Annotating Appraisal in English and Spanish product reviews from mobile application stores: a contrastive study for linguistic and computational purposes
Evaluation, opinion and subjectivity are related phenomena which are currently receiving attention both in the linguistic and the computational communities. One of the most influential theories that deal with the phenomenon of evaluation and subjectivity is the ‘Appraisal’ framework, which proposes...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis doctoral |
| Fecha de publicación: | 2018 |
| País: | España |
| Institución: | Universidad Complutense de Madrid (UCM) |
| Repositorio: | Docta Complutense |
| Idioma: | inglés |
| OAI Identifier: | oai:docta.ucm.es:20.500.14352/15670 |
| Acceso en línea: | https://hdl.handle.net/20.500.14352/15670 |
| Access Level: | acceso abierto |
| Palabra clave: | 81'322(043.2) Computational linguistics Lingüística computacional Lingüística 57 Lingüística |
| Sumario: | Evaluation, opinion and subjectivity are related phenomena which are currently receiving attention both in the linguistic and the computational communities. One of the most influential theories that deal with the phenomenon of evaluation and subjectivity is the ‘Appraisal’ framework, which proposes that linguistic expressions of evaluative meanings such as emotion, attitude and opinion can be divided into three different axes: Attitude, Engagement and Graduation. To date, there is no large-scale cross-linguistic work in the computational or in the linguistic communities which has set out the task to validate the different features of Appraisal Theory empirically through corpus annotation, and to investigate its application to a relatively new genre, namely, mobile application reviews The work developed in this dissertation is an attempt at validating aspects of Appraisal Theory in a contrastive manner (i.e. comparing English and Spanish), and at providing a cross-linguistic characterisation of this new review genre in terms of Appraisal features. The categories, empirically validated with a high degree of inter-annotator agreement, are used to annotate a larger bilingual corpus. The results showed interesting language-specific differences, such as a higher degree of straightforward strategies in the Spanish texts in opposition to a higher modulation of meanings in the English ones. In addition, some tendencies may be pointed out among different products (applications and games, vs. books and films) as well as specific aspects that are typically included in negative reviews (Pseudo-Questions and Judgement) or in positive reviews (Capacity). |
|---|