Idioma: Español
Fecha: Subida: 2021-04-10T00:00:00+02:00
Duración: 15m 24s
Lugar: Conferencia
Visitas: 981 visitas

How is information content distributed in RA introduction moves across disciplines? (...)

Jin Liu and Wei Xiao (Chongqing University)

Descripción

Research article (RA thereafter), serving as the central genre of knowledge
production and dissemination and the key medium for the legitimating of claims and
disciplines, has been actively studied in the last few decades. Researchers have
dedicated themselves to the study of the different subsections of RA encompassing
abstracts, method sections, discussions and results. Among a variety of detailed
analysis in almost all sections of the research articles, introductions, being a “crafted
rhetorical artifact” and a “manifestation of rhetorical maneuver” (Swales, 1990: 155),
have been given special attention especially after Swales’ (1990) proposal of CARS
model for introductions. Most of the previous studies in this regard were based on
qualitative methods, mainly focusing on the identification of the move structures of
certain disciplines, the rhetorical functions performed by the moves as well as the
linguistic realizations, but there were relatively few quantitative investigations to
reveal the rhetorical organization of RA introductions and their disciplinary variations.
Besides, although texts are considered as the realization of the process of mediating
messages, disseminating knowledge and conveying information, there have been few
attempts to investigate the information content of the texts from the perspective of the
information theory and little to none research delved into the exploration of the
informative distribution of RA introduction moves.The current study employed an
entropy-based approach to investigate the distributional patterns of RA introduction
moves in terms of information content and their variations across disciplines (natural
sciences, social sciences and humanities). Three indices, i.e. 1-/ 2-/ 3-gram entropies,
were used to analyze 120 RA introductions (40 introductions from each discipline).
All samples were annotated for rhetorical moves using an adapted version of the
CARS model (Swales, 2004) and the information content of each move was calculated in terms of 1-gram, 2-gram, and 3-gram entropies through a home-made R
script.The results revealed that rhetorical moves in RA introductions, given any
discipline, tended to show a similar unevenly distributional pattern in terms of
information content, with the information content of Move 1 being the highest, Move
2 being the lowest and Move 3 lying in the middle position. Meanwhile, the entropies
of the three different grams shared a resembling distribution across disciplines, though
the information content of the same move measured by 1-gram entropies was the
highest, followed in sequence by 2-gram and 3-gram entropies. Further explorations
revealed that while insignificant variation across disciplines was found in Move 1 and
Move 2, there was a significant difference between natural sciences and social
sciences in Move 3. The unevenly distributed pattern of information content could be
explained by the different rhetorical functions of moves and distinct disciplinary
features.
The current study is significant in the following aspects. Firstly, this study has
widened the applications of quantitative linguistic methods in RA introductions to
reveal the distinctive features of this specific genre. Secondly, this study also has
pedagogical implications for RA introduction writing instruction and practice. In
addition, the present study may make contributions to the automatic move recognition
of RA introductions in the field of NLP (natural language processing) by introducing
an entropy algorithm, thus improving the accuracy of a move recogniser. It is believed
that such a recogniser, if widely used, will be conducive to academic writing and
practice.

Propietarios

Congreso Cilc 2021

Comentarios

Nuevo comentario

Serie: CILC2021: Lingüística computacional basada en corpus / Corpus-based computational linguistics (+información)