The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

an overview of methodology and findings

(Biber, D., 1992)

Tobias Anton

Corpus Linguistics

05.06.2019

Course: 8.3409

Tobias Anton

  1. Intro & Motivation
  2. Methodology
    - Data preparation
    - Factor Analysis
  3. Findings
    - e.g. genres over time
    - statistical extension
  4. Critique / Discussion
    - ...on the basis of a sloppy reproduction

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Schedule

Introduction

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

  • genres
  • register / dialect
  • historical development
  • languages

How to find meaningful structure?

The multidimensional approach aims to find meaningful underlying structure to explain linguistic variation across several domains like:

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Basic assumptions of the model:

  1. it is corpus based (seems 'twas important to note in the days of yore)
  2. it's computer based, thus enabling distributional analysis
  3. aims for linguistic/statistical analysis, not individual stylistic constructions
  4. different kinds of text differ linguistically
  5. multiple parameters of variation are operative in any domain
  6. it's a quantitative approach
  7. quantities are interpreted functionally
    (statistical co-occurence implies shared functionality)
  8. it combines macro- and microscopic approaches (e.g. variation across all genres as well as within specific ones)

Introduction

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

basic assumptions of the model:

  1. it is corpus based (seems 'twas important to note in the days of yore)
  2. it's computer based, thus enabling distributional analysis
  3. aims for linguistic/statistical analysis, not individual stylistic constructions
  4. different kinds of text differ linguistically
  5. multiple parameters of variation are operative in any domain
  6. it's a quantitative approach
  7. quantities are interpreted functionally
    (statistical co-occurence implies shared functionality)
  8. it combines macro- and microscopic approaches (e.g. variation across all genres as well as within specific ones)

Introduction

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

basic assumptions of the model:

5. multiple parameters of variation are operative in any domain

no single dimension is adequate in itself to account for the range of linguistic variation in a language

- Douglas, 1992

Introduction

Methodology

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Methodology

Biber (1992) used:

  • LOB- and London-Lund Corpora
    - LOB: 500 texts / 2000 words each
    - LL: 100 (spoken) texts / 5000 words each
  • CLAWS tagging system
  • ultimately resulting in:
    - 481 texts (written & spoken)
    - 67 linguistic features
    - 23 major genre categories

(and small recap)

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Selected examples of Feature-Classes:

  • tense and aspect markers
  • questions
  • lexical specifity
  • pronouns and proverbs
  • place and time adverbials
  • passives
  • negation
  • modals

Methodology

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Methodology

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Dimension composition in detail:

Dimension 2: Narrative vs. Non-narrative

positive loadings:

  • past tense verbs
  • third person pronouns
  • perfect aspect verbs
  • public verbs
  • synthetic negation
  • present participle clauses
0.90
0.73
0.48
0.43
0.40
0.39

negative loadings:

  • present tense verbs
  • attributive adjs.
-0.47
-0.41

Methodology

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Dimension composition in detail:

Dimension 2: Narrative vs. Non-narrative

positive loadings:

  • past tense verbs
  • third person pronouns
  • perfect aspect verbs
  • public verbs
  • synthetic negation
  • present participle clauses
0.90
0.73
0.48
0.43
0.40
0.39

negative loadings:

  • present tense verbs
  • attributive adjs.
-0.47
-0.41

Factor/Dimension labels are highly interpretative!

Methodology

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

5 basic dimensions of variation

  1. Informational vs Involved
    + nouns, word length
    - private verbs, that deletions, contractions
  2. Narrative vs Non-narrative
  3. Elaborated vs Situated Reference
    + WH relative clauses on sub- & object positions
    - time & place adverbials
  4. Overt Expression of Persuasion (Argumentation)
    + infinitives, prediction modals
  5. Abstract vs Non-abstract Style
    + conjuncts, agentless passives, BY-passives

Methodology

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Methodology

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Methodology

Findings

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Findings

Grabe (1987):

  • investigation between genres and audience level
    - genres: social science, natural science, humanities...
    - audience: academic, introductory university, popular
  • four dimensions identified
  • associations / differences found among both
    ==>implying linguistic importance

Findings

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Biber (1991)

Analysis of primary school reading texts

  • texts are "notably different from adult genres" ¯\_(ツ)_/¯
  • also huge differences among each other
  • 2nd grade science texts are more
    - involved and interactive (dim 1)
    - situated (dim 3)
    than more advanced science texts
  • but also more involved & interactive than basal reading texts

primary school science text are probably no good marker for scientific literacy

Findings

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Biber & Finnegan (1992)

Study on historical development of English genres

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

What is (a) genre?

text varieties that are readily recognized and "named" within a culture.

opposed to text types, which are

text varieties that are defined linguistically

Findings

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

What is (a) genre?

text varieties that are readily recognized and "named" within a culture.

opposed to text types, which are

text varieties that are defined linguistically

categorized perceptually
(not necessarily linguistically coherent)

categorized by linguistic dimension
(coherent by definition)

Findings

Findings

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Statistical extension

Cluster Analysis on the basis of dimension scores (Biber, 1989):

  • identifies Intimate Interpersonal & Informational Interaction
    (instead of a single interactive/dialog type)
  • no single expository type, but...
    - Scientific Exposition; informational, elaborated, abstract
    - Learned Exposition; less abstract and technical
    - General Narrative Exposition; informational elaboration
  • no single narrative type:
    - General Narrative Expostion
    - Imaginative Narrative; rather involved

Findings

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

MD Analysis of other languages

Dimensions in other languages convey different function

Kim & Biber (1993) analyzed Korean and found:

  1. Informational Interaction versus Explicit Elaboration
  2. Discourse Chaining versus Discourse Fragmentation
  3. Stance
  4. Narrative Concern
  5. Honorification

Discussion / Critique

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Roundup

  • Distributional analysis of co-occuring linguistic features across a wide range of genres
  • Identification of underlying linguistic and textual constructs
  • Model/Method can be used to address sociolinguistic and stylistic issues regarding their respective linguistic variation

Discussion / Critique

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Discussion

  • What do you think about this approach?
    • are the assumptions justified?
  • ...or the model?
    • are the dimensions appropriate?
    • are they sufficient?
  • Any (b)ad-hoc hypotheses interesed in to able to test with this?
    • Is InfoWars actually waging war on information?
    • How do hi-fantasy, satirical sci-fi and scientific literature differ?

Discussion / Critique

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Discussion

no single dimension is adequate in itself to account for the range of linguistic variation in a language

- Douglas, 1992

Discussion / Critique

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Discussion

no single dimension is adequate in itself to account for the range of linguistic variation in a language

- Douglas, 1992

Discussion / Critique

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Discussion

Discussion / Critique

The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

Course: 8.3409

Corpus Linguistics

Critique

  • They don't mention any statistics regarding the dimensions
    - Variance explanied per dimension / in total
    - How they dealt with unclear loadings (if at all)

MDA to genre analyses

By Tobias Anton

MDA to genre analyses

Corpus Linguistics (SoSe'19): The Multi-Dimensional Approach to Linguistic Analyses of Genre Variation

  • 495