of specific conjunctions was the number of students using them.This number increased steadily throughout the study and doubled at the final measurement (from four students in Recordings 1 and 2, through six in Recordings 3, 4, and 5, to seven in Recording 6 and to eight in Recording 7).

Figure 2 Specific conjunction overall development with S1 and S2 rejected
The use of specific conjunctions by the teachers was not uniform, hence few conclusions can be drawn in this respect.The level of 0.000533, although higher than the students' ratio, could be incidental.Only 11 out of 17 teachers used specific conjunctions, with some of them exceeding the native reference level (e.g., Teacher 7's [T7] ratio at 0.0032 or T8's ratio at 0.0015 compared with the native reference level of 0.001140).What also needs to be underscored is the similarity between the students' final use of specific conjunctions and that of the teachers', as the number of teachers whose talk included them was the aforementioned 11 out of 17 (65%), and the number of students was eight out of 13 (62%), yet with more even intensity results in individual cases.This suggests that notwithstanding a higher teacher overall conjunction use level, as compared with the students' overall conjunction use level, specific conjunctions may not be an integral part of many teachers' repertoire.Their use by teachers may not be dependent on class type and seems to be an individual quality of the teacher.

Formal Conjunctions
The analysis of formal conjunction use development shows more dynamic changes in their use throughout the study from the average ratio at the first measurement at 0.000111 to the final high 0.000533.The intensity of formal conjunction use more than quadrupled over the course of the research.That it radically increased is no surprise as the use of formal conjunctions is directly linked to grammar competence, which was a dominant element of the students' college education.What is astonishing is the fact that, as illustrated in Figure 3, the students' use of formal conjunctions remained at low levels throughout the first year of the study, notwithstanding the intense training in conjunction use which they received in the first semester of the writing and grammar courses.Although it would be an overstatement to claim that formal instruction had little influence on the students' actual discourse competence, it certainly had no immediate effect.
Whether year one linguistic training materialized in this respect only in year two is also difficult to determine.In the third semester, the students took a course in descriptive grammar, during which conjunctive aspects of discourse construction were discussed.This could have been reflected in the temporary increase in the use of formal conjunctions in Recording 4 (0.00029), which decreased shortly after to the stable 0.00025, only to rise to the high 0.00053 in the final measurement.Whether it was so, however, is sheer speculation.The increase in the use of formal conjunctions might as well have been caused by a more extensive exposure of the students to authentic English, richer in formal conjunctions (native reference level of 0.000489), beginning in year two, and less intensive contact with teacher discourse, relatively deficient in the use of formal conjunctions (0.00022), in the same period.This interpretation has solid grounds, since whereas in the fourth and fifth semesters the students' use of formal conjunctions remained around the teacher reference level, as shown in Figure 3, it reached the native reference level in the final measurement.It is also likely that both the aforementioned factors had a facilitative effect on the increase in the use of formal conjunctions by the students.

Figure 3 Overall formal conjunction use development
The teachers' use of formal conjunctions cannot be linked to any particular subject or class type.As illustrated in Table 1, some of the teachers showed high ratios, for example, T13 (U.S. history), with a ratio of 0.000468, and T3 (listening), with a ratio of 0.000498.A similar irregularity can be found at low intensity levels: For example, T12 (American literature), T4 and others (EFL) used no formal conjunctions.In the authentic communications analyzed, all three samples showed higher ratios than the average teacher reference level, with the college student setting the native reference level at the high 0.000489.The high native reference level ratio, as compared with formal conjunction use levels found in the additional two native samples (0.000311, 0.000305), might have resulted from the partly formal contexts of the student recordings.Each measurement included one task which could promote the use of formal conjunctions.Also, the low levels of the teacher use of conjunctions should not be attributed to their possible linguistic deficiency.It seems rather that some of the teachers are still not able to alleviate the apparent classroom limitations and fail to employ natural discourse devices in their classroom communication.
The contrastive analysis of the students' use of Polish and English formal conjunctions shows no correlation (see Table 2).In fact, some of the students demonstrated high Polish ratios and low English intensity levels (e.g., S5 and S12), or low Polish ratios and high English intensity levels (e.g., S3 and S13).The correlation between L1 and L2 formal conjunction use was -0.1857 with p = .544.This finding suggests that the students' use of formal English conjunctions could be independent of L1 influence.

Conjunctions Diversity
For the analysis of conjunction diversity no ratio was used as the number of conjunctions is a finite one.The results will be given in absolute numbers showing how many different conjunctions were used by individual subjects.The analysis shows less radical changes than those in formal conjunction use development, yet the progress is still significant (see Table 3).Although the overall results do show a rising trend, as in the specific conjunction development analysis, two cases (S1 and S2) were rejected from calculations, as their high initial ratios distorted the development trend.As illustrated in Figure 4, the students' conjunction diversity level grew steadily from a low 4.8 at the first measurement to 6.55, a level slightly higher than the teacher reference level of 6.06.In contrast to formal conjunction use, the students' level remained far lower than the native reference level of 11.The six most common conjunctions used by the students were and, but, then, because, so, and or.Formal conjunctions showed little diversity.The natural and so or but then, already discussed in the previous sections, were used sporadically.In addition to the common six conjunctions, the college reference student used however, even though, and on the other hand and the aforementioned specific conjunctions.So did the other two native referents.
The distribution of conjunctions throughout the teacher discourse was relatively even, not exceeding the bounds of 5 (for T17) to 7 (for T5 and T12), with the average of 6.058824.Since the observed individual differences were insignificant, no conclusion as to the type of class or subject taught can be drawn.
The students' Polish conjunction diversity levels were comparably high and reached the English native reference level with the average of 11.62.This indicates that it was the students' discoursal deficiency in the use of conjunctions, and not classroom limitations, that brought about the low levels of their conjunction diversity.At the same time, for all the students the correlation between L1 and L2 conjunction diversity was virtually nonexistent: 0.0603 with p = .845.Yet, after the rejection of five students: S6, S9, S10, S11 and S12, it amounted to a significant 0.7054 with p = .049,which suggests that, at least in individual cases, L1 conjunctive diversity can affect the diversity of conjunctions in L2.

Classroom Procedures
The analysis of classroom procedures indicates that the development of discourse competence may take a peripheral position even in teaching English to advanced learners.As shown in Table 4, illustrating the place of discourse competence in teaching materials used in EFL classes, discourse competence was taught predominantly with regard to grammar (31.16, as compared to oral skills at 14.31), with the stable level of discourse-oriented grammar teaching at approximately 5.0 throughout the study and oral discourse development between a low 0.87 in the first semester and 4.23 in the forth semester.Interestingly, sociolinguistic competence also appears to have been dismissed in the teaching process (7.57).
This peripheral position of discourse competence development was confirmed by classroom observation.Although the number of observations ( 6) is not representative of all the EFL courses, since the observed classes were speaking-oriented, a somewhat gloomy picture of classroom practices emerges.In all six lessons, discourse competence was promoted, yet it happened indirectly through the negotiation of meaning in pairwork or groupwork.Table 5 illustrates this finding.Only in two lessons were the students instructed on discourse construction.Both lessons were taught by one teacher, which suggests that it is not a syllabus, but individual teaching beliefs that determine the content of language instruction.If the statistics were to be trusted in this respect, out of 150 contact hrs of speaking oriented classes, 16% were devoted to direct development of discourse competence, which clearly indicates that, throughout their 3-year college education, the students received fewer than 30 hrs of language instruction that directly promoted spoken discourse competence as contrasted with 760 contact hrs of EFL classes.It is a safe statement, then, that discourse competence may still be left to its own self-adjustment, which casts doubt on the adequacy of teaching practices.

Conclusions
The analysis of possible factors contributing to the development of conjunction use was carried out in two ways.Linearly, overall scores in conjunction use intensity levels were related to possible factors.The other way in-volved correlating individual subjects' conjunction use development with types of their language exposure throughout the study.As the results show, it is teacher talk, exposure to authentic input, and teaching procedures, although not necessarily in the order given, that appear to be fundamental underlying characteristics of the development of discourse competence in advanced L2 learners.These results will now be discussed.
The analysis of formal conjunction use produces interesting results with regard to the effect of formal instruction on discourse development.Notwithstanding the intensive training in conjunction use which the students received in the first semester of the grammar and writing course, their levels of formal conjunction use did not increase until the second year, which indicates that formal instruction may have no immediate effect on the development of discourse competence.It seems that fully internalizing a discourse device so that its use can pass situational rigors of communication, as in the case of formal conjunctions, is a long-term process.It is also possible that the subjects' progress in the use of formal conjunctions was stimulated by factors other than language instruction, that is, teacher talk or authentic L2 input.However, whether the subjects' eventually progressed in formal conjunction use in the second year due to exposure to teacher talk, natural discourse or other factors is not certain.
The analysis of the development of specific conjunction use and of conjunction diversity has produced different results.Unlike the use of formal conjunctions, which increased radically throughout the study, the levels of both the employment of specific conjunctions and conjunction diversity rose only slightly, approaching the reference level set by teacher talk.In this respect, the native reference level was far higher than both the teacher reference level and the student level at the final measurement.This finding suggests that teacher talk, no doubt, has an effect on learners' conjunction use, yet not always a positive one.
The effect of teacher talk on the development of conjunction use definitely deserves more than a thought in this discussion.It seems that the development of specific conjunction use and conjunction diversity was affected primarily by teacher talk, which showed relatively low levels in these respects.The six most common conjunctions used by the students and the teachers were and, but, then, because, so, and or and they did perform their function adequately, making the discourse coherent and cohesive.The natural and so or but then were sporadically used.This finding, however, should not be interpreted as a mere criticism of the teachers' discourse competence.A foreign language classroom has its apparent limitations, and for various reasons, including educational ones, teacher talk is, and sometimes must be artificially formalized, focused mainly on knowledge transfer or factual teaching and, consequently, deficient in communicative devices, thus departing from natu-ral, standard discoursal conventions.The finding under discussion also indicates that it is not only the textuality of speech that defines the whole of the discourse, but also the natural and appropriate use of discourse devices, which in the case of conjunctions suffered a significant deficit on the part of both the students and the teachers.And this very coherence and cohesion factor appears to account for the reason why the students refrained from a more resourceful use of conjunctions.The students, as it seems, may not have deemed it necessary to use other conjunctions if those at hand ensured the coherence and cohesion of their discourse.Sensitizing both L2 learners and teachers to the necessity of a more resourceful and natural use of conjunctions is, therefore, advisable.
Natural speaking standards can be enforced by intensive contact with authentic English, whether through individual interaction with L2 native speakers or passive exposure to input.However, exposure to authentic English, as this study indicates, may not have a remedial effect on all discourse domains.Although the overall results show a clear tendency of the subjects' development towards the native reference level, a correlational analysis of all 13 individual subjects failed to produce results indicating regularities.
Apart from the subjects' mother tongue, teacher talk and type of L2 exposure, the study also examined classroom procedures applied in EFL classes through classroom observation and teaching materials evaluation.The teaching materials, collected on a weekly basis throughout the study, clearly show that oral discourse competence took a peripheral position in classroom procedures.The same conclusion can be drawn from classroom observations, which indicate that the students might have received fewer than 30 hrs of language instruction that directly promoted spoken discourse competence throughout their 3-year college education, as contrasted with the total of 780 contact hrs of EFL classes.Although, the sample of classroom observation is by no means representative, these findings may point to the inadequacy of teaching procedures with respect to discourse competence development at university level.
This gives a somewhat gloomy picture of classroom practices.Teaching EFL may still be viewed as the development of communicative competence mainly with regard to grammar competence and sociolinguistic competence.In contrast, discourse competence appears to be stranded on pedagogical peripheries or optimistically left to its own self-adjustment, which could materialize on the condition that a sufficient amount of naturalistic instruction or exposure to large quantities of authentic input is provided.If, however, the predominant educational setting is a foreign language classroom, more emphasis should be placed on techniques helpful in natural discourse construction, especially at advanced levels.
Although the present research has shown a number of developmental patterns in conjunction use and identified possible factors determining it, there are areas which require further investigation.A major question confronting future researchers is whether the number of 13 students that participated in this study is representative enough to make valid generalizations for a larger population.In addition, future research should explore the development of discourse domains other than those included in this investigation, such as use of back-channeling devices or references as well as discourse marking.Future research could also focus on identifying other factors that most likely determine the construction of discourse and the development of discourse competence.Possible factors include personalities, individual differences, or learning styles of the speaker.

Figure 4
Figure 4 Conjunction diversity development with S1 and S2 rejected

Table 1
Teacher formal conjunction use

Table 2
Polish formal conjunction vs. English formal conjunction use

Table 3
Conjunction diversity development

Table 4
Teaching materials vs. communicative competence development

Table 5
Classroom observation results: type of discourse promotion