Hypothesis 3 predicted that both corpora would provide more dependent noun clauses than other types of dependent clauses and is confirmed within the interview corpus but not within the linguistic corpus, which contained an even number of adverbial, noun and relative clauses. Hypothesis 4 predicted that the extract taken from a transcribed interview would contain more contractions than the linguistic extract, and is borne out by the results.
This study compares two small corpora, each comprising 300 words, taken from a linguistic journal and a transcribed interview. The texts are therefore diverse in terms of content, style, register and their proposed audience, and are compared and contrasted in terms of the linguistic properties pertaining to the number and length of sentences, the number and type of dependent clauses, and the use of contractions.
The focus on sentence constructions begins by ascertaining the number of sentences within each corpus. The purpose of this is that the number of sentences will provide insight as to the length of the sentences, which is measured in terms of the number of words within a punctuated sentence. The higher the number of sentences found within a 300 word corpus, then the shorter the sentences would have to be. The fewer the number of sentences therefore, would realise longer, compound and complex sentences and thus would be expected to render more dependent clauses.
The second property to be measured in this study is the number and type of dependent clauses used within each text. A dependent clause does not convey a complete idea and therefore is unable to stand by itself; in other words a dependent clause relies on an independent clause for meaning. Tallerman (2005) refers to dependent clauses as subordinate clauses and the independent clause, which must contain a finite verb, as the matrix clause. She also tells us that subordinate clauses are often considered as embedded clauses because