—through the analysis of larger patterns and cohesive ties
For discourse analysis, we usually analyze two main categories of discourse, the spoken discourse and written discourse. When we analyze a piece of spoken discourse, we will exam the identify of the speaker, the purpose of the utterance, the perlocutionary effect of the utterance, and the context of the utterance. Elements like intonation, tone, and genre of the utterance also are included in the analysis. We can get a close look by the speaking mode presented by Dell Hymes's SPEAKING model. The SPEAKING model is a method to analyze a piece of utterance, in which we, through the analysis of Setting and Scene, Participants, Ends (Purposes, goals, and outcomes), Act Sequence, Key(Cues that establish the "tone, manner, or spirit" of the speech act), Instrumentalities(Forms and styles of speech), Norms, Genre, deconstruct a discourse.
As for the written discourse, it is, in some sense, much the same. A text is written within a certain context, aimed at specific readers. The writer’s purpose is realized with the structure and vocabulary used. This can be exemplified by the following list of everyday written texts; instruction leaflet, letter to/from friend, public notice, product label, newspaper obituary, poem, news report, academic article, small ads, postcard to/from friend, business letter. Though the analyses of written and spoken discourse are the same in some sense, they rely on different methods for exact interpretation. For spoken language, we can rely on visual and aural clues. The clues for written texts are not so obvious. But as the writer produced texts for particular purposes, he or she would employ difference structures. Connections between sentences and ideas are possible because all texts have structure. This structure is created through an overall textual pattern, lexical signals, inter-clause relations, and lexical and grammatical cohesive links.