Download presentation
1 -

Stance in the BAWE Corpus New Revelations from Multidimensional


AnalysisHilaryNesi Coventry University UK and Sheena Gardner Coventry University UK The British Academic Written English BAWE corpus contains about 65 million words of proficient university student wr

felicity's Recent Documents

Getting Together
Getting Together

1COVID-19Tips on SafelyCOVID-19 is on the rise in New York City Gathering with other people especially indoors increases the risk of spreading COVID-19You should not host or attend a gathering But if

published 0K
determine which students are engaged in your course
determine which students are engaged in your course

io button tistics Tracking On/Off 2 Statistics Tracking section to turn button to save your changes Note tatistics Tracking appear ation 4 meaning only new information

published 0K
Tracking Analyst Tutorial
Tracking Analyst Tutorial

Copyright 1995-2010 Esri All rights reservedTable of ContentsArcGIS Tracking Analyst TutorialTracking Analyst tutorial exercises3Analyzing hurricanes using Tracking Analyst4Tracking Analyst TutorialC

published 0K
MakingSpaceforthePlaceboEffectinPainMedicineDanielEMoermanPhDandAnneHa
MakingSpaceforthePlaceboEffectinPainMedicineDanielEMoermanPhDandAnneHa

Abroadviewoftheplaceboeffectincorporatingneurobiologyindividualpsychologyepistemologyhistoryandculturedeeplyenrichesourunderstandingofthesecomplexandpowerfulforcesandindeedurgesustoabandonthatnarrowan

published 0K
Historical Legacies of Christianity
Historical Legacies of Christianity

1in East AsiaCo-sponsored byThe Kirishitan Bunkoat Sophia UniversityThe Ricci Institute for Chinese-Western Culturalat the University of San FranciscoSaturday October 5 2019Sophia University Central L

published 0K
Pharmaceutica Sci 251262 2005
Pharmaceutica Sci 251262 2005

Sclentia O Variations of naphthoquinone micropropaaated vitro under qreenhouse and outdoor growth conditions Christoph Wawrosch Eva Vackar Department of Pharmacognosy University Althanstr A-1 090 Vien

published 0K
Letter of Map Amendment
Letter of Map Amendment

-Out As Shown OAS InstructionsLOMA-the property and/or buildings as to whether it is located within the Special Flood Hazard Area SFHA Only Obtain MT-EZ formFound onFEMAs site at -ezshtm also availab

published 0K
She offers the iceberg metaphor for culture The products and practices
She offers the iceberg metaphor for culture The products and practices

They will bring those questions up with her later she says so that together they can iden-tify the perspectives behind the practicesfor example how their Chinese hosts react when given a gift and what

published 1K
Download Section

Download - The PPT/PDF document "" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.






Document on Subject : "Stance in the BAWE Corpus New Revelations from Multidimensional"— Transcript:

1 Stance in the BAWE Corpus: New Revelatio
Stance in the BAWE Corpus: New Revelations from Multidimensional Analysis Hilary Nesi (Coventry University, UK) and Sheena Gardner (Coventry University, UK) The British Academic Written English (BAWE) corpus contains about 6.5 million words of proficient university student writing, categorised in terms of ‘genre families’ and distributed fairly equally across levels of study and disciplinary groupings. The corpus has been examined from various perspectives, including multidimensional analysis (M DA) ( Nesi & Gardner 2012) using Biber’s 1988 dimensions. This analysis revealed clear distinctions between texts produced by first, second and final year undergraduates and Masters students, and between texts belonging to different disciplinary groupings. In terms of stance, epistemic modality was the most discriminating feature, being considered a tool for persuasion on Dimension 4 . Published academic prose from the Lancaster - Oslo - Bergen (LOB) Corpus had a neutral score on Dimension 4 in Biber’s own 1988 study, but scores for BAWE assignments were negative on Dimension 4 , and became increasingly so across levels of study (Table 1). Scores for Arts and Humanities assignments were particularly low (Table 2). Level Score 1 st year undergraduate - 1.4 2 nd year undergraduate - 1.4 3 rd year undergraduate - 1.5 Masters - 2.0 Table 1: BAWE Dimension 4 scores by level Disciplinary Grouping Score Physical Sciences - 1.2 Social Sciences - 1.3 Life Sciences - 1.5 Arts and Humanities - 2.3 Table 2: BAWE Dimension 4 scores by disciplinary grouping Biber’s 1988 dimensions still have validity, because the results from studies which employ them can easily be compared. For example, the 1988 dimension 4 scores for BAWE and Biber’s LOB subcorpus can be taken as an interesting indicator of differences between student and expert writing. S tudent writing below PhD level appears to have fewer overtly persuasive features than expert writing, perhaps because und ergraduate and Masters students tend to discuss the work of others rather than presenting their own original research. Stance theory has developed considerably since 1988, and features have now been added to Biber’s tagset, enabling clearer distinctions t o be made between registers which differ predominantly in terms of stance. Biber (2006) uses this tagset as the basis for a detailed linguistic description of registers encountered by university students, such as classroom teaching, coursepacks and institu tional writing, but does not include any analysis of texts (such as assignments) produced by students themselves. A new multidimensional analysis of BAWE (BAWE2016) has built on this work, using the extended tagset to create new dimensions specific to univ ersity student writing. These bespoke dimensions distinguish between BAWE texts grouped by discipline and genre family, without reference to registers in other corpora. Following MDA methodology, 39 linguistic features were retained and four new factors w ere identified ( Gardner, Nesi & Biber , under review), to be treated as dimensions along which the BAWE registers could be situated . Two of the four new dimensions depend strongly on stance features. The variation along each dimension w as then interpreted by qualitative means, manually examining high - and low - scoring texts to ascertain the communicative effect created by the clustering of feature s . S tance adverbials and stance nouns controlling that - clauses cluster ed with 3rd person pronouns, proper nouns and communication verbs at the negative pole of Dimension 1. T ogether these have been interpreted as indicating ‘ s tance toward the work of others’. Essays tend to have the lowest scores on th is dimension , particularly in Arts and Humanities disciplines ( Table 3). Discipline Score Comparative American Studies (n=74) - 6.78 History (n=95) - 6.89 English (n=106) - 7.38 Classics (n=82) - 10.53 Philosophy (n=106) - 10.67 Table 3: Disciplines with the lowest scores on Dimension 1 Stance nouns controlling that - clauses were relatively rare in Biber’s university language corpus, with the exception of fact (Biber 2006:112 ). I n texts with low scores on Dimension 1 most of these nouns ha ve an epistemic function, but s ome fall into the category of ‘Communication’ or ‘Attitude ’ nouns (Biber 2006:93). Stance nouns in the 20 lowe

2 st - scoring texts are listed in Table
st - scoring texts are listed in Table 4 . Table 4: Stance nouns in the 20 lowest - scoring texts on Dimension 1. Likewise s tance adverbials mainly serve as epistemic devices , particularly to indicate certainty . Table 5 lists these items under the categories identified by Biber (2006:92) . Likelihood Certainty Style Attitude Apparently 3 Actually 33 According to 9 Importantly 5 Arguably 3 Always 26 Clearly 13 Ironically 2 Evidently 3 Certainly 15 Frankly 1 Paradoxically 1 Kind of 13 Decidedly 2 Fundamentally 3 Surprisingly 1 Maybe 1 Doubtless 1 Generally 4 Tragically 2 Perhaps 10 Evidently 3 Mainly 4 Unfortunately 1 Likelihood Certainty Communication Attitude Belief 9 Certainty 2 Argument 1 Fear 4 Assumption 2 Conclusion 3 Confession 1 Feeling 1 Claim 1 Fact 30 Explanation 1 Hope 1 Hint 1 Precept 1 Interpretation 1 Illusion 1 Hypothesis 1 Principle 1 Intimation 1 Indication 2 Idea 11 Proof 1 Objection 1 Understanding 2 Interpretation 1 Realization 1 Proposal 1 View 9 Intimation 1 Revelation 2 Proposition 3 Notion 3 Statement 1 Possibility 1 Premise 3 Presumption 1 Sign 2 Suggestion 4 Total 41 Total 42 Total 10 Total 20 Possibly 3 Indeed 21 Primarily 4 Presumably 3 In fact 17 Technically 1 Probably 12 Inevitably 5 Typically 2 Sort of 3 Never 33 Usually 3 No doubt 1 Obviously 3 Of course 4 Really 20 Undeniably 1 Undoubtedly 3 Unquestionably 1 Total 54 Total 189 Total 44 Total 12 Table 5 : Stance adverbials in the 20 lowe st - s coring texts on Dimension 1. At the positive pole of Dimension 2, stance verbs controlling to - and that - clauses occur alongside mental verbs, that - deletions, 1 st person pronouns and past tense verbs. Together, these have been interpreted as indicating ‘Personal stance’. Genres that are not typically considered ‘academic’ - r eflective writing, legal Problem Questions discussing the implications of everyday situations, and Empathy Writing engaging with non - experts – tend to contain more of these features . Scores a re highest in the soft and applied disciplines ( Table 6). Disc ipline Score Philosophy (n=106) 8.24 Health (n=81) 7.60 Psychology (n=95) 5.51 Linguistics (n=115) 5.29 Classics (n=82) 4.55 Table 6 : Disciplines with the highest scores on Dimension 2 Stance verbs in the 20 highest - scoring texts on Dimension 2 are shown in Tables 7 and 8 , using the categories identified by Biber (2006:92) . Likelihood Certainty Communication Attitude Assume 1 Conclude 3 Argue 1 Agree 2 Believe 15 Discover 3 Claim 1 Ensure 2 Imagine 1 Find 6 Explain 4 Expect 1 Think 3 Know 3 Insist 1 Feel 38 Learn 8 Mention 1 Forget 1 Mean 9 Propose 1 Hope 5 Notice 4 Say 16 Wish 4 Observe 2 State 2 Worry 3 Realise 10 Suggest 1 Recognise 1 Tell 1 See 4 Show 7 Understand 1 Total 20 Total 53 Total 20 Total 56 Table 7: Stance verbs controlling that - clauses in the 20 highest - scoring texts on Dimension 2. P robability / C ognition / P erception Desire/ Intention / Decision Speech Act/ Communication Causation/ Modality / Effort Appear 3 Agree 2 Ask 3 Attempt 5 Believe 1 Choose 9 Be said 1 Enable 22 Expect 4 Decide 7 Teach 7 Encourage 9 Find 1 Hope 3 Fail 2 Forget 1 Intend 2 Help 31 Know 1 Like 6 Manage 5 Learn 8 Mean 3 Persuade 1 Seem 12 Need 37 Require 4 Tend 2 Plan 2 Try 29 Suppose 3 Prefer 3 Prepare 1 Want 78 Wish 3 Total 28 Total 156 Total 11 Total 108 Table 8: Stance verbs controlling to - clauses in the 20 highest - scor

3 ing texts on Dimension 2. At the po
ing texts on Dimension 2. At the positive pole of Dimension 2 stance features indicate a different register, more expressive of personal attitudes, desires and efforts. Stance verbs suit texts that are less like conventional university assignments because they appeal to readers at a more emotional level, and are essentially designed to persuade. These fi ndings indicate some distinct differences in the expression of stance across genres and disciplines, and provide evidence to support the more general claims made in prior studies. For example Hyland has argued that t he soft disciplines are “more inclined to explicitly recognise the ro le of human agency in constructing knowledge” (2002:124) and to take “ more involved and personal positions than those in the science and engineering fields” (2015:34). Epistemic nouns and adverbials are particularly prevalent in soft discipline E ssays , because argumentative claims in this register tend to be developed on the basis of certainty and likelihood (see, fo r example McEnery & Kifle 2002). Stance nouns which control that - clauses also enable arguments to unfold, by serv ing as ‘ ‘shell nouns’ ( Hunston & Francis 1999) or ‘signalling nouns’ (Flowerdew & Forest 2014) which label and encapsulat e information presented elsewhere in the text . Epistemic and attitude adverbials were strongly associated with spoken registers in Biber’s university language corpus (2006:103 - 4) , but in texts at the negative pole of Dimension 1 they ar e used to boost and hedge , signalling the writer’s direct involvement in the text. Again, this is typical of the soft disciplines where , in contrast to the ha rd sciences, there is “less control of variables, more diversity of research outcomes, and fewer clear bases f or accepting claims” (Hyland 2015:35). Our study has also shown that there is a distinctive register found in student writing that employs first p erson pronouns and ‘stance’ to - and that - clauses. Future studies could investigate the extent to which the personal stance registers of student writing a re also found in expert writing, perhaps extending beyond research writing to include texts produced in the professions. References Biber, D. (2006). University Language: A Corpus - based Study of Spoken and Written Registers . Amsterdam: John Benjamins. Flowerdew, J. & Forest, R. (2014). Signalling Nouns in English: A Corpus - Based Discourse Approach . Cam bridge: Cambridge University Press. Gardner, S. Nesi , H. & Biber, D. ( under review ) Discipline, level, genre: Integrating multiple situational perspectives in a new MD analysis of university student writing. Hyland, K. (2015) . Genre, discipline and identity. Journal of English for Academic Purposes 19 32 - 43. Hyland, K. (2002) . Activity and evaluation: Reporting practices in academic writing. In J. Flowerdew (Ed.), Academic Discourse (pp. 115 - 130) Harlow, Essex: Pearson Educatio n. Hunston, S. & Francis, G. (1999). Pattern Grammar . Amsterdam: John Benjamins. McEnery, T., & Kifle, N.A. (2002). Epistemic modality in argumentative essays of second language writers. In J. Flowerdew (Ed.), Academic Discourse (pp. 182 - 195) . Harlow, Esse x: Pearson Education . Nesi, H. & Gardner, S. (2012). Genres across the Disciplines: Student writing in higher education. Cambridge: Cambridge University Press. expert writing, perhaps extending beyond research writing to include texts produced in the University Language: Corpus-based Study of Spoken and Written & Forest, R. (2014). Signalling Nouns Corpus-Based Discourse Cambridge University Press. Gardner, S. & Biber, (under review) Discipline, in a new MD analysis (2015). Genre, discipline and identity. English for Academic Purposes(2002). Activity and evaluation: Reporting practices academic writing. Flowerdew (Ed.), Academic Discourse(pp. 115-130) Harlow, Essex: Pearson McEnery, T., & Kifle, N.A. (2002). Epistemic modality argumentative essays language writers. In J. Flowerdew (Ed.), Academic Discourse (pp. 182-195). Harlow, Essex: Pearson Education. Nesi, H. & Gardner, S. (2012). Genres across the Disciplines: Student writing in higher education. Cambridge: Cambridge University Press. P robability / C ognition / P erception Desire/ Intention / Decision Speech Act/ Communication Causation/ Modality / Effort Appear 3 Agree 2 Ask 3 Attempt 5 Believe 1 Choose 9 Be said 1 Enable 22 Expect

4 4 Decide 7 Teach 7 Encourage
4 Decide 7 Teach 7 Encourage 9 Find 1 Hope 3 Fail 2 Forget 1 Intend 2 Help 31 Know 1 Like 6 Manage 5 Learn 8 Mean 3 Persuade 1 Seem 12 Need 37 Require 4 Tend 2 Plan 2 Try 29 Suppose 3 Prefer 3 Prepare 1 Want 78 Wish 3 Total 28 Total 156 Total 11 Total 108 clauses in the 20 highest-scoring texts on Dimension At the positive pole of Dimension 2 stance features indicate a different register, more expressive of personal attitudes, desires and efforts. Stance verbs suit texts that are less like conventional university assignments because they appeal to readers at a more These findings indicate some distinct differences in the expression of stance across genres and disciplines, and provide evidence to support the more general claims made in prior studies. For example Hyland has argued that the soft disciplines are “more inclined to explicitly recognise the role of human agency in constructing knowledge” (2002:124) and to take more involved and personal positions than those in the science and engineering Epistemic nouns and adverbials are particularly prevalent in soft discipline Essays, because argumentative claims in this register tend to be developed on the basis of certainty and likelihood (see, for example McEnery & Kifle 2002). Stance nouns which control -clauses also enable arguments to unfold, by serving Hunston & Francis 1999) or ‘signalling nouns’ (Flowerdew & Forest 2014) label and encapsulate information presented elsewhere in the text. and attitude adverbials were strongly associated with spoken registers in Biber’suniversity language corpus (2006:103-4, but in texts at the negative pole of Dimension 1 they are used to boost and hedge, signalling the writer’s direct involvement in the text. Again, this is typical of the soft disciplines where, in contrast to the hard there is “less control of variables, more diversity of research outcomes, and fewer clear bases foOur study has also shown that there is a distinctive register found in student writing that employs first person pronouns and ‘stance’ and clauses. Future studies could investigate the extent to which the personal stance registers of student writing are also Possibly 3 Indeed 21 Primarily 4 Presumably 3 In fact 17 Technically 1 Probably 12 Inevitably 5 Typically 2 Sort of 3 Never 33 Usually 3 No doubt 1 Obviously 3 Of course 4 Really 20 Undeniably 1 Undoubtedly 3 Unquestionably 1 Total 54 Total 189 Total 44 Total 12 Table 5: adverbials in -scoring texts on At the positive pole of Dimension 2, stance verbs controlling clauses occur alongside mental verbs, -deletions, 1person pronouns and past tense Together, these have been interpreted as indicating ‘Personal stance’. Genres that are not typically considered ‘academic’ - reflective writing, legal Problem Questions discussing the implications of everyday situations, and Empathy Writing engaging with tend to contain more of these features. Scores are highest in the soft and applied disciplines (Table 6). Disc ipline Score Philosophy (n=106) 8.24 Health (n=81) 7.60 Psychology (n=95) 5.51 Linguistics (n=115) 5.29 Classics (n=82) 4.55 Table 6: Stance verbs the 20 highest-scoring texts Dimension 2 are shown Tables 7 and 8, using the Likelihood Certainty Communication Attitude Assume 1 Conclude 3 Argue 1 Agree 2 Believe 15 Discover 3 Claim 1 Ensure 2 Imagine 1 Find 6 Explain 4 Expect 1 Think 3 Know 3 Insist 1 Feel 38 Learn 8 Mention 1 Forget 1 Mean 9 Propose 1 Hope 5 Notice 4 Say 16 Wish 4 Observe 2 State 2 Worry 3 Realise 10 Suggest 1 Recognise 1 Tell 1 See 4 Show 7 Understand 1 Total 20 Total 53 Total 20 Total 56 clauses in 20 highest-scoring texts on qualitative means, manually examining high- and low-scoring texts to Stance adverbials and stance nouns controlling clauses clusterewith 3rd person pronouns, proper nouns and communication verbs at the negative pole of Dimension 1. Together these have been interpreted as indicating stance toward the of others’. Essays tend to have the lowest scores on thidimension, particularly in Arts and Humanities disciplines (Table 3). Discipline Score Comparative American Studies (n=74) - 6.78 History (n=95) - 6.89 English (n=106) - 7.38 Classics (n=82) - 10.53 Philosophy (n=106) - 10.67 with the scores on Dimensi

5 on Stance nouns controlling -clauses wer
on Stance nouns controlling -clauses were relatively rare in Biber’s university language corpus, with the exception of (Biber 2006:112) In texts with low scores on Dimension 1 most of these nouns an epistemic function, but some fall into the category of ‘Communication’ ornouns (Biber 2006:93). Stance nouns in the 20 lowest- lowest-scoring texts on Likewise stance adverbials mainly serve as epistemic devices, particularly to indicate certainty. Table 5 under the categories Likelihood Certainty Style Attitude Apparently 3 Actually 33 According to 9 Importantly 5 Arguably 3 Always 26 Clearly 13 Ironically 2 Evidently 3 Certainly 15 Frankly 1 Paradoxically 1 Kind of 13 Decidedly 2 Fundamentally 3 Surprisingly 1 Maybe 1 Doubtless 1 Generally 4 Tragically 2 Perhaps 10 Evidently 3 Mainly 4 Unfortunately 1 Likelihood Certainty Communication Attitude Belief 9 Certainty 2 Argument 1 Fear 4 Assumption 2 Conclusion 3 Confession 1 Feeling 1 Claim 1 Fact 30 Explanation 1 Hope 1 Hint 1 Precept 1 Interpretation 1 Illusion 1 Hypothesis 1 Principle 1 Intimation 1 Indication 2 Idea 11 Proof 1 Objection 1 Understanding 2 Interpretation 1 Realization 1 Proposal 1 View 9 Intimation 1 Revelation 2 Proposition 3 Notion 3 Statement 1 Possibility 1 Premise 3 Presumption 1 Sign 2 Suggestion 4 Total 41 Total 42 Total 10 Total 20 Hilary Nesi and Sheena Gardner (CoventryUniversity, UK) The British Academic Written English (BAWE) corpus contains about 6.5 million words of proficient university student writing, categorised in terms of ‘genre families’ distributed fairly equally across levels of study and disciplinary groupings. The corpus has been examined from various perspectives, including multidimensional analysis (MDA) (Nesi & Gardner 2012) using Biber’s 1988 dimensions. This analysis revealed clear between texts produced by first, second and final year undergraduates and Masters students, and between texts belonging to different disciplinary groupings. In terms of stance, epistemic modality was the most discriminating feature, being considered a tool for persuasion on Dimension 4. Published academic prose from the Lancaster-Oslo-Bergen (LOB) Corpushad a neutral score on Dimension 4 in Biber’sown 1988 study, but scores for BAWE assignments were negative on Dimension 4, and became increasingly so across levels of study (Table 1). Scores for Arts and Humanities Level Score 1 st year undergraduate - 1.4 2 nd year undergraduate - 1.4 3 rd year undergraduate - 1.5 Masters - 2.0 Table 1: BAWE Dimension 4 scores by level Disciplinary Grouping Score Physical Sciences - 1.2 Social Sciences - 1.3 Life Sciences - 1.5 Arts and Humanities - 2.3 by disciplinary grouping Biber’s 1988 dimensions still have validity, because the results from studies which employ them can easily be compared. For example, the 1988 dimension scores for BAWE Biber’s LOB subcorpus can be taken as an interesting indicator of differences between student and expert writing. Student writing below PhD level appears to have fewer overtly persuasive features than expert writing, perhaps because undergraduate and Masters students tend to discuss the work of others rather than presenting their own original research. Stance theory has developed considerably since 1988, and features have now been added to Biber’s tagset, enabling clearer distinctions to be made between registers which differ predominantly in terms of stance. Biber (2006) uses this tagset as the basis for a detailed linguistic description of registers encountered by university students, such as classroom teaching, coursepacks and institutional writing, but does not include any analysis of texts (such as assignments) produced by students themselves. A new multidimensional analysis of BAWE (BAWE2016) has built on this work, using the extended tagset to create new dimensions specific to university student writing. These bespoke dimensions distinguish between BAWE texts grouped by discipline and genre family, without reference to registers in other corpora. Following MDA methodology, 39 linguistic features were retained and four new factors were identified (Gardner, Nesi & Biber, under review), to be treated as dimensions along which the BAWE registers could be situated. Two of the four new dimensions depend strongly on stance features. The variation along each dimension