{"title":"Scaling as method: A three-stage, mixed-methods approach to digital discourse analysis","authors":"Jannis Androutsopoulos","doi":"10.1016/j.dcm.2024.100817","DOIUrl":null,"url":null,"abstract":"<div><div>Drawing on research on graphic contextualization cues in punctuation and typography, this paper describes a three-stage, mixed-methods approach to digital discourse analysis. It introduces the terms ‘scale’ and ‘scaling’ as methodological metaphors for a researcher’s planned, yet contingent movement through formations of digital textual data that differ in terms of volume, method of collection, processing, and analysis. ‘Scaling-as-method’ aims to replace static binaries (such as ‘micro’ and ‘macro’, ‘small’ and ‘big’ data, ‘manual’ and ‘automated’ processing) by the vision of a researcher who shifts their degree of abstraction, or ‘distance’, towards digital data, while moving from close to distant reading and back again. The paper exemplifies this three-stage process on the example of the indignation mark, aka <!!1>, a twist on the iterated exclamation mark that is attested in digital discourse in various languages as a cue of double-voicing. The explorative examination of a small dataset (Stage 1) leads to the computational collection and distributional analysis of a much larger dataset (‘scaling up’, Stage 2), followed by the manual annotation of a selected subset of this data (‘scaling down’, Stage 3). Each stage draws on a different amount of data, which enables different techniques of processing and analysis, and relies on a specific combination of abductive, deductive, and inductive reasoning. Yet all three stages complement one another in a kaleidoscopic way towards understanding connections between punctuation practices and participatory political discourse online. Scaling as method is not a closed recipe, but an adaptable procedure that can be applied to a variety of discrete digital features. It does not aim to replace established methods of computational social media analysis, but to boost research that is predominantly based on the manual collection and annotation of social media data, and to enables a dialogue between multiple understandings of context.</div></div>","PeriodicalId":46649,"journal":{"name":"Discourse Context & Media","volume":"62 ","pages":"Article 100817"},"PeriodicalIF":2.3000,"publicationDate":"2024-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Discourse Context & Media","FirstCategoryId":"98","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2211695824000631","RegionNum":2,"RegionCategory":"文学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMMUNICATION","Score":null,"Total":0}
引用次数: 0
Abstract
Drawing on research on graphic contextualization cues in punctuation and typography, this paper describes a three-stage, mixed-methods approach to digital discourse analysis. It introduces the terms ‘scale’ and ‘scaling’ as methodological metaphors for a researcher’s planned, yet contingent movement through formations of digital textual data that differ in terms of volume, method of collection, processing, and analysis. ‘Scaling-as-method’ aims to replace static binaries (such as ‘micro’ and ‘macro’, ‘small’ and ‘big’ data, ‘manual’ and ‘automated’ processing) by the vision of a researcher who shifts their degree of abstraction, or ‘distance’, towards digital data, while moving from close to distant reading and back again. The paper exemplifies this three-stage process on the example of the indignation mark, aka <!!1>, a twist on the iterated exclamation mark that is attested in digital discourse in various languages as a cue of double-voicing. The explorative examination of a small dataset (Stage 1) leads to the computational collection and distributional analysis of a much larger dataset (‘scaling up’, Stage 2), followed by the manual annotation of a selected subset of this data (‘scaling down’, Stage 3). Each stage draws on a different amount of data, which enables different techniques of processing and analysis, and relies on a specific combination of abductive, deductive, and inductive reasoning. Yet all three stages complement one another in a kaleidoscopic way towards understanding connections between punctuation practices and participatory political discourse online. Scaling as method is not a closed recipe, but an adaptable procedure that can be applied to a variety of discrete digital features. It does not aim to replace established methods of computational social media analysis, but to boost research that is predominantly based on the manual collection and annotation of social media data, and to enables a dialogue between multiple understandings of context.