Corpus Pragmatics

, Volume 2, Issue 1, pp 1–26 | Cite as

If-Conditionals in Economics Research Articles: From Keywords to Language Teaching/Learning in the L2 Writing-for-Publication Class?

Original Paper

Abstract

Using the tools of corpus linguistics, this study identifies the basic means of knowledge construction in research articles in economics. The results suggest that discourse-signals realize conditional prediction and empirical hypothesis within the macro-speech acts of hypothesis, analysis/interpretation/generalization and prediction, with if being the most key of all connectors and cohesive devices. If-conditionals are described based on categories such as factual and theoretical conditionals, case-specifying and rhetorical conditionals. Given the vast array of forms and functions, the complexity of conditionals can be considered relatively high from the point of view of explicit knowledge. Turning to implications and applications of the research for an elective L2 writing-for-publication program intended for Ph.D. students and researchers in economics, however, it is clear that scholars in economics can use their domain expertise and L1 genre awareness in the L2 classroom. In this context, some recommendations are given for developing consciousness-raising tasks, activities and materials about if-conditionals. Materials are intended to promote semantic processing, noticing and/or reflecting based on nontechnical vocabulary and working explanations that are comprehensible to the learner and adequate to his/her background knowledge, needs and goals. They comprise grammaticality judgments, in-class comparison of well-formed examples for rule identification, explicit corrective (peer) feedback and discussion of multiple-choice items and gap-fills.

Keywords

Disciplinary discourse Economics research articles English If-conditionals Language and genre awareness L2 writing-for-publication course 

Introduction

In the economics tradition (for one, Brandis 1968), research articles unravel through the manifold combinations of hypothesis, analysis/interpretation/generalization and prediction. Taking inspiration from Merlini Barbaresi’s (1983) qualitative investigation into the issue, this paper uses the tools of corpus linguistics—Keyword tool and Concordance tool (Scott’s 2010 [1997] WordSmith Tools suite)—to look at knowledge construction in research articles in economics. More particularly, we shall first concentrate on keywords and key lexical bundles in economics as compared to cognate disciplines. They are categorized based on Hyland’s (2008) influential work on the uses and meanings of lexical bundles in disciplinary discourses (“Keyness in Research Articles in Economics” section). This will help us identify conditional prediction as an important part of economics research articles, come out with especially frequent discourse signals, and relate them to the macro-speech acts of hypothesis, analysis/interpretation/generalization and prediction.

On these grounds, we will then concentrate on the concordances of if-conditionals so as to provide an accurate description of their uses and meanings in research articles in economics. If-conditionals are thus investigated adapting Declerck and Reed (2001) to our purposes: we shall see that possible world (the factual/theoretical dimension) and function in discourse (case-specifying as opposed to rhetorical conditionals) provide a solid descriptive account for the interpretation of conditionals in research articles in economics. This is a move away from the traditional form-based distinction into realis-potentialis-irrealis conditionals or canonical tense patterns 1–3 (“ If-Conditionals” section).

If-conditionals, we shall see, turn out to be key framing devices with a multiplicity of meaning and uses—which makes them functionally complex form-meaning connections in the grammar (in the sense of Purpura 1996). Taking these observations in an applied direction, we conclude our analysis by turning to effective L2 ‘writing-for-publication’ in the disciplines (Hyland 2013: 68). We work on the assumption that small elective EAP programs intended for non-native scholars in economics might benefit from combining materials that enhance language proficiency and genre-awareness. We therefore ask ourselves whether the formal and functional complexity of if-conditionals, which are particular to research articles in economics, should and can be verbalized in simple, non-technical rules of thumbs that can explain a number of formal and grammatical features pertaining to conditionals (“From Keywords to Language Teaching” section).

Keyness in Research Articles in Economics

Banking on work by Brandis (1968), professor of economics, Merlini Barbaresi (1983) carries out extensive qualitative investigation into the macro-speech acts (Searle and Vandervecken 1985) of economics research articles. She shows that genre- and disciplinary-specific knowledge construction results from the complex interplay of hypothesis, analysis/interpretation/generalization and prediction. More to the point, Merlini Barbaresi (1983) identifies macro- and micro-speech acts, relevant lexico-grammatical and organizational devices, and comes to the conclusion that her analysis provides evidence for the key role of conditional prediction in economics research articles. Prediction is probed to bear on the interaction of sets of factors and variables; based on the application of a particular model, prediction follows from and interacts with hypothesis, data analysis and generalization.

It is generally agreed that comparison across cognate disciplines helps shed light onto disciplinary variation—not only in terms of ‘aboutness’ but also as regards specific discourse signals. In the following we foresee that, though not unique to the particular discipline, discourse signaling devices that are variously related to prediction can be shown to be key, or more frequent than chance (Scott 2010 [1997]), in economics research articles. We therefore concentrate on keyness in a small corpus of research articles in economics (HEM-Economics: HEM-E) as compared to research articles in history (HEM-History: HEM-H). Still limiting our scope to the humanities, research articles in marketing, business and management (HEM-Marketing: HEM-M) are also used as a reference corpus to further validate the results.1 In line with recent developments in EAP (e.g. Biber et al. 1999, 2004; Hyland 2008), we take the task of corroborating information gathered from individual keywords (“Keywords” section) with discussion of 3-word, 4-word and 5-word lexical bundles in economics research articles (“Key Lexical Bundles” section). Because lexical bundles (Scott 2010 [1997]: clusters) are word sequences that appear in a genre more frequently than chance (Biber et al. 1999), it makes sense to assume that they are more readily able to return a reasonably comprehensive view of disciplinary-specific discourse practices.

Keywords

In this section we provide a brief outline of discourse-signaling keywords in economics research articles. Once we set aside keywords that provide information about the subject under study or the research field, we are left with the text- and research-oriented items given in Tables 1 and 2. Table 1 compares data from research articles in economics and research articles in history; Table 2 gives a breakdown of the keywords returned from research articles in economics vis-à-vis marketing. (We use italics to mark words that are present, though with different frequencies, both in the HEM-E and in the reference corpus used for comparison; single underlining signals words that are key in the three corpora).
Table 1

Keywords (keyness) [HEM-E; HEM-H]; [HEM-E; HEM-M] (Cacchiani 2011, revised)

N

Word

Freq.

HEM-E.LST  (%)

Freq.

HEM-H-LST  (%)

Keyness

N

Word

Freq.

HEM-E.LST  (%)

Freq.

HEM-M.LST  (%)

Keyness

Keywords (keyness) [HEM-E; HEM-H]

 7

Model

6575

.25

594

0.02

5443.3

218

Follows

1012

.04

171

 

605.8

 11

Table

4405

.17

329

0.01

3900.2

246

Analysis

2109

.08

784

.03

542.1

 15

Results

4242

.16

341

0.01

3663.7

252

Hence

1180

.05

283

.01

530.8

 26

Variables

2640

.1

94

 

2800.1

263

Assumption

993

.04

209

 

503

 46

Variable

1821

.07

83

 

1839.7

330

Due

1014

.04

277

.01

397.9

 47

If

7728

.3

3037

0.12

1806.3

353

Therefore

1854

.07

793

.03

367.4

 48

Data

2749

.11

417

0.02

1765.3

360

Assumed

905

.03

243

 

362

 53

Sample

1997

.08

162

 

1718.6

377

Findings

503

.02

70

 

339.1

 59

Will

5794

.22

2051

0.08

1613.4

383

Denote

379

.01

28

 

336.3

 65

Can

6663

.26

2756

0.11

1414.0

384

Suppose

492

.02

67

 

335.7

 73

Probability

1467

.06

107

 

1307.3

384

Mean

879

.03

246

.01

335.5

 92

Estimated

1227

.05

77

 

1.115.1

402

Shows

1091

.04

374

.02

318.7

 103

Estimates

1157

.04

75

 

1.069.5

409

Then

3252

.12

1828

.07

312.4

 111

Models

1527

.06

219

 

1.013.7

423

Expectations

593

.02

123

 

304.1

 135

Assume

1225

.05

156

 

868.3

437

Conditional

313

.01

19

 

294.7

 141

Hypothesis

1146

.04

138

 

835.7

439

Denotes

259

 

6

 

292.9

 146

May

4838

.19

2228

0.09

816.2

449

Likely

1245

.05

494

.02

268.8

 179

Implies

948

.04

105

 

719.5

454

Find

1320

.05

543

.02

282.7

 203

Case

3341

.13

1453

.06

639.7

460

Analyse

340

.01

33

 

273.3

Table 2

Keywords (keyness) [HEM-E; HEM-M] (Cacchiani 2011, revised)

N

Word

Freq.

HEM-E.LST  (%)

Freq.

HEM-M.LST  (%)

Keyness

N

Word

Freq.

HEM-E.LST  (%)

Freq.

HEM-M.LST  (%)

Keyness

Keywords (keyness) [HEM-E; HEM-M]

 11

Fig.

1912

.07

42

 

3980.7

172

Probability

1467

.06

1364

.02

472.7

 32

If

7728

.3

8888

.16

1515.5

201

Follows

1012

.04

411

.01

298.7

 37

Table

4405

.17

4114

.07

1407.9

220

Assumed

905

.03

724

.01

381.5

 75

Case

3341

.13

3364

.06

910.3

233

Expected

2340

.09

2904

.05

361.6

 95

Since

2369

.09

2190

.04

772.3

246

Estimated

1227

.05

1250

.02

361.8

 131

Then

3252

.12

3836

.07

584.5

247

Assumption

993

.04

886

.02

346.4

 136

Suppose

492

.02

132

 

578.2

272

Denote

379

.01

170

 

314.7

 150

Assume

1225

.05

966

.02

526.7

493

Shall

206

 

88

 

178.1

The typical configuration of research articles in economics, Tables 1 and 2 suggest, is a combination of the macro-speech acts of hypothesis, data analysis, interpretation, generalization, and, though less represented, prediction. Research articles in economics proceed through the interaction of empirical hypothesis and conditional prediction with data analysis, which bears on models, factors and exogenous and endogenous variables specified by the analyst. This explains recourse to keywords that realize or name the underlying (macro-)speech act (i), which may come with different degrees of certainty and probability, expressed by modals or other verbs (ii). With content words that describe and present data and statistics, self-reflexive markers pointing to the disciplinary-specific reliance on visuals are also key (iii). Lastly, markers of hypothesis or cause and reason relations between elements (cf. Siepmann 2005) relate to interpretation and generalization (iv). The following examples provide an illustration based on Tables 1 and 2. (Ranking and keyness are specified for the top-150 keywords in the [HEM-E; HEM-H] list and/or in the [HEM-E; HEM-M] wordlist).
  1. (i)

    Keywords that realize or name the underlying speech act:

    Hypothesis: assume [HEM-E; HEM-H: 135, 863.3; HEM-E; HEM-M: 150, 526.7], hypothesis [HEM-E; HEM-H: 141, 835.7], suppose [HEM-E; HEM-M: 136, 587.2; HEM-E; HEM-H], assumption [HEM-E; HEM-H], assumed [HEM-E; HEM-H];

    Analysis/interpretation/generalization: estimated [HEM-E; HEM-H: 92, 1115.10], estimates [HEM-E; HEM-H: 103, 1069.5], analysis [HEM-E; HEM-H];

    Prediction: expected [HEM-E; HEM-M], expectations [HEM-E; HEM-M].

     
  2. (ii)

    Keywords pointing to different degrees of certainty and probability:

    Modals: will [HEM-E; HEM-H: 59, 1613.40], can [HEM-E; HEM-H: 65, 1414], may [HEM-E; HEM-H: 146, 816.2], likely [HEM-E; HEM-H];

    Lexical verbs: denote [HEM-E; HEM-M and HEM-E; HEM-H], denotes [HEM-E; HEM-M and HEM-E; HEM-H], implies [HEM-E; HEM-M and HEM-E; HEM-H], shows [HEM-E; HEM-H];

    Nouns: probability [HEM-E; HEM-H: 73, 1307.30; HEM-E; HEM-M].

     
  3. (iii)

    Content words that describe present data: results [HEM-E; HEM-H: 15, 3900.20], data [HEM-E; HEM-H: 48, 1760.30], sample [HEM-E; HEM-H: 53, 1307.30], findings [HEM-E; HEM-H], denote [HEM-E; HEM-M and HEM-E; HEM-H];

    Content words that describe statistics: model [7, 5443.30], variables [HEM-E; HEM-H: 26, 2800.10], variable [HEM-E; HEM-E: 46, 1839.70], models [HEM-E; HEM-H: 111, 1013.70], probability [HEM-E; HEM-H: 73, 1307.30; HEM-E; HEM-M], hypotheses [HEM-E; HEM-H], mean [HEM-E; HEM-H], conditional [HEM-E; HEM-H];

    Self-reflexive markers pointing to the disciplinary-specific reliance on visuals: table [HEM-E; HEM-H: 11, 3900.20; HEM-E; HEM-M: 37, 1407.90], Fig. [HEM-E; HEM-M: 11, 3980.70].

     
  4. (iv)

    Markers of hypothesis or cause and reason relations (cf. Siepmann 2005) within the clause or sentence: if [HEM-E; HEM-M: 32, 1515.50; HEM-E; HEM-H: 47, 1806.3], since [HEM-E; HEM-M: 95, 777.3], case [HEM-E; HEM-M: 75, 910.3; HEM-E; HEM-H], hence [HEM-E; HEM-H: 95, 777.3; HEM-E; HEM-M], then [HEM-E; HEM-M: 131, 494.1; HEM-E; HEM-H], follows [HEM-E; HEM-M and HEM-E; HEM-H], due [HEM-E; HEM-H], therefore [HEM-E; HEM-H].

     

Key Lexical Bundles

A first look at Keywords gave us some rough indication of the knowledge construction processes at play in research articles in economics. While we acknowledge that occurrence in two or all three sub corpora may suggest similarity across text-oriented and research-oriented metadiscourses, variation in raw frequencies points to the contrary. To probe more thoroughly the presence of disciplinary-specific mechanisms of knowledge construction and the importance of prediction in particular, we will now have a careful look at lexical bundles across research articles in economics (HEM-E), history (HEM-H) and marketing (HEM-M).

Let us recall that lexical bundles (Scott 2010 [1997]: clusters) are word strings that (a) appear in a genre more frequently than chance, and (b) are found in multiple texts in that genre (Biber et al. 1999). Given corpus size and research purpose, we set our frequency cut-off point at 2 hits per million words. While identified on the basis of frequency, individual lexical bundles can serve one or more functions. Key lexical bundles in economics research articles are grouped under three major categories: research-oriented, text-oriented and participant-oriented bundles (Hyland 2008: 13–19). The associated subcategories are identified and discussed by complementing Hyland’s (2008) classification with insights from Biber et al. (2004)2; illustrative examples are gathered from our lists of key lexical bundles.

As regards research-oriented bundles, the following categories can be identified:
  1. (i)

    Location bundles, which serve as referential expressions that specify text deixis (in Table 2) or indicate time and place, for instance in analyses and process descriptions (in the first period);

     
  2. (ii)

    Procedure bundles, marking research methodology and purpose (it is shown that);

     
  3. (iii)

    Quantification bundles, comprising referential expressions (Biber et al. 2004) associated with a scale of quantity (is the number of, an increase in the), classificatory adjectives denoting an objective frequency measure (not significantly different, negative and significant), and Biber et al.’s (2004) identification bundles (e.g. existential there in that there is no, there exists a, there are a number of);

     
  4. (iv)

    Description bundles, covering a wide range of vocabulary used to present data, and usually concerning the possible outcomes of the application of a model (the first-order condition for);

     
  5. (v)

    Focus bundles, which provide information on the subject under investigation within a given research field (of central bank independence).3

     
Whereas we can set aside subtype (v), focus bundles, as not relevant to our investigation, additional examples of subtypes (i)–(iv) are given in Tables 3 and 4, which compare wordlists from research articles in economics and history ([HEM-E; HEM-H]), and from research articles in economics and marketing ([HEM-E; HEM-M]). To be as systematic as possible, the top-ten bundles in each subcategory also come with information about ranking, frequency in the HEM-E and keyness. Single underlining is used for the top-25 5-word, 4-word and 3-word bundles in the respective wordlist. 4-word and 3-word bundles are not included when they can be subsumed under one or more top-ranking 5-word lexical bundles.
Table 3

Research-oriented bundlesKeywords (keyness) for 5-word, 4-word and 3-word bundles [HEM-E; HEM-H] (Cacchiani 2011, revised)

Research-oriented bundlesKeywords (keyness) [HEM-E; HEM-H]

Location

 5-word: 7 in table # and table [58; 80.3]; 21 panel B of table # [38; 51.4]; 50 the left-hand side of [29; 38.2]; 74 in columns # and # [25; 33]; 88 on the right-hand side [24; 31.6]; 101 column # of table # [22; 30.3]

 4-word: 42 in the second period [74; 97.6]; 208 in table # are [37; 48.8]; 498 a model in which [24; 31.6]; 499 in section # and [24; 31.6]

 3-word: 40 in the model [207; 272.9]; 59 in our model [166, 218.8]

Procedure

 5-word: 17 it is assumed that the [50; 52.7]; 37 the null hypothesis of no [32; 42,2]; 108 are summarized in table # [31; 28.7]

 4-word: 10 we assume that the [133; 175.5]; 11 is assumed to be [130; 171.4]; 24 it is assumed that [114; 116,1]; 35 is defined as the [78; 108.2]; 45 can be written as [70; 92.3]; 50 is a function of [94; 87.3]; 71 the change in the [90; 78.6]; 92 is the number of [65; 70.6]; 117 is the sum of [21; 27.7]; 140 can be used to [87; 59]; 219 the hypothesis that [131; 111.3]

 3-word: 341 I assume that [67; 88.3]

Quantification

 5-word: 114 more than # of the [21; 27.7]

 4-word: 1 an increase in the [326; 354.4]; 3 significant at the # level [102; 134.5]; 26 significantly different from zero [86; 113.4]; 30 the # level of significance [35; 46.1]; 62 of an increase in [61; 80.4]; 47 is equal to the [81; 90.8]; 89 to an increase in [74; 71.8]; 92 is the number of [65; 70.6]; 193 the total amount of [38; 50.1]

 3-word: 208 is less than [118; 116.3]; 220 the higher the [97; 111.2]; 222 magnitude of the [106; 111.2]; 245 is higher than [81; 106.8]; 246 equal to # [98; 106.6]; 463 is lower than [58; 76.5]

Description

 5-word: 70 the relative size of the [26; 34.3]; 72 the standard deviation of the [25; 33]; 73 the present value of the [25; 33]; 79 the dependent variable is the [24; 31.6]; 100 the first-order condition for [23; 30.3]; 103 the short-run Phillips curve [22; 29]; 58 is a decreasing function of [28; 36.9]; 59 is a function of the [37; 36.9]; 69 is an increasing function of [26; 34.3]; 102 an increasing function of the [22; 29]; 124 there is a continuum of [20; 26.4]

 4-word: 7 as a function of [184; 192.7]; 33 the distribution of the [80; 105.5]; 61 the effect of a [73; 80.7]; 63 the standard deviation of [61; 80.4]; 71 the change in the [90; 78.6]; 73 the coefficient on the [59; 77.8]; 74 the first-order conditions [59; 77.8]; 92 is the number of [65; 70.6]; 94 the effect of the [86; 70.3]; 99 the probability that the [52; 68.6]; 104 as a percentage of [51; 67.2]; 112 the probability of a [50; 65.9]

 3-word: 375 probability of a [64; 84.4]

Table 4

Research-oriented bundlesKeywords (keyness) for 5-word, 4-word and 3-word bundles [HEM-E; HEM-M] (Cacchiani 2011, revised)

Research-oriented bundlesKeywords (keyness) [HEM-E; HEM-M]

Location

 5-word: 4 at the end of the [85; 193.5]; 5 are presented in table # [78; 177.6]; 6 at the beginning of the [62; 141.2]; 11 in table # and table [58; 132]; 18 the right-hand side of [45; 104.5]; 40 panel B of table# [37; 84.2]; 65 the second half of the [31; 70.6]; 76 the left-hand side of [29; 66]; 106 in columns # and # [25; 56.9]; 134 in the rest of the [23; 52.4]; 184 in the second half of [20; 45.5]

 4-word: 343 in the present model [21; 30.4]; 348 the model presented in [21; 30.4]

 3-word: 466 in our model [166; 54,3]; 494 in panel A [37; 53.5]

Procedure

 5-word: 14 it is assumed that the [50; 111.6]; 25 can be thought of as [43; 97.9]; 37 is a function of the [37; 84.2]; 159 is the sum of the [21; 47.8]; 203 can be interpreted as a [19; 43.3]; 220 as a measure of the [18; 41]; 225 I estimate the model using [18; 41]

 4-word: 15 is assumed to be [130; 91.5]; 17 it is assumed that [114; 89.1]; 38 it is shown that [49; 68.8]; 79 we assume that the [133; 53.5]; 91 is the ratio of [36; 52.1]; 264 where T is the [23; 33.3]

 3-word: 44 the null hypothesis [233; 150.6]; 153 suppose that the [84; 89.6]

Quantification

 5-word: 46 to an increase in the [36; 82]; 50 the # level of significance [35; 79.7]; 61 the rate of growth of the [31; 70.8]; 133 there are two types of [23; 52.4]; 147 there are a number of [22; 50.1]; 156 more than # of the [21; 47.8]; 159 is the sum of the [21; 47.8]; 198 to the size of the [19; 43.3]; 450 increase in the rate of [13; 29.6]; 452 at the # level in [13; 29.6]

 4-word: 3 an increase in the [326; 47]; 5 the growth rate of [135; 251.4]; 8 significant at the # [110; 172.1]; 9 the size of the [322; 127.6]; 14 significantly different from zero [86; 104.1]; 250 is equal to # [24; 34.7]; 401 there is some evidence that [14; 31.9]

 3-word: 21 is equal to [215; 208.1]; 34 the increase in [223; 166.8]; 251 rise in the [73; 71.1]

Description

 5-word: 8 as a function of the [77; 175.5]; 37 is a function of the [37; 84.2]; 52 effect of a change in [34; 77.4]; 74 percentage change relative to the [30; 68.3]; 81 the probability of success of [29; 66]; 86 is a decreasing function of [28; 66]; 100 is an increasing function of [26; 59.2]; 121 in the form of a [24; 54.6]; 113 the standard deviation of the [25; 56.9]; 122 the first-order conditions for [24; 54,6]

 4-word: 48 the present value of [65; 64.5]; 63 a positive relation between [40; 57.8]; 66 the distribution of the [80; 57.7]; 91 is the ratio of [36; 52.1]; 94 the opportunity cost of [77, 51.8]; 131 probability of success of [31; 44.8]; 132 the real value of [31; 44.8]; the coefficient on the [59; 44.1]; 140 the ratio of the [72; 44.1]; 148 percentage change relative to [30; 43.4]; 157 the elasticity of substitution [29; 41.9]

 3-word: 131 long-run equilibrium [94; 94.4]; 163 present value of [163; 86.6]; 256 a positive relation [49; 70.6]; 273 the variance of [117; 68.1]; 276 the symmetric equilibrium [47; 68]; 500 the critical value [50; 53.2]

Another set, text-oriented bundles, comprises signals that help organize the text and structure its meaning as a message or argument:
  1. (i)

    Structuring signals, or text-reflexive cues that organize stretches of discourse or direct the reader elsewhere (paper is organized as follows);

     
  2. (ii)

    Resultative signals, or markers of inferential or causative relations (it follows that the);

     
  3. (iii)

    Transition signals, which make explicit coherence relations that do not fit in (i) and (ii), also additive and contrastive relations between elements (on the other hand);

     
  4. (iv)

    Framing signals, which delimit an argument by specifying cases, circumstances of application of one or more models, and variables within the model(s) (if the number of).4

     
Tables 5 and 6 give a more comprehensive selection of text-oriented bundles that are key in research articles in economics. Structuring signals are excluded because they are concerned with the organization and structure of papers rather than with knowledge construction.
Table 5

Text-oriented bundlesKeywords (keyness) for 5-word, 4-word and 3-word bundles [HEM-E; HEM-H] (Cacchiani 2011, revised)

Research-oriented bundles—Keywords (keyness) [HEM-E; HEM-H]

Transition signals

 4-word: 77 on the other hand [562; 281]

Resultative signals

 5-word: 26 table # reports the results [36; 47.5]; 44 these results are consistent with [30; 39.5]; 49 table # presents the results [29; 39.5]; 77 results are presented in table [25; 33]; 78 table # shows the results [24; 33]; 84 this is due to the [24; 31.6]; 111 this result is consistent with [21; 27.7]; 128 does not depend on the [20; 26.4]; 129 leads to an increase in [20; 26.4]

 4-word: 66 results are consistent with [61; 80.4]; 87 is determined by the [66; 71.9]; 110 the reason is that [66; 65.9]; 157 these results suggest that [42; 55.4]; 166 this implies that the [41; 54]; 170 the results of this [55; 53.3]; 179 is due to the [61; 52.3]; 458 may be due to [25; 33]

 3-word: 38 depends on the [281; 282.9]; 89 implies that the [195; 184.5]; 232 so that the [196; 108.3]; 381 such that the [88; 84.3]

Framing signals

 5-word: 23 in the presence of the [112; 124.2]; 48 with respect to the [164; 90.6]; 67 if and only if the [26; 34.3]; 109 in the case of a [45; 28.7]

 4-word: 5 if and only if [150; 197.7]; 155 the case in which [42; 55.4]; 157 for the case of [42; 55.4]; 165 the case where the [41; 54]; 457 if the number of [25; 33]; 478 other things being equal [24; 31.6]

Table 6

Text-oriented bundlesKeywords (keyness) for 5-word, 4-word and 3-word bundles [HEM-E; HEM-M] (Cacchiani 2011, revised)

Text-oriented bundles—Keywords (keyness) [HEM-E; HEM-M]

Transition signals

 5-word: 353 as is the case with [14; 31.9]; 354 as well as in the [14; 31.9]; 357 such a way that the [14; 31.9]; 427 in the same way as [13; 29.6]

 4-word: 7 on the other hand [152; 150.9]

Resultative signals

 5-word: 10 as a result of the [59; 134.3]; 23 due to the fact that [44; 100.2]; 45 table # reports the results [36; 82]; 34 the results in table # [38; 86.5]; 45 table # reports the results [36; 82]; 48 on the basis of the [35; 79.7]; 70 these results are consistent with [30; 68.3]; 82 table # presents the results [29; 66]; 95 is due to the fact [26; 59.2]; 11 results are presented in table # [25; 56.9]; 116 this is due to the [24; 54.6]

 4-word: 84 in such a way that [29; 66]; 400 it follows that the [31; 28.5]

Framing signals

 5-word: 19 in the context of the [45; 102.5]; 20 in the case of a [45; 102.5]; 25 as in the case of [44; 100.2]; 29 from the point of view [41; 93.3]; 55 in the context of a [33; 75.1]; 94 if and only if the [26; 59.2]; 210 in the sense that the [18; 41]; 221 the extent to which the [18; 41]; 278 in the absence of any [16; 36.4];298 in the presence of a [15; 34.1]; 343 if the probability of success [14; 31.9]; 345 if this is the case [14; 31.9]

 4-word: 6 if and only if [150, 151.3]; 152 in the first case [48; 42.9]

 3-word: 98 with respect to [594; 108.3]

Third, Hyland’s (2008) participant-oriented bundles focus on personal and impersonal (Biber et al. 2004) stance and engagement features:
  1. (i)

    Stance features, which describe the writer’s attitude and evaluation (are more likely to be, may be due to);

     
  2. (ii)

    Engagement features, which address the reader directly (it is straightforward to show).

     
For purposes of this paper, there is no need to separate stance and engagement features, which may overlap to some extent. Tables 7 and 8 return the top-ranking examples from the corpus as well as information about their keyness and frequency in research articles in economics.
Table 7

Participant-oriented bundlesKeywords (keyness) for 5-word, 4-word and 3-word bundles [HEM-E; HEM-H] (Cacchiani 2011, revised)

Participant-oriented bundles—Keywords (keyness) [HEM-E; HEM-H]

Stance and engagement features

 5-word: 12 can be thought of as [43; 59.3]; 20 are more likely to be [38; 50.1]; 31 it can be shown that [48; 46.1]; 40 is easy to see that [31; 40.9]; 89 it can be seen that [24; 31.6]; 139 it is easy to see that; 146 can be interpreted as a [19; 25]; 161 is straightforward to show that [19; 25]

 4-word: 188 it can be shown [56; 50.3]; 458 may be due to [25; 33]

 3-word: 186 may not be [199; 123.1]; 318 consider the case [83; 93.4]; 471 we consider the [88; 76.2]

Table 8

Participant-oriented bundlesKeywords (keyness) for 5-word, 4-word and 3-word bundles [HEM-E; HEM-M] (Cacchiani 2011, revised)

Participant-oriented bundles—Keywords (keyness) [HEM-E; HEM-M]

Stance and engagement features

 5-word: 16 it can be shown that [48; 109.3]; 21 it should be noted that [45; 102.5]; 25 can be thought of as [43; 97.9]; 33 it is important to note [39; 88.8]; 49 it is easy to see [35; 79.7]; 66 is easy to see that [30; 68.3]; 87 it is clear that the [28; 63.7]; 88 it is not possible to [28; 63.7]; 99 it is well known that [26; 59.2]; 108 it is interesting to note [25; 56.9]; 112 it is possible that the [25; 56.9]

Having gone through the key lexical bundles of research articles in economics, we can now complete our assessment of key data and relate this variety of examples to specific mechanisms of knowledge construction, which develop around the complex interaction of hypothesis, analysis/interpretation/generalization and prediction. More to the point, though not exclusive to research articles in economics, it is now apparent that some key lexical bundles in particular can be seen as serving five major functions, with participant-oriented signals cutting across other categories. Points (i)–(vi) provide full support for this claim. (Discourse signals under (i)–(v) are ordered by overall ranking for (keyness) [HEM-E; HEM-H] and/or (keyness) [HEM-E; HEM-M]; ranking is specified for the top-25 bundles).
  1. (i)

    Procedural bundles that point to empirical hypothesis or conditional prediction: we assume that the [HEM-E; HEM-H: 10, 175.5; HEM-E; HEM-M], it is assumed to be [HEM-E; HEM-H: 11, 171.4], is assumed to be [HEM-E; HEM-H: 11, 171,4; HEM-E; HEM-M: 15, 91.5], it is assumed that the [HEM-E; HEM-H: 17, 52,7; HEM-E; HEM-M: 14, 111.6], it is assumed that [HEM-E; HEM-M: 17, 89.1; HEM-E; HEM-H: 24, 116.1];

     
  2. (ii)

    Procedural and description bundles used for definitions and parameter setting, seen as instrumental in carrying out data analysis and furthering discussion: as a function of [HEM-E; HEM-H: 7, 192.7; HEM-E; HEM-M: 8, 175.5], can be thought of as [HEM-E; HEM-M: 25, 97.9], is defined as the [HEM-E; HEM-H], is a function of the [HEM-E; HEM-H and HEM-E; HEM-M], is the ratio of [HEM-E; HEM-H];

     
  3. (iii)
    Given the reliance of economics on statistics and figures, graphs and tables:
    • Quantification bundles (scoring the highest also among the top-25 bundles): an increase in the [HEM-E; HEM-M: 1, 344.4; HEM-E; HEM-M: 3, 47], significant at the # level [HEM-E; HEM-H: 3, 134.5], the growth rate of (HEM-E; HEM-M: 5, 251.4), significant at the # [HEM-E, HEM-M: 8, 172.1], the size of the [HEM-E, HEM-M: 9, 127.9], significantly different from zero [HEM-E; HEM-M: 14, 104.1; HEM-E; HEM-H];

    • Location signals: at the end of the [HEM-E; HEM-M: 4, 193.5], at the beginning of the [HEM-E; HEM-M: 6, 141.2], in table # and table [HEM-E; HEM-H: 7, 80.3; HEM-E; HEM-M: 11, 132;], panel B of table # [HEM-E; HEM-H: 21, 51.4; HEM-E; HEM-M];

    • Though not in the top-25, description signals such as the probability of success of the [HEM-E; HEM-M] or the standard deviation of the [HEM-E; HEM-M];

    • Structuring bundles: table # and table # [HEM-E; HEM-H: 1, 242.6; HEM-E; HEM-M: 1, 418.9], are presented in table # [HEM-E; HEM-M: 5, 177.6], are reported in table [HEM-E, HEM-H: 8, 76.5; HEM-E, HEM-M: 12, 127.5];

     
  4. (iv)

    Ranking high among the top-25 bundles, framing signals that delimit an argument and specify cases, conditions and circumstances of application of variables within a model: if and only if [HEM-E; HEM-H: 5, 197.7; HEM-E, HEM-M: 6, 151.3], in the context of the [HEM-E; HEM-M: 19, 102.5], in the case of a [HEM-E; HEM-M: 20, 102.5], in the presence of the [HEM-E; HEM-H: 23, 124.2];

     
  5. (v)

    Following from reliable inferencing, resultative signals that allow the analyst to discuss research processes and outcomes: as a result of the [HEM-E; HEM-M: 10, 134.3], due to the fact that [HEM-E; HEM-H: 23, 100.2], and a multiplicity of less key signals, e.g. table # reports the results [HEM-E; HEM-H and HEM-E; HEM-H], on the basis of the [HEM-E; HEM-M], the reason is that [HEM-E; HEM-H], these results suggest that [HEM-E; HEM-H], implies that the [HEM-E; HEM-H], it follows that the [HEM-E; HEM-H].

     
  6. (vi)

    As is obvious, stance and engagement bundles are used to modulate certainty, also in conditional predictions: can be thought of as [HEM-E; HEM-H: 12, 59.3], it can be shown that [HEM-E; HEM-M: 16, 109.3; HEM-E; HEM-H], are more likely to be [HEM-E; HEM-H: 20, 50.1], it is easy to see that [HEM-E; HEM-M], it may not be [HEM-E; HEM-H].

     

If-Conditionals

Taking a genre-based perspective on the teaching and learning of writing in English for Academic Purposes, effective L2 writing in the disciplines can be seen as combining L2 proficiency and genre awareness.

Knowledge construction within research articles in economics, we have seen, appears to develop through the complex interplay of the macro-speech acts of hypothesis, analysis/interpretation/generalization and prediction. This is no surprise: economic analysis is characterized by recurrence to model(s) as well as to exogenous and endogenous variables as specified by the model; the model enables analysts to make claims and predictions in the form of hypotheses based on facts. Particularly, the keyness of if, assum*, therefore, thus, suggest*, denote, impl*, hypothes*, estimate*, result*, find*, likely or of lexical bundles such as if and only if and these results suggest that sits in well with the centrality of empirical hypothesis and conditional prediction.

These claims are important for any attempt at developing academic literacy in the context of a L2 writing-for-publication course5 especially intended for scholars in economics.6 In this regard, if-conditionals are a good point to start: if is the most key cohesive device in research articles in economics, regularly instantiating a premise-conclusion or cause-effect coherence relation. This section thus addresses the formal and functional complexity of if-conditionals, which are particular to research articles in economics. The section “From Keywords to Language Teaching” will then go on to discuss some applications and implications for language teaching.

Drawing on Declerck and Reed (2001), we define if-conditionals as If P, Q structures which comprise a conditional clause, also called protasis or if-clause (P/P-clause), and a main clause, or apodosis (Q/Q-clause).

Traditionally, grammars identify three types of predictive conditionals (term from Dancygier 1998) using tense as a diagnostics. In canonical (tense) pattern 1 conditionals (realis conditionals), P is in the present tense and Q is in the future tense (1a). Canonical (tense) pattern 2 conditionals (potentialis conditionals) exhibit simple past in P and present conditional in Q (1b). Canonical (tense) pattern 3 conditionals (irrealis conditionals) have past perfect in P and conditional perfect tense in Q (1c).

  1. (1a)

    Realis: If we reduce environmental standards, the production in the pollution-intensive sector will become more efficient. (constructed)

     
  2. (1b)

    Potentialis: If we reduced environmental standards, the production in the pollution-intensive sector would become more efficient. (constructed)

     
  3. (1c)

    Irrealis: If we had reduced environmental standards, the production in the pollution-intensive sector would have become more efficient. (constructed)

     
Crucially, a manual investigation into if-concordances shows that while pattern 3 is absent from the HEM-E corpus, canonical patterns 1 and 2 are extremely rare. This appears to be fully consistent with the mechanisms of knowledge construction that are particular to research articles in economics—that is, empirical analysis and conditional prediction.

It should also be noted that there is no one-to-one correspondence between patterns 1–3 and the validity of P, or the extent to which the condition in P is interpreted as more or less likely to actualize in general or specific situations. Pattern 1 conditionals can represent P as potentialis or realis. Potentialis conditionals (1a) represent the validity of P as possibly true: the conditional is assumed to be fulfilled in the actual world. Realis conditionals do not presuppose the actualization of P at the time of speaking; they represent P as generically possible and Q as atemporal (a state). Third, irrealis refers either to conditionals that represent P as unlikely to be true (canonical pattern 2: 1b) or to conditionals that represent the validity of P as contrary to fact (as in canonical pattern 3: 1c).

Based on the above, where I distinguish myself from traditional classifications is in focusing on corpus examples that necessarily depart from canonical tense patterns and, second, in trying to account for conditionals in terms of possible worlds and the validity of P. By doing this, I draw heavily on work by Declerck and Reed (2001) on possible worlds (“Possible Worlds” section) and conditionals interpretation (“Interpreting Conditionals” section).7

Possible Worlds

Declerck and Reed (2001) provide a comprehensive typology of English conditionals based on possible worlds, form and interpretation of the P–Q connection. As regards possible worlds, in (i) factual-P conditionals, the P-clause refers to a situation that forms part of the actual world (that is, its present state and history) (Declerck and Reed 2001: 483). Conversely, in (ii) theoretical-P conditionals, P creates a supposed world. Theoretical-P conditionals comprise: neutral-P, closed-P, open-P, tentative-P and counterfactual-P conditionals. For convenience, Table 9 gives a list of illustrative examples.
Table 9

The possible-world typology of conditionals

Conditionals and possible words

(i) factual-P conditionals

 (2a) Manufacturers of category A items could apply for prices increases if they had incurred increases in allowable costs. (HEM-E~\joeb\511)

 (2b) If, our survey showed, these debtors are unable to pay their own debts, they are insolvent. (constructed)

(ii) theoretical-P conditionals

 (ii.a) neutral-P

(3) You cannot marry a man if you are a man. (HEM-E ~\eer\432)

 (ii.b) closed-P

(4) If environmental standards are reduced, production in the pollution-intensive sector becomes more efficient. (HEM-E~\ejope\4)

 (ii.c) open-P

(5) [It would be sensible for manufacturers of category A items to apply for prices increases.] If they did, they might compensate for increases in allowable costs. (constructed)

 (ii.d) tentative-P

(6) […] it is possible that we would prefer A over B if we took these correlations into account. (HEM-E~\iref\93)

 (ii.e) counterfactual-P

(7) If products had been sufficiently differentiated, technology transfer would have been profitable. (constructed)

In (i) factual-P (realis) conditionals, the P-situation is known to be true in the actual world. Consider (2a) and (2b). Both examples represent the possible P world as the actual world. To put it with Declerck and Reed (2001), in this world “things are [represented] the way they are in the real world as we know it”. In (2a) factuality is grounded in past repetitive habits. In (2b), reference to facts through ‘our survey showed’ makes the validity of P known in context. If can be readily replaced by when or whenever (Declerck and Reed 2001).
  1. (2a)

    Manufacturers of category A items could apply for prices increases if they had incurred increases in allowable costs (HEM-E~\joeb\511).

     
  2. (2b)

    If, our survey showed, these debtors are unable to pay their own debts, they are insolvent. (constructed)

     
Conversely, in (ii) theoretical-P conditionals the P-clause refers to a possible world, or a world other than the actual world.
  1. (iia)

    Neutral-P conditionals are not interested in the actualization of P. The P-clause “expresses a supposition without presupposing anything about the relation of compatibility between the supposed world and the actual world” (Declerck and Reed 2001: 490) (3).

     
  1. (3)

    You cannot marry a man if you are a man. (HEM-E~\eer\432)

     
  1. (iib)

    Closed-P conditionals treat the condition in P as assumed to be fulfilled in the actual world. Given that is a possible alternate of if.

     
  1. (4)

    If environmental standards are reduced, production in the pollution-intensive sector becomes more efficient. (HEM-E~\ejope\4)

     
  1. (iic)

    Open-P conditionals present P as possibly true (5). Because the speaker treats fulfilment of the condition (supposition) as a real possibility, though uncertain, if can be substituted for supposing that, assuming and exhortatives such as let, suppose, or assume (Declerck and Reed 2001).

     
  1. (5)

    [It would be sensible for manufacturers of category A items to apply for prices increases.] If they did, they might compensate for increases in allowable costs. (constructed; based on 2a)

     
  1. (iid)

    Tentative-P conditionals represent the P-proposition as unlikely to be true, as in canonical pattern 2 conditionals (6).

     
  1. (6)

    […] it is possible that we would prefer A over B if we took these correlations into account. (HEM-E~1\iref\93)

     
  1. (iie)

    Counterfactual-P conditionals represent the validity of P as contrary to fact and, thus, false in the actual world. One constructed example is (7), a canonical pattern 3 conditionals.

     
  1. (7)

    If products had been sufficiently differentiated, technology transfer would have been profitable. (constructed)

     

Interpreting Conditionals

If we now look at the meanings and functions of conditionals in economics research articles, it is easy to draw a line between (i) case-specifying and (ii) rhetorical conditionals (Declerck and Reed 2001). Whereas in case-specifying conditionals the P-clause specifies the case or cases in which Q obtains, in rhetorical conditionals the P-clause is used “for rhetorical purposes (rather than being case-specifying)” (Declerck and Reed 2001: 503).

Case-specifying conditionals in the HEM-E corpus comprise (ia) actualization conditionals, (ib) direct inferentials and (ic) purely case-specifying conditionals. More particularly:
  1. (ia)

    Actualization conditionals treat the P-clause as a condition for (or against) the actualization of the Q-situation. Consider examples (2a), (4), (6)—repeated for simplicity—and (8). What licenses an actualization reading in (2a) is the interpretation of P as triggering (causing) Q in the past. In (4), the P-clause describes the condition that actualizes, or licenses, Q and the P–Q relation can be made explicit via recourse to then. In like manner, the clauses in (6) and (8) are interpreted as licensing Q. Specifically, in (8), a closed conditional, If we solve and [… ] denote refers to the conditions that license subsequent steps in the analysis. Importantly, in case can replace if when the emphasis lies onto the anticipation of the actualization of the P-clause rather than onto its actualization.

     
  1. (2a)

    Manufacturers of category A items could apply for prices increases if they had incurred increases in allowable costs. (HEM-E~1\joeb\511)

     
  1. (4)

    If environmental standards are reduced, production in the pollution-intensive sector becomes more efficient. (HEM-E~1\ejope\4)

     
  2. (6)

    […], and it is possible that we would prefer A over B if we took these correlations into account. (HEM-E~1\iref\93)

     
  3. (8)

    If we solve Eq. 2.16 for w(r) and denote the solution by wh(r) then it represents the wage rate that industry h can bear under zero-profit condition. (HEM-E~1\eer\432)

     
  4. (ib)

    It is not always easy to distinguish between some actualization conditionals and (ib) direct inferentials, which feature a premise-expressing P and an implicative link between P and Q (that is, a resultative, licensing, preclusive or non-preclusive relation). In (9a), then and would infer directly point to this type of relation. Another example is (9b): with its reduced P-clause, it illustrates the inferential premise-conclusion link that is particular to mathematical-statistical analysis.

     
  1. (9a)

    If the limiting pivot magnitude were different when the Official is more effective than when she is ineffective, then for a large n the voters would infer that the event of being pivotal was overwhelming evidence in favour of the Official’s type that yields the higher pivot magnitude […] (HEM-E~1\eer\5).

     
  2. (9b)

    Clearly, *−0>2* if 2+/(+) = 1+/, i.e., if 22 = 2. However, if small enough then *−0<2−*. (HEM-E~1\ijoio\723)

     
  1. (ic)

    Unlike actualization conditionals and implicative conditionals, purely case-specifying-P conditionals are non-implicative and, therefore, do not allow then in the Q-clause. Examples here are (1a) and (3). (1a) is a non-predictive unbounded set-identifying conditional in which if could be replaced by atemporal or restrictive when- or whenever-clauses. It is virtually synonymous with Debtors who cannot pay their debts are insolvent. (3) is a gnomic conditional, generally understood to be a universal truth till recently.

     
  1. (1a)

    If debtors are unable to pay their own debts, debtors are insolvent. (constructed)

     
  2. (3)

    You cannot marry a man if you are a man. (HEM-E~1\eer\432)

     
Given the centrality of empirical hypothesis and conditional prediction in research articles in economics, actualization conditionals and direct inferentials largely outnumber purely case-identifying-P conditionals in the HEM-E corpus.
In (ii) rhetorical conditionals, the case-specifying meaning is secondary or absent. Examples from the HEM-E corpus comprise (iia) anchoring-P, (iib) downtoning-P and (iic) speech-condition-defining-P conditionals.
  1. (iia)

    In anchoring-P conditionals, P serves a framing function within the text and cannot therefore follow Q as a postscript-P clause (Declerck and Reed 2001). One example is (10), where If this is so could be replaced by If this is true or the like and then can be inserted between P and Q: the P-clause only anchors the Q-clause into the ongoing discourse.

     
  1. (10)

    If this is so, the data would over-represent close relationships and under-represent diverse, arm-length’s contracts. (HEM-E~1\ijoio\171)

     
  1. (iib)

    Downtoning-P conditionals are commenting-P rhetorical conditionals in which the P-clause tones down, or casts doubt on, the claim made or implied in Q. The P-clause is often elliptical, as in (11), with reduced if any.

     
  1. (11)

    Countries that produce a same commodity usually face different values and signs of this correlation coefficient (Table 1) and, therefore, different gains, if any. (HEM-E~1\ijope\4)

     
  1. (iic)

    One third subtype of rhetorical conditionals are speech-condition-defining-P conditionals. One example is If one grants the theory, in (12). Notice that (12) also exhibits a combination of commenting and metalinguistic Q-clauses in it is an easy step to conclude, which fleshes out an implicative relation between P and Q. Other examples are (8)—where P (If we solve […] and denote) can be interpreted as taking on a performative function—and (9a), where then and infer can serve both an inferential and a speech-condition-defining function while making the inferential relation between P and Q explicit.

     
  1. (12)

    If one grants the theory, it is an easy step to conclude that the transition from an ex-socialist to a market economy is appropriately guided by what I characterize as the efficiency standard. (HEM-E~1\jose\29)

     
  2. (8)

    If we solve Eq. 2.16 for w(r) and denote the solution by wh(r) then it represents the wage rate that industry h can bear under zero-profit condition. (HEM-E~1\eer\432)

     
  3. (9a)

    If the limiting pivot magnitude were different when the Official is more effective than when she is ineffective, then for a large n the voters would infer that the event of being pivotal was overwhelming evidence in favour of the Official’s type that yields the higher pivot magnitude […] (HEM-E~1\eer\5).

     

From Keywords to Language Teaching

The analysis to this point has shown that knowledge construction in research articles in economics involves a configuration and diverse combination of the (macro-)speech acts of hypothesis, data analysis/interpretation/generalization and prediction, where hypothesis is empirical and prediction conditional. In this context, if-conditionals emerge as key discourse-signaling devices and a functionally complex category in which different types can be identified based on possible P-worlds and on P–Q relations. While this marks a major departure from traditional form-based categorizations into patterns 1–3 conditionals and the realis–potentialis–irrealis distinction, intermediate and advanced course-books still concentrate on predictive conditionals and specify canonical pattern 1–3 conditionals for instruction and practice. As extensively argued in “pedagogy-driven corpus-based research” (Gabrielatos 2006), this precludes an adequate representation of the variety of types, forms and functions of if-conditionals (Gabrielatos 2013).

The Complexity of If-Conditionals

In Second Language Acquisition terms, the multiplicity of forms and vast array of types and functions served by if-conditionals is a clear violation of Andersen’s (1984) One-to-One Principle, and a move towards functional complexity (De Graaf 1997: 41; see also Purpura 1996) and complexity of the accompanying explanation (Robinson 1996). What ensues is that learning difficulty can be understood as explicit declarative (as opposed to procedural) knowledge of the grammar (Ellis 2009)—to be measured in terms of (elaborate) metalanguage. In the context of a writing-for-publication course intended for researchers in economics, this suggests a need for language teachers to develop sufficient teacher language awareness 8 (Andrews 2003) as well as genre- and disciplinary-specific grammar of if-conditionals. Indeed, sufficient genre- and disciplinary-specific awareness can serve as a basis for self-directed preparation of tailor-made materials, tasks and activities that are supported by acceptable explanations for L2 learners, as well as useful explicit instruction and explicit types of corrective feedback.

One obvious prerequisite for effective language teaching, however, is preliminary consideration of learner profiles, needs and goals. Overall, Ph.D. students and scholars that join small elective L2 writing-for-publication courses are fully engaged learners (Svalberg 2007, 2009).9 They are motivated by the need to develop writing literacy in the discipline as a means to gaining global recognition and furthering their careers. At the same time, they exhibit varying levels of language competence, though they are often intermediate to advanced L2 users.

As regards disciplinary knowledge, suitable domain knowledge or high domain expertise correlate positively with research age. Particularly, nationally recognized scholars show mastery of written and spoken genres in their native language. Also, they share with Ph.D. students and young researchers full awareness of the centrality of empirical hypothesis and conditional prediction in research articles in economics, as well as some implicit knowledge of the forms and meanings of conditionals in research articles in economics written in their native language. This should greatly facilitate learners’ ability to gain declarative and procedural knowledge of conditionals in research articles in economics.

Implications and Applications: Preliminary Remarks

Reflection on the facilitating role of genre-awareness, learner L2 competence as well as learner goals and objectives (acquiring writing literacy in the discipline) and resulting engagement with language, suggests creating tailor-made teaching materials, activities and tasks especially intended for consciousness raising (Svalberg 2007). Crucially, effective language teaching of if-conditional and their analogs would start by briefly recycling and consolidating previously acquired structures along the lines of traditional student grammars, workbooks and course-books. At this stage, the main focus would be placed onto pattern 1 and 2 conditionals rather than onto pattern 3 conditionals, which were not found in the HEM-E. At a later stage, instruction and in-class discussion would follow in order to make more transparent to the learner form-meaning links in canonical and non-canonical conditionals, as regards both possible P-worlds and set-identifying and rhetorical P–Q relations. Considering the particular learners, their goals and objectives, as well as their considerable to extensive domain expertise, acceptable in-class explanations would have to do away with elaborate metalanguage and translate it into categories and questions that are more comprehensible and adequate to the type of knowledge shared by the learners.
  • Questions about possible P worlds might include the following:
    1. (i)

      To what extent is the if-clause assumed to obtain/be true when uttering the sentence (from certain to obtain, or closed, through open to tentative and counterfactual)?

       
    2. (ii)

      Is it an issue at the time of speaking? (Recall that neutral conditionals are not interested in the actualization of the if-clause).

       
    3. (iii)

      Is the situation described by the if-conditional known (as opposed to assumed) to be true in the actual world/at the time of speaking?

       
  • For set-identifying and rhetorical conditionals the following questions could be appropriate:
    1. (ia)

      Is there a premise-conclusion relation?; (ib) Can we infer the main clause from the if-clause?; (ic) Is the if-clause a necessary condition for realizing the main clause?

      Or, conversely:

       
    2. (ii)

      Does the if-clause serve other functions, such as (iia) anchoring the main clause to the preceding text via repetition (If this is so), (iib) commenting on the main clause, e.g. on its validity (if at all), or (iic) saying what one is doing (If one grants)?

       
  • Additionally, questions about the Q-clause would be:
    1. (i)

      Does the main clause describe facts?

       
    2. (ii)

      Does the main clause make the inferential link to the if-clause explicit and/or express some type of evaluation (e.g. for likelihood)? One example here is It is an easy step to conclude that.

       
Comprehensible metalanguage could feed in a number of consciousness rising tasks.10 Let us take possible P worlds: following recycling and consolidation of partly acquired structures (step 1 in the lesson plan), step 2 would stimulate familiarization with types of possible P-clauses via in-class discussion of enriched (also, constructed) input for neutral conditionals and conditionals that see P as true, probably true, unlikely to be true and certainly not true. This would be a strategy for explicit learning (Ellis 2008). Explicit rules might be given first, e.g.:
  • ‘When the if-clause is assumed to be true at the time of speaking, the if-clause is in the present indicative and the main clause is in the present or future indicative.’

  • ‘When the if-clause is not presupposed the time of speaking, both if-clause and main clause are in the present indicative.’

Alternatively, learners might be encouraged to discover and formulate the rule(s) themselves.
Also (step 3), analogs can be introduced and discussed in relation to the possible uses and meanings of if-clauses in context:
  • Assuming, assuming that, supposing, supposing that, exhortatives let, assume and suppose are used when P is represented as possibly true.’

  • Given that can replace if when the conditional is assumed to be fulfilled in the actual world.’

  • ‘Atemporal/restrictive when and whenever can replace if when the actualization of P is not presupposed at the time of speaking.’

  • In case describes situations that are part of the actual world.’

Having said that, consciousness rising activities that can be used for in-class discussion, (peer) corrective feedback, rule discovery and discussion of satisfactory explanations, might comprise a number of activities and tasks based on authentic corpus data and authored examples. Consider the following suggestions:
  • Expressing grammaticality judgments in relation to possible P-worlds. A good option would be to give the rule(s) or information about context of use first, identify, correct and explain the error. To give one example, if item instruction clarifies that P is known to be true at the time of speaking, then if or in case can be used (10, adapted from 4), though not *given that or *assuming that. (the asterisk and single underlining are used for agrammatical choices). Notice that (4) and (10) are non-canonical pattern 2 conditionals. P is known to be true in the past, with reference to past habits. Also, in case serves an actualizing function in that it triggers the situation described in the main clause (or, better, anticipates the actualization of Q).

  1. (2a)

    Manufacturers of category A items could apply for prices increases if they had incurred increases in allowable costs. (HEM-E~1\joeb\511)

     
  2. (13)
    Manufacturers of category A items could apply for prices increases in case/* assuming that/* given that they had incurred increases in allowable costs. (HEM-E~1\joeb\511)
    • Practicing and testing ability to use conditionals by selecting the right key in a multiple-choice/gap-fill exercise. One example is (13), adapted from (1a). Given the instruction ‘Express a conditional in which P is not presupposed at the time of speaking’, ‘a’ is the right key (single underlining). Another possible option would be to develop a matching task linking if-clauses and main clauses.

     
  3. (1a)

    If debtors are unable to pay their debts, debtors are insolvent. (constructed)

     
  4. (14)
    We _______________ of insolvent debtors if they _______________ pay their own debts.
    1. (a)

      cannot speak … are able to

       
    2. (b)

      could not speak … were able to

       
    3. (c)

      will not speak … will be able to

       
    4. (d)

      would not speak … would be able to

       
     
  • Practicing procedural and declarative knowledge of form-meaning pairings with reference to the interpretation of if-conditionals. For instance, identifying, correcting and explaining errors in (15), a non-canonical conditional. In (15), based on (10), the if-clause is anaphoric, it refers back to the previous sentence and, therefore, cannot follow the main clause.

  1. (10)

    If this is so, the data would over-represent close relationships and under-represent diverse, arm-length’s contracts. (HEM-E~1\ijoio\171)

     
  2. (15)

    The data would over-represent close relationships and under-represent diverse, arm-length’s contracts, *if this is so.

     
  • Noticing and reflecting on form-meaning connections. For instance, trying to explain how pairs of sentences are different. Consider (16a) and (16b), where the if-clause licenses the main clause: whereas in (16a) if might be replaced by when or whenever and be interpreted as neutral (that is, the applicability of the if-clause is not an issue), in (16b) the if-clause is only seen as possibly true—which precludes substitution via when or whenever.

  1. (16a)

    If the products are sufficiently differentiated (i.e., <0), technology transfer is always profitable. (HEM-E~1\iref\363)

     
  2. (16b)

    If products are sufficiently differentiated, technology transfer might be profitable.

     
  • Practicing set-identifying and rhetorical conditionals. For instance, asking the learner to explicitate the premise-conclusion relation between if-clause and Q-clause (17b, adapted from 17a) and then provide tailored feedback on the keys, e.g. then or other options (17c).

  1. (17a)

    Countries that produce a same commodity usually face different values and signs of this correlation coefficient (see Table 1) and therefore different welfare gains, if any. (HEM-E~1\ejope\775)

     
  2. (17b)

    If countries that produce a same commodity usually face different signs of this correlation coefficient, _______________ they also face different welfare gains. (adapted)

     
  3. (17c)

    If countries that produce a same commodity usually face different signs of this correlation coefficient, then/we can conclude that/it is an easy step to conclude that they also face different welfare gains.

     

Conclusions

Using the tools of corpus linguistics, this study identified the basic means of knowledge construction in research articles in economics (HEM-E) as against research articles in history (HEM-H) and business/management/marketing (HEM-M). The results suggest that discourse-signals in research articles in economics realize conditional prediction and empirical hypothesis within the macro-speech acts of hypothesis, analysis/interpretation/generalization and prediction, with if being the most key of all connectors/cohesive devices in the corpus. Accordingly, a second analysis was undertaken to focus exclusively on if-conditionals. Drawing on work by Declerck and Reed (2001), the study specified the forms and functions involved, arguing that the traditional distinction into pattern 1–3 conditionals does not provide a complete picture. Indeed, the data shows that if-conditionals in economics research articles tend to realize non-canonical patterns 1 and 2 and are more adequately differentiated according to the factual/non-factual meanings expressed by the if-clause or the case-specifying or rhetorical relation holding between P and Q in discourse. Given the vast array of forms and functions, the complexity of conditionals can be considered relatively high from the point of view of explicit knowledge, or the ability to verbalize the underlying grammar rule(s) depending on the concepts involved (factuality and case-specification) and the labels needed to express the rule(s).

Turning to implications and applications of the research for an elective L2 writing-for-publication program intended for Ph.D. students and researchers in economics, the long-term aim of this research is to design materials that combine corpus-driven, corpus-based and genre-based discourse analysis so as to cover a wide range of discourse signaling devices. As far as roles and relationships, needs and goals of key stakeholders in the teaching/learning process are concerned, a basic distinction can be drawn between teachers that need to develop genre- and disciplinary-specific teacher language awareness, and fully engaged learners that can use their domain expertise and L1 genre awareness in the L2 classroom. Fairly obviously, the issue was exploring ways of developing consciousness-raising tasks, activities and materials about if-conditionals. Having first described some features of if-conditionals and identified categories such as factual and non-factual conditionals, case-specifying and rhetorical conditionals, it was important to translate these labels and definitions into non-technical vocabulary and working explanations that are comprehensible to the learner and can be used in consciousness rising tasks that promote semantic processing, noticing and/or reflecting: not only grammaticality judgment tasks/tests, but also in-class comparison of well-formed examples for rule identification, explicit (peer) corrective feedback and discussion of multiple-choice items and gap-fills.

Footnotes

  1. 1.

    The HEM corpus of research articles was built and is currently held at the University of Modena and Reggio Emilia, Italy (http://www.cla.unimo.it/cofin). The corpus spans the years 1999–2000 and comprises 2.7 m running words equally distributed across three modules—HEM-Economics, HEM-History and HEM-Marketing. All texts were downloaded from a panorama of prestigious international journals named by disciplinary specialists.

  2. 2.

    Strictly speaking, these (sub-)categories are not mutually exclusive and may overlap to different extents. Take, for example, procedure bundles in definitions, assumptions, parameter and variable setting: they readily combine with participant-oriented features, e.g. can be interpreted as, can be thought of as, I estimate the model using. More importantly, some lexical bundles may serve multiple functions, e.g. when there are (quantification signal; framing signal), results are consistent with the (resultative signal; procedure signal).

  3. 3.

    Note that we do not use focus signals in the same sense as Biber et al. (2004). Rather, taking sides with Hyland (2008) we exclude from this set discourse organizers that introduce a topic (if you look at, in this chapter we) and referential expressions that signal identification (is one of the).

  4. 4.

    Note as well that Hyland‘s (2008) framing signals only represent a small fraction of Biber et al.’s (2004) intangible framing attributes, a subset of referential expressions in their classification.

  5. 5.

    In line with research on variation across genres and disciplines (see, e.g., Hyland and Bondi 2006; Hyland 2008), I would express concern over the effectiveness of programs intended for doctoral students and scholars in separate disciplines or disciplines that are not cognates. Crucially, there is an urgent need for small, elective EAP ‘writing-for-publication’ (Hyland 2013: 68) programs that center on the development of disciplinary- and genre-specific grammar ability, use of lexico-grammar and knowledge of rhetorical structures. However, these courses are still in their infancy in Italian university language centers.

  6. 6.

    The logic of the argument is this. Written literacy instruction is to be seen as a support mechanism that can assist peripheral members of the disciplinary community appropriate expert practices. If keywords and key lexical bundles are ultimately motivated by the rationale for the genre and their use and selection are informed by epistemology of the discipline and underlying mechanisms of knowledge construction, they are also inseparable from how economists see and understand the world. Writing literacy skills are thus reconceptualized as a set of text-structural and lexico-grammatical choices that are effectively used in disciplinary practice. In this context, extensive discourse-analytic, corpus-based and corpus-informed investigation into disciplinary- and genre-specific conventions serves as a first step towards designing teaching materials and tasks. In the writing for publication class, these materials should presumably integrate guidelines that are not language-oriented or only address MA students (e.g. Neugeboren 2005), practice based on EAP textbooks intended for a wide audience in different disciplines (e.g. Swales and Feak 2012 [3nd ed.]), and teaching materials that cover common grammatical structure and vocabulary (as in traditional student grammars).

  7. 7.

    One anonymous reviewer expressed concern over our choice of Declerck and Reed’s (2001) categorization because it is a relatively old publication. Yet, to the best of our knowledge this extensive monograph still represents the most comprehensive empirical analysis of English conditionals to date.

  8. 8.

    For purposes of this paper, we define teacher language awareness as “explicit knowledge about language, and conscious perception and sensitivity in language learning, language teaching and language use” (www.lexically.net/la/la_defined.htm; see also van Lier and Corson 1997 (eds.): knowledge about language).

  9. 9.

    Highly engaged learners are assumed to be attentive learners, with a positive attitude towards (the) language/languages and what it represents/they represent. Additionally, they are willing to interact, to reflect on language with peers, and to provide corrective feedback themselves (Svalberg 2007, 2009).

  10. 10.

    With Ur (2012), we understand a task as a learning activity with two objectives: learning some aspect of language and providing an outcome that can be discussed and evaluated.

References

  1. Andersen, R. W. (1984). The one to one principle of interlanguage construction. Language Learning. doi: 10.1111/j.1467-1770.1984.tb00353.x.Google Scholar
  2. Andrews, S. J. (2003). Teacher language awareness and the professional knowledge base of the L2 teacher. Language Awareness. doi: 10.1080/09658410308667068.Google Scholar
  3. Biber, D., Conrad, S., & Cortes, V. (2004). ‘If you look at…’: Lexical bundles in university teaching and textbooks. Applied Linguistics. doi: 10.1093/applin/25.3.371.Google Scholar
  4. Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (1999). Longman grammar of spoken and written English. Harlow: Pearson.Google Scholar
  5. Brandis, R. (1968). Principles of economics. Homewood: Irwin.Google Scholar
  6. Cacchiani, S. (2011). Keywords and key lexical bundles as cues to knowledge construction in RAs in economics. In S. Goźdź-Roszkowski (Ed.), Explorations across languages and corpora PALC 2009 (Vol. 24, pp. 335–350). Frankfurt am Main DEU: Peter Lang.Google Scholar
  7. Dancygier, B. (1998). Conditionals and prediction. Time, knowledge, and causation in conditional constructions. Cambridge: Cambridge University Press.Google Scholar
  8. Declerck, R., & Reed, S. (2001). Conditionals. A comprehensive empirical analysis. Berlin: Mouton de Gruyter.CrossRefGoogle Scholar
  9. De Graaf, R. (1997). Differential effects of explicit instruction in second language acquisition. Netherlands: Holland Institute of Generative Linguistics.Google Scholar
  10. Ellis, N. (2008). The study of second language acquisition (2nd ed.). Oxford: Oxford University Press.Google Scholar
  11. Ellis, R. (2009). Investigating learning difficulty in terms of implicit and explicit knowledge. In R. Ellis, S. Loewen, C. Elder, R. Erlam, J. Philp, & H. Reinders (Eds.), Implicit and explicit knowledge in second language learning, testing and teaching (pp. 143–166). Bristol: Multilingual Matters.Google Scholar
  12. Gabrielatos, C. (2006). Corpus-based evaluation of pedagogical materials: If-conditionals in ELT coursebooks and the BNC. In 7th teaching and language Corpora conference. University Paris 7 Denis Diderot, 1–4 July 2006. http://eprints.lancs.ac.uk/882. Accesesed 15 July 2016.
  13. Gabrielatos, C. (2013). If-conditionals in ICLE and the BNC: A success story for teaching or learning? In S. Granger, G. Gilquin, & F. Meunier (Eds.), Twenty years of learner corpus research: Looking back, moving ahead (pp. 155–166). Louvain la Neuve: Presses Universitaires de Louvain.Google Scholar
  14. Hyland, K. (2008). As can be seen. Lexical bundles and disciplinary variation. English for Special Purposes. doi: 10.1016/j.esp.2007.06.001.Google Scholar
  15. Hyland, K. (2013). Writing in the university: Education, knowledge and reputation. Language Teaching. doi: 10.1017/S0261444811000036.Google Scholar
  16. Hyland, K., & Bondi, M. (Eds.). (2006). Academic discourse across disciplines. Bern: Peter Lang.Google Scholar
  17. Merlini Barbaresi, L. (1983). Gli atti del discorso economico: La previsione. Status illocutorio e modelli linguistici nel testo inglese. Parma: Edizioni Zara.Google Scholar
  18. Neugeboren, R. H. (2005). The student’s guide to writing. London: Routledge.Google Scholar
  19. Purpura, J. E. (1996). Assessing grammar. Cambridge: Cambridge University Press.Google Scholar
  20. Robinson, P. (1996). Learning simple and complex rules under implicit, incidental, rule-search, and instructed conditions. Studies in Second Language Acquisition. doi: 10.1017/S0272263100014674.Google Scholar
  21. Scott, M. (2010 [1997]). WordSmith tools. Oxford: Oxford University Press.Google Scholar
  22. Searle, J. R., & Vandervecken, D. (1985). Foundations of illocutionary logic. Cambridge: Cambridge University Press.Google Scholar
  23. Siepmann, D. (2005). Discourse markers across languages. A contrastive study of second-level discourse markers in native and non-native text with implications for general and pedagogical lexicography. London: Routledge.Google Scholar
  24. Svalberg, A. M.-L. (2007). Language awareness and language learning. Language Teaching. doi: 10.1017/S0261444807004491.Google Scholar
  25. Svalberg, A. M.-L. (2009). Engagement with language. Interrogating a construct. Language Awareness. doi: 10.1080/09658410903197264.Google Scholar
  26. Swales, J., & Feak, C. B. (2012). Academic Writing for graduate students. Essential tasks and skills (3rd ed.). Michigan: Michigan University Press.CrossRefGoogle Scholar
  27. Ur, P. (2012). Course in English language teaching. Cambridge: Cambridge University Press.Google Scholar
  28. Van Lier, L., & Corson, D. (Eds.). (1997). Encyclopedia of language and education, 6: Knowledge about language. Amsterdam: Kluver.Google Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  1. 1.Department of Studies on Languages and CultureUniversity of Modena and Reggio EmiliaModenaItaly

Personalised recommendations