Login (DCU Staff Only)
Login (DCU Staff Only)

DORAS | DCU Research Repository

Explore open access research and scholarly works from DCU

Advanced Search

Simplifying, reading, and machine translating health content: an empirical investigation of usability

Rossetti, Alessandra orcid logoORCID: 0000-0002-2162-9639 (2019) Simplifying, reading, and machine translating health content: an empirical investigation of usability. PhD thesis, Dublin City University.

Abstract
Text simplification, through plain language (PL) or controlled language (CL), is adopted to increase readability, comprehension and machine translatability of (health) content. Cochrane is a non-profit organisation where volunteer authors summarise and simplify health-related English texts on the impact of treatments and interventions into plain language summaries (PLS), which are then disseminated online to the lay audience and translated. Cochrane’s simplification approach is non-automated, and involves the manual checking and implementation of different sets of PL guidelines, which can be an unsatisfactory, challenging and time-consuming task. This thesis examined if using the Acrolinx CL checker to automatically and consistently check PLS for readability and translatability issues would increase the usability of Cochrane’s simplification approach and, more precisely: (i) authors’ satisfaction; and (ii) authors’ effectiveness in terms of readability, comprehensibility, and machine translatability into Spanish. Data on satisfaction were collected from twelve Cochrane authors by means of the System Usability Scale and follow-up preference questions. Readability was analysed through the computational tool Coh-Metrix. Evidence on comprehensibility was gathered through ratings and recall protocols produced by lay readers, both native and non-native speakers of English. Machine translatability was assessed in terms of adequacy and fluency with forty-one Cochrane contributors, all native speakers of Spanish. Authors seemed to welcome the introduction of Acrolinx, and the adoption of this CL checker reduced word length, sentence length, and syntactic complexity. No significant impact on comprehensibility and machine translatability was identified. We observed that reading skills and characteristics other than simplified language (e.g. formatting) might influence comprehension. Machine translation quality was relatively high, with mainly style issues. This thesis presented an environment that could boost volunteer authors’ satisfaction and foster their adoption of simple language. We also discussed strategies to increase the accessibility of online health content among lay readers with different skills and language backgrounds.
Metadata
Item Type:Thesis (PhD)
Date of Award:November 2019
Refereed:No
Supervisor(s):O'Brien, Sharon
Uncontrolled Keywords:Readability; Comprehensibility; Translatability
Subjects:Computer Science > Machine translating
Humanities > Linguistics
Humanities > Translating and interpreting
DCU Faculties and Centres:DCU Faculties and Schools > Faculty of Humanities and Social Science > School of Applied Language and Intercultural Studies
Research Initiatives and Centres > Centre for Translation and Textual Studies (CTTS)
Use License:This item is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 3.0 License. View License
Funders:Faculty of Humanities and Social Sciences at Dublin City University, Irish Research Council Government of Ireland Postgraduate Scholarship Programme., International Network on Crisis Translation (INTERACT), European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 734211.
ID Code:23124
Deposited On:20 Nov 2019 12:32 by Sharon O'brien . Last Modified 05 Feb 2020 12:28
Documents

Full text available as:

[thumbnail of 15211745_Alessandra Rossetti_April 2019_final.pdf]
Preview
PDF - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
2MB
Downloads

Downloads

Downloads per month over past year

Archive Staff Only: edit this record