Leksikosyntaktiske trekk og skriveverktøy. En kvantitativ undersøkelse av tekster skrevet for hånd og på tastatur av elever i VG1
Not peer reviewed
MetadataShow full item record
This PhD thesis investigates lexicosyntactic features in writings in Norwegian L1 by 60 16- year-old pupils written using two different writing tools: hand-writing and typing. Its methodological approach is corpus-based and statistical, and its theoretical foundations are mainly those of theories of complexity and register variation.
The main focus of the thesis is on the analysis of 5 lexical and 8 syntactic variables. The lexical variables are average word-length, average word-length in lexical words, lexical density, global TTR and local TTR, both adjusted or neutralised for text-length. The syntactic variables are t-unit length, clause length, frequency of short subclauses, number of prepositional phrases per clause, number of adverbial subclauses per clause, number of subclauses per t-unit, ratio of t-units with a short frontal constituent and frequency of attributive adjectives. All these are analysed using anova on four pupil features, which are gender, general writing skills, total text length of the two texts and ratio of text lengths in the two texts. Ten of the variables are included in an overall principal component analysis.
The main findings come in four general categories:
Some variables seem unaffected by the writing tool. This applies to average clause length and frequency of attributive adjectives.
Some variables display a shift in the direction of more spontaneous, "oral" features in the typed texts. This applies to subclause frequency and perhaps average t-unit length.
Some variables have different properties in boys' and girls' writings; they display more spontaneous features in the typed texts written by girls and more planned or edited features in the typed texts written by boys. This applies to average word-length, average word-length in lexical words, global TTR and ratio of short frontal constituents.
Some variables have different properties in the writings of pupils who write considerably longer texts when typing compared to the writings of pupils who write texts of approximately equal lengths with both writing tools. The former pupils have more planned or edited features in their typed texts, whereas the latter pupils have more spontaneous features in their typed texts. This applies to local TTR, global TTR and ratio of adverbial clauses.
The principal component analysis confirms the two effects splitting the sample of pupils in terms of gender and production length.
In addition, some variables display types of patterns which do not fit neatly into any of the four categories above, namely average t-unit length (longer in typed texts), ratio of short subclauses (complex interactions of several pupil-related factors), lexical density (interactions of two pupil-related factors), frequency of prepositional phrases (higher in typed texts by productive writers, lower in typed texts by terse writers), and ratio of short frontal constituents (interactions between difference in text length and gender).
An important goal of the project was the further development of methods for this kind of study based on statistical stylistics. As part of the study, several lexical variables are constructed, compared and evaluated with respect to reliability and validity, among these entropy-based lexical distribution measures and frequency-based measures termed logarithmic frequency index. Also, emphasis has been put on investigating what relevant and valid textual or linguistic features are possible to extract from a corpus using a combination of automatic procedures and a limited extent of manual procedures, namely the manual segmentation of the texts into t-units and clauses, and correction of orthography and punctuation.