XXII Открытая конференция студентов-филологов в СПбГУ

The Effect of Visual Noise on Good-Enough Language Processing in Reading: an Eye-Tracking Study

Анастасия Дмитриевна Антонова
Докладчик
магистрант 1 курса
Национальный исследовательский университет «Высшая школа экономики»

184
2019-04-17
16:40 - 17:00

Ключевые слова, аннотация

Good-enough language processing is a theory in psycholinguistics that says that people are more likely to rely on lexical-semantic heuristics than algorithmic syntax processing in situations of bigger cognitive load. My work is concerned with studying linguistic noise as a factor capable of increasing the degree of reliance on GE-processing. In order to understand how much this noise influences GE-processing, I have conducted two experiments on the material of syntactically complex unambiguous sentences (in Russian).

Тезисы

Good-enough language processing is a theory in psycholinguistics that says people are more likely to rely on lexical-semantic heuristics and language experience than algorithmic syntax processing in situations of bigger cognitive load. Unlike deep algorithmic processing, this mechanism can lead to incomplete or erroneous understanding of the sense of a phrase, such as interpretation of the word dog as a subject in a phrase The dog was bitten by a man. Despite the fact that there are many works on GE-processing and this research topic is attracting increasingly more attention, the reasons why people rely on GE-processing to such an extent are still not studied. My work is concerned with studying the linguistic visual noise as a factor capable of increasing the degree of reliance on GE-processing.
In order to understand how much this noise influences GE-processing, I have conducted two experiments (with and without noise) on the material of syntactically complex unambiguous sentences that contain an adjunct that is attached to a complex noun group and is expressed by a participle clause (in Russian). These sentences are divided into «plausible» (Я поздоровался с учителем школьника, держащим в руках указку) and «implausible» (Я поздоровался с учителем школьника, одетым в школьную форму) depending on their meaning. The correctness of interpretation of their meaning enables me to assess the degree of reliance on GE-processing. The participants were 22 neurologically healthy people, Russian native speakers (age: 18—38 years, average 23 years, mean 4,7 years, 4 males and 18 females, 19 right-handed and 3 left-handed). In my research, I applied eye-tracking method as ecologically valid and the most precise. For the data analysis I used statistical analysis in R and repeated measures ANOVA in SPSS.
The following variables have proven to be statistically significant: plausibility for dwell time (p=0.027); visual noise for dwell time participle fixations (p=0.031); noise for accuracy (p=0.0002).
People spent more time for reading implausible sentences which is quite predictable. What is more, they were more attentive while reading participle clause in noise conditions as they spent more time reading them. The most surprising thing was that participants were more accurate in interpretation sentences in noise conditions. This fact allows us to discuss the future plans for conducting experiments to compare different types of noise.
The study was supported by the RFBR grant, the project №18-012-00640.