Formulaic sequences signalling discourse organisation in ELF academic lectures: a disciplinary perspective

3.1 Data

The data for the present study form part of the lecture subset used in Wang (2017), which in turn was drawn from a one-million-word corpus of transcribed spoken academic lingua franca English (ELFA) (Mauranen 2008). Specifically, the dataset used for the present study comprises nine lectures of about ten hours’ transcribed material, covering three broad disciplinary domains: Social Sciences (SS), Natural Sciences (NS), and Medicine (Med). Each lecture is of a distinct topic, given either by an invited speaker or the main lecturer of a course module. Questions from the audience are minimal and are excluded from the investigation because the study deals only with the lecturers’ output. Table 1 presents the word counts of each file involved.

In total, a corpus of 74,350 words was manually examined to identify and annotate FSs signalling discourse organisation, using the UAM Corpus Tool (O’Donnell 2013).

It should be mentioned that with such small quantity of data, it is difficult to prove whether a certain characteristic is disciplinary or unique to a specific lecturer. However, given that disciplinary variation has been ascertained in Wang (2017) with more corpus data, including these nine lectures, the selection of data in the present study was based on the topic (distinctive of a particular discipline) and the structure (largely monologic) as representing the disciplines involved for the sake of a more qualitative analysis.

3.2 Identification criteria

The present study aims to be as inclusive as possible when it comes to the identification of FSs in the data. Therefore, mixed criteria drawn from previous FL research were adopted in this study to decide whether or not a word sequence is formulaic, based on the rationale that “most examples will be captured one way or another” (Wray 2008: 110). If a multi-word sequence satisfies any one of these criteria below, it is regarded an FS (see Wang [2018] for a more detailed explanation).

Grammatical irregularity and/or semantic opacity (Wray 2008; Martinez and Schmitt 2012; Schneider et al. 2014): this means that as long as some aspect of the form or meaning of a word sequence is not strictly predictable from its component parts or from regular grammar, the expression is an FS, e.g. on the other hand, at least, for example, first of all. In order to ensure that this criterion is applied consistently in the study and replicable results can be produced by other researchers, dictionaries (primarily Oxford Advanced Learner’s Dictionary) and the list of phrasal expressions provided by Martinez and Schmitt (2012) were constantly consulted. If a word sequence is highlighted in the dictionaries (either as a separate entry or emphasised in boldface) or occurs on the list, it was considered to contain some kind of irregularity and therefore an FS.

Underlying frame (Wray 2008): this refers to a formulaic frame that links parallel structures with open slots to be filled, often by items of similar characteristics, e.g. not only … but also, as … as.

Situation/register/genre-specific formula (Wray 2008; Buerki 2016): what is idiomatic about this type of FSs is not their internal semantics or syntax, but the fact that they are the common ways (judged by frequency of occurrence) of saying things in a particular situation, e.g. as i said, which means that, i will show you. In this regard, IDIOM Search (Colson 2016), an online tool for the extraction of multi-word phrases, was used. Also included in the analysis are expressions that were not identified by the program but nevertheless share an underlying pattern with those that have been recognised. As will be seen in Section 4.2, such instances are key to our understanding of formulaicity in ELF communication and therefore need to be included.

3.3 A functional taxonomy

The classification of functions for the present study is based on Systemic Functional Linguistics (SFL), which focuses on the underlying communicative functions of language and the systemic choices that are made available by the language system (Halliday 2014; see Wang [2018] for justification of using this framework). Key to the use of SFL is the notion of metafunction, which refers to three separate strands of meaning: ideational, textual, interpersonal. The present study focuses on the textual metafunction that is related to expressions whose primary purpose is to tell the listener how the discourse is organised and to create cohesion as it moves along. The three options within the system of textual metafunction are elaboration, extension, and enhancement (Halliday 2014), all serving to link a clause to its surrounding context. Elaboration means making an element more precise by means of restatement, summarising, or in some other way for discourse purposes. Extension expresses an additive, contrasting, or replaceable relationship between prior and coming discourse. Enhancement qualifies the context by reference to the physical environment, topics of discussion, logical connection, or manner. Each of the three categories gives entry to a more specific system with its own options, which are presented with illustrative examples in Table 2.

While most FSs occurring in the data have a distinct function that is fairly easy to identify, some are less straightforward, in particular those that may potentially have more than one function depending on the context. The FS you know is a typical example. Rintaniemi (2017) lists ten different functional categories of you know in academic ELF interaction, operating on both textual and interactional (a.k.a. interpersonal) levels, but primarily the former, which tallies well with House’s (2009) findings. In the present study, the use of you know as in example (1) is regarded as serving a textual function as its main purpose seems to be to introduce an explanation. However, one may argue that it is also used to establish rapport with the audience by assuming shared knowledge, and therefore serving an interpersonal function as well.

1. the ethics that i have been erm explaining to you in the beginning of my lecture er with justice er you know equality solidarity blah blah blah erm all those erm are vital are very important are very important for the er sustainable development … (ULECD030)

In such cases where a FS may have more than one function, as long as it fulfils a textual function, it was included in the analysis. A more refined analytical framework is needed in future research to deal with the issue of multifunctionality of FSs in a systematic way.

4.1 Frequency and distribution of textual FSs across disciplines

Altogether, 2,249 FSs were identified with a discourse organisation function. Table 3 presents the frequencies (both raw and normalised) of such FSs in the sub-corpora.

As can be seen in Table 3, such FSs are most frequent in the sub-corpus of Natural Sciences (NS) and least in Social Sciences (SS), with Medicine (Med) lying in between. No outliers were identified in each disciplinary group. The log-likelihood test was carried out throughout the study, and the results here show that the differences across the three disciplinary groups are all statistically significant. However, as mentioned in Section 3.1, the small quantity of data involved needs to be taken into consideration while interpreting these results.

Figure 1 shows the distribution of the functional categories involved in the dataset. FSs as spatio-temporal markers form the largest group. Table 4 presents the frequencies of such FSs across the disciplines. Again, disciplinary differences can be seen – spatio-temporal FSs occur most frequently in the NS sub-corpus and least in SS, with Med in between.

Figure 1:

Distribution of FSs of different functional categories in the data.

Table 5 lists the most frequent textual FSs in each disciplinary group. Overall, we can see that the majority of these frequent textual FSs are two- or three-word sequences, whereas four-word sequences, which seem to have attracted most attention in the literature, are scarce with only six types (e.g. on the other hand, you can see here). At the same time, many of the textual FSs are fixed expressions (e.g. for instance, and so on, at least, in fact, because of), unlike in the case of four-word lexical bundles as found in Wang (2017), where most are so-called genre-specific formulas (e.g. as you can see, what do you think, is one of the).

The NS sub-corpus yielded not only the highest frequency of textual FSs, but also a greater variety of different FSs, some of which have a high repetition rate, suggesting that the lecturers in this group relied on discourse-structuring devices to a greater extent than those in the other two groups. The top three most frequent FSs in the NS sub-corpus are shared with the Med counterpart. Two of them, this is (a/the) and you (can) see (here/that), suggest that the lecturers in these two disciplines made use of visual aids, corroborating Wang (2017), which is based on a larger dataset. Apart from these, the Med sub-corpus also shares some FSs with the SS sub-corpus (e.g. so that, because of, et cetera), while the NS and SS sub-corpora display distinctly different preferences.

In the NS sub-corpus, we see in Table 5 a large number of spatio-temporal FSs, used to either introduce the macro-structure of the lecture or to direct the audience’s attention to a particular entity (e.g. as i said, that’s all, i would like to show/talk about, i will show you, on the X side). The same observation is made by Kashila and Heng (2014), who explain that hard science fields often deal with a variety of concrete entities such as a particular procedure, materials, and instruments, which need to be identified and explained in the lecture (see also Schleef 2008). A wide range of expressions such as those mentioned above may thus have been adopted to meet this need, and have gradually acquired the status of formulas in these disciplines because of their high frequency of occurrence.

As mentioned in Section 2, using a frequency-based approach, Wang’s (2017) study yielded a number of fairly frequent sequences that contain repetitions of a single word or sound (e.g. in the in the, and and and and) in Social Sciences, which in turn can be explained in terms of the structure of knowledge in different academic divisions. Through manual identification, however, a clearer picture emerged with regard to the other “real” lexical options that may be typical of this disciplinary group. As shown in Table 5, the SS sub-corpus features in particular phrases of elaboration (e.g. i mean, which means (that), that is, let’s say, for example/instance), logical connection (e.g. because of, that’s why, as long as), and transition (e.g. on the other hand). Again, this observation is consistent with Kashila and Heng (2014) (cf. Section 2). One possible explanation for the prevalence of such FSs in SS is related to the need for more discussion of ideas in this discipline. It should be noted that most previous studies such as Kashila and Heng (2014) and Schleef (2008) are based on ENL settings; the results of the present study as discussed above suggest that lecturers in ELF settings follow more or less the same patterns distinctive of their own disciplinary practices. As Wray (2002) claims, FSs develop to serve important communicative needs of a given discourse community. The lecturers’ preferences for different FSs in the NS and SS sub-corpora may thus be intrinsically bound up with different needs that arise from conveying different domains of knowledge. In disciplines such as SS where one uses language to construct research findings, as opposed to disciplines such as NS where language is used to report findings, the formulaic structures are bound to be different.

The following extracts (2)–(4) from the data should give a taste of the contrast in the use of textual FSs between SS and NS lectures. Example (2) was drawn from an SS lecture, where coherence is achieved as the topic, which involves several related ideas, progresses seamlessly in the stretch of discourse, through the use of a range of textual FSs. To start with, a formulaic frame (not … but) is used to link two parallel ideas, with the emphasis (new information) being placed on the latter part, which is then followed by a comment (that is important) and an explanation led by an exposition marker that means that. In the explanation, logical relationships between segments are made explicit by the use of expressions such as because and so that. The clarification marker let’s say would be superfluous in written language as it adds no meaning to the sentence. The use of it here in the lecture, however, may serve to give the lecturer time to think of the upcoming proposition – a specific time period in this case (since the eighteenth century), followed immediately by another clarification led by at least. The connector and also introduces another two related ideas, with i mean serving a similar purpose as let’s say in the preceding sentence, and the whole stretch of discourse ends with a brief summary (clarification), led by so to sum up.

1. and modernisation is not just a change of social structures, towards what we have now but i would emphasise that modernisation includes also change in the ways of thinking, and that is important, and that means that during the past 150 years people have understood that the society is changing, and they have wanted to change things and change as such has been evaluated as a positive matter, because change has been seen as a promise of better future, so that people could have better future on earth not only in heaven, and this way of thinking has changed rapidly let’s say since the eighteenth century, also in finland people, at least in the first half of the nineteenth century still believed that things go better if the if things (won’t) change. and also two major ideologies of the of the twentieth century i mean capitalism and socialism both have been aware very future oriented ways of thinking they both give promises from change towards a better society, so to sum to sum up clo- er shortly er … (ULEC020)

Wood’s (2006) discussion of how FSs can facilitate fluency in speech may be relevant here. The use of the textual markers as delineated above, mostly multi-word sequences, allows a continuing flow of speech to occur while giving time to both the speaker and the listener for mental processing of other aspects of speech. On the part of the speaker, these markers can help shorten the processing route of speech by bypassing the need for assembling components so that the conscious mind can be focused elsewhere during speech (e.g. planning and formulating the following utterance); on the part of the audience, these markers serve as important signposts to prepare them for what they need to understand, and in so doing help ease their processing load as well.

1. i’ll talk aboutthe synthesis of 2-methylanthraquinone is this thing commercially available then i’ll talk about some of the colour problems in the early stages of bleaching and then i’ll talk about lignin condensation which appears to be w- the root of the problem, er our route of making synthesising 2-methylanthraquinone is to replace benzine with toluene and i’ll show you that reaction in a little while … (ULEC080)

2. the way we got the pure product is just by washing it with hot water and this thing with the carboxylic group wash washes out so in a commercial synthesis you could actually wash the crude-product the mixture of this and this with hot water and then vacuum dry your evaporate your water and get this product and recycle it back er to the start of the cyc- cyclisation stage the ring-closing stage so if you if you recycle the unconverted product you could get a theoretical yield of a 100 per cent for this stage of the reaction and then the yield of the overall reaction would be close to the 92 per cent that for the intermediate, so basically er w- you don’t get any degradation the only product you get from this reaction is this product and er you get some unreacted material and all you’d have do is rec- keep on recycling it and the only thing that would leave is this product it’s a very simple reaction and water is the only solvent er you use … (ULEC080)

Examples (3) and (4) were taken from a NS lecture. Example (3) demonstrates how spatio-temporal FSs are typically employed in the NS sub-corpus. Most of them (i’ll talk about, i’ll show you, in a little while) are what Flowerdew and Miller (1997: 38) call macro-markers, which outline the structure of the lecture, announce discourse goals, or refer to outside the lecture. As we can see in example (3), the signposts are arranged purely in chronological order, linked by the connectors such as then, and then, and. When it comes to the description of the actual research procedure itself, as we can see in example (4), it is mainly the connectors, mostly single-word expressions (such as and, so) that are used to structure the narrative in a linear way, as a response to how the sequence of actions is arranged, namely in chronological order.

4.2 Congruency with frequency-based identification

One of the purposes of the present study is to give an idea of how many and what types of FSs tend to be overlooked by frequency-based methods (cf. Section 1). As introduced in Section 3.2, IDIOM Search was used as a pointer to potential FSs on account of frequency of occurrence. The identified FSs were subsequently coded according to whether or not they were recognised by IDIOM Search. In some cases such as in another occasion in example (5), only part of it (another occasion) was identified by the frequency-based program. Sometimes the sequences captured by the program may contain elements adjacent to a complete formulaic unit such as you in so that you and have in have at least in example (6). These cases were coded as “Partly.”

1. … and maybe if there is not enough discussion now we can do it later in another occasion, … (ULEC01A)

2. … and er i understand that two or three dis not get it yet so er will bring i hope before ten o’clock so that you have at least when he begins, … (ULEC01A)

The remainder of this section is divided into two sub-sections. Section 4.2.1 provides an overall picture, followed by a discussion of common features found in the data with regard to the use of some textual FSs. Section 4.2.2 is devoted to a close look at the types of FSs not identified by IDIOM Search.