Chapter 2 Literature Review
III. REFERENTIAL EXPRESSIONS : make direct references to physical or abstract entities or to textual contexts
(1) identification/focus: identify an entity or part of it as noteworthy (e.g., that’s one of the, of the things that, one of the most)
(2) imprecision: communicate that the previous text is imprecise (e.g., or something like that, and stuff like that)
(3) specification of attributes: focus on some particular attribute of the entity (i) quantity (e.g., there’s a lot of, how many of you, per cent of the) (ii) tangible framing attributes (e.g., the size of the, in the form of) (iii) intangible framing attributes (e.g., the nature of the, in the case of) (4) time/place/text reference
(i) place reference (e.g., in the United States)
(ii) time reference (e.g., at the same time, at the time of) (iii) text deixis (e.g., as shown in figure)
(iv) multifunctional reference (e.g., the end of the, the beginning of the) IV. SPECIAL CONVERSATIONAL FUNCTIONS: occur only in the conversation
subcorpus
(1) politeness routines (e.g., thank you very much) (2) simple inquiry (e.g., what are you doing) (3) reporting (e.g., I said to him)
The method proposed in Biber et al. (1999) has been widely adopted to identify lexical bundles in a corpus. However, different studies may modify the method to serve their research purposes or to overcome the limitations of the corpus. The methods of the follow-up studies are summarized in the following table, which is admittedly not an exhaustive list, and each study will be reviewed immediately afterwards. Methodological concerns for the identification of lexical bundles will be discussed in more detail in the next chapter, which is devoted to the methodology of the present study.
Table 2.4. Methods for identifying lexical bundles (sorted in chronological order).
Study Language L1/L2 Length
(words)
Other criteria Database (e.g., corpus, size, registers)
Biber et al. (1999) English L1 3, 4, 5, 6 10 (3-, 4-word bundles);
5 (5-, 6-word bundles)
5 texts NA The Longman Spoken and Written English Corpus
■ conversation: c. 4,000,000 words (British English); c. 3,000,000 words (American English)
■ academic prose: c. 5,300,000 words
Cortes (2002) English L1 4 20 5 texts NA a self-built corpus
■ freshman compositions: 360,704 words Culpeper and Kytö
The Corpus of English Dialogues 1560-1760
■ late trials: 211,426 words; early trials: 40,727 words
■ late comedy drama: 104,494 words; early comedy drama: 102,817 words
Biber et al. (2004) English L1 4 40 5 texts NA The TOEFL 2000 Spoken and Written Academic
Language Corpus (T2K-SWAL)
■ university classroom teaching: c. 1,248,800 words
■ university textbooks: c. 760,600 words
Cortes (2004) English L1, L2 4 20 5 texts NA a self-built corpus
■ published academic writing (history): 966,187 words
■ published academic writing (biology): 1,026,344 words
■ student writing (history): 493,109 words
■ student writing (biology): 411,267 words Partington and
Morley (2004)
English L1 2, 3, 4, 5, 6, 7
occur more than 3 times NA NA The Newspool Corpus
■ editorials: c. 500,000 words
■ press briefings: c. 250,000 words
■ political news interviews: c. 250,000 words
Nesi and
Basturkmen (2006)
English L1 4 10 NA NA The British Academic Spoken English Corpus:
882,980 words
The Michigan Corpus of Academic Spoken English:
387,818 words Biber and Barbieri
(2007)
English L1 4 40 3 texts NA T2K-SWAL
■ spoken (5 registers): ranging from 39,255 words to 1,248,811 words
■ written (3 registers): ranging from 52,410 words to 760,619 words
The Michigan Corpus of Academic Spoken English
■ university speech: c. 1,700,000 words (200 hours) three comparison corpora
■ The Corpus of Spoken Professional American English
■ The Bank of English National Public Radio
■ The Switchboard Corpus Tracy-Ventura et
al. (2007)
Spanish L1 4 30 20 texts NA a self-built corpus
■ sociolinguistic interviews: 2,222,025 words
■ academic texts: 1,002,550 words Cortes (2008) English;
■ 4 academic disciplines by 3 text types: ranging from 107,700 words to 670,000 words
Kim (2009) Korean L1 3 20 5 texts NA The Spoken and Written Sejong Corpus
■ conversation: 2,604,054 words
■ academic texts: 3,407,020 words
Chen (2010) English L1 4 20 5 texts NA The Electrical Engineering Introductory Textbook
Corpus: 247,346 words
The English for Specific Purposes Textbook Corpus: 99,774 words
Chen and Baker
■ native expert writing: 164,742 words The British Academic Written English Corpus
■ native peer writing: 155,781 words
■ learner writing: 146,872 words
Wood (2010) English L1 4 20 NA NA a self-built corpus compiled from six textbooks
■ a textual subcorpus: 187,959 words
■ an instructional subcorpus: 391,386 words Kopaczyk (2012) Middle
Scots
L1 3, 4, 5, 6, 7, 8
occur more than 10 times 10 texts NA a compilation of legal and administrative texts: c.
600,000 words
■ The Edinburgh Corpus of Older Scots
■ The Helsinki Corpus of Older Scots
■ an unpublished transcript of a burgh court book from the south-west of Scotland
Leńko-Szymańska (2014)
English L2 3 7.6 (in COCA) 5 texts or more in
any of the learner data sets
NA ■ target bundles: The Corpus of Contemporary American English (c. 425,000,000 words)
■ learner bundles: The International Corpus of Crosslinguistic Interlanguage (6 native languages by 3 proficiency levels; ranging from 4,023 words to 16,089 words)
■ target bundles: sample texts from the Health Science Corpus (2,082,409 words)
■ non-native bundles: a self-compiled corpus (120,718 words)
As can be seen from the above table, studies on lexical bundles vary greatly in many dimensions of their methods (e.g., the frequency threshold, the corpus size).
Also noteworthy is that there are several lines of studies centering around lexical bundles.
First, many studies are concerned with the use of lexical bundles in the academic arena, and most of them consider register differences. Biber et al. (2004) examine corpus texts from university classrooms and textbooks. The use of lexical bundles in conversation, academic prose, university classrooms, and textbooks is compared. The overall patterns are summarized in the following table.
Table 2.5. Functional distribution of common lexical bundles in four registers.
Stance bundles
Discourse organizers
Referential bundles
Conversation 29 10 3
University classrooms 33 19 32
Textbooks 4 3 20
Academic prose 3 1 15
The extremely high density of lexical bundles in university classrooms is attributed to the fact that classroom teaching relies heavily on both oral bundles and literate bundles. Though being an information-oriented register, classroom teaching is similar to conversation and features many bundles that are fragments of declarative and interrogative clauses. On the other hand, textbooks are like academic prose, featuring many bundles that are noun phrases and prepositional phrases. Furthermore, a strong relationship between the structure and the function of a lexical bundle is revealed again: for example, a noun phrase serves mainly as a referential bundle, and so does a prepositional phrase.
Biber and Barbieri (2007) further extend Biber et al. (2004) and investigate the use of lexical bundles in a wide range of spoken and written university registers: the former includes classroom teaching, classroom management, office hours, study groups, and service encounters, and the latter includes textbooks, course management, and institutional writing. It is found that each register relies on different functional categories. For example, though both being written registers, course management and institutional writing show significantly different functional distributions: over 50% of the lexical bundles in course management are stance bundles (e.g., students are expected to, you are encouraged to), while nearly 70% of the lexical bundles in institutional writing are referential bundles (e.g., the first day of, the semester in which). This reflects that the use of lexical bundles is influenced not simply by the physical mode (i.e., spoken versus written) but also by the communicative needs of different registers.
Simpson (2004) also examines lexical bundles in academic speech, identifying lexical bundles typical of academic speech and conducting two within-corpus comparisons (i.e., interactive speech vs. monologic speech, and the speech of professors vs. students). A list of 54 lexical bundles occurring significantly more often in academic speech (e.g., you can see, and so on, what I mean) is provided.
Additionally, a functional framework of two main categories, which is summarized in Table 2.6, is proposed: (i) discourse organizing functions, and (ii) interactional functions. These two functions also highlight the dual pragmatic characteristics of academic speech (see also Biber et al. 2004): academic speech is an information-rich genre, so lexical bundles structuring the discourse are common; on the other hand, the interaction between speakers is also important in academic speech, so lexical bundles related to the speakers’ interactivity are common as well.
Table 2.6. Selected functions of lexical bundles in academic speech (Simpson 2004).
Function Subcategory Example
focuser, introducing examples and in fact, what happens is meta-discourse expressions when we talk,
you could say cause-effect markers so that’s why,
the reason why Discourse organizing
functions
summarizers it turns out (that)
questions, sentence stems what do you mean, how do you know explaining, demonstrating as you can see,
I’ll show you spatial organizers, locatives in this class,
on the web hedges, mitigators in some sense,
more or less
Nesi and Basturkmen (2006) zoom in to investigate lexical bundles used as linking devices in university lectures, a register where the information load is so heavy that connections between prepositions are required to be made clear. It is found that as in university classrooms (see also Biber et al. 2004), both oral bundles and literate bundles occupy a crucial role in university lectures. Like classroom teachers, lecturers face real-time constraints and thus feel the urge to use prefabricated bundles;
on the other hand, a lecture is similar to academic prose in that both are less interactive and more pre-planned. When used as cohesive devices in a lecture, lexical bundles usually signal how an idea or a concept is related to another (e.g., and this is the) or signal how a topic or an activity in a lecture is related to another (e.g., I want to you). It is also found that lexical bundles often co-occur with conventionally recognized cohesive devices: for example, the bundle if you look at is found to co-occur with the linking adverbial for instance/example.
Cortes and Csomay (2007) then explore the relationship between the position and the function of lexical bundles in university lectures, with the focus on the opening phase. The opening phase of a university lecture “sets students and instructors in the organization of the class about to start, contextualizing the content to be delivered or discussed further” (Cortes and Csomay 2007: 69). It is found that in the opening phase of a university lecture, lexical bundles are often used as discourse organizers to introduce a topic or highlight the forthcoming discourse (e.g., take a look at).
Furthermore, disciplinary differences in the use of lexical bundles have also drawn some attention. Cortes (2004) examines lexical bundles in published academic prose from two disciplines, i.e., history and biology. Some bundles are associated more closely with one discipline than with the other. For example, the power of the, the creation of the, and in the context of, all identified in the subcorpus of published history, are strongly related to social events or issues; various stance bundles (e.g., are likely to be, the probability that the) are used more often in biology to hedge the effect of an affirmation or make it more tentative. Another interesting pattern is that lexical bundles in history are either noun phrases or prepositional phrases, while lexical bundles in biology vary greatly in their structure. Cortes (2004) also compares lexical bundles in publish academic prose and those in student writing. Many lexical bundles
frequently occurring in published prose (e.g., from the perspective of) are rarely or never used by students. With respect to discourse functions, the way a lexical bundle is used in student writing is sometimes distinctly different from the way it is used in published prose: for example, the bundle at the same time, which is usually used for simultaneity in published prose, is often used for addition in student writing.
Hyland (2008) also examines lexical bundles in four disciplines (i.e., electrical engineering, microbiology, business studies, and applied linguistics), and another framework is proposed specifically for lexical bundles in academic writing (see Table 2.7).9 Many differences in the use of lexical bundles in the four disciplines are observed. For example, nearly half of the lexical bundles in electrical engineering and microbiology (e.g., was added to the, the performance of the) are research-oriented, describing research methods, specifying some aspects of the research environment, etc., and bundles of this kind usually take the form of ‘noun phrase + of’. Based on the findings, Hyland (2008) argues against the assumption that there is a core vocabulary/phraseology in the academic arena (cf. Coxhead 2000, Simpson-Vlach and Ellis 2010), and suggests instead that every discipline has its own phraseology (Swales 1990).
9 The first two disciplines are applied and pure sciences, while the latter two are social sciences.
Table 2.7. Functional framework of lexical bundles in academic writing (Hyland 2008).
I. RESEARCH-ORIENTED: help writers to structure their activities and