A coherent piece of writing is one where the parts connect to the whole. That sounds obvious, but it can be difficult to achieve. That’s partly because the whole is surprisingly easy to lose sight of, especially in long texts where the writing happens across days or weeks or months. The whole is never entirely visible at the same time.

Coherence is sometimes confused with cohesion. Cohesion is surface-level stitching. Stitching a text together means using meta-commentary that guides, foreshadows, maps and links. Stitching a text together can also mean using transition words and phrases, pronouns that refer back to earlier nouns, sentence openings that pick up the thread of the previous sentence. This is important for the reader because clumsy cohesion makes reading hard work. 

But, and always with the but, you can have excellent cohesion and still produce an incoherent piece of writing. You can produce a text where every paragraph flows gracefully into the next. It won’t cohere if the paragraphs are not all pulling in the same direction. Coherence is structural. It is about whether the argument has a spine, sometimes called a red thread.

The spine holds the text together. The spine of academic writing is the central claim or question that the piece is organised around and the argument that this affords. In a journal article, the claim or question is usually stated explicitly near the end of the introduction with the follow on – this is what the paper argues, this is what it shows. Every section of the paper then does something in relation to that claim – establishing its context, providing evidence, complicating or qualifying it, drawing out its implications. A section that does no work for the argument creates a coherence problem, regardless of how well it is written in itself. 

So the coherence test is not “is this good material?” but “what is this doing here in relation to what this piece is trying to do?”

Doctoral writers can run into coherence problems at predictable points. One is the literature review that becomes a laundry list. The writer moves through their reading source by source, summarising what each one says, and the review demonstrates wide reading without ever committing to a line of argument that connects back to the study’s own question. Another is the chapter that was written early in the doctorate and never revised to fit what the thesis became. The argument in that chapter was coherent with a different project and it now sits slightly at an angle to everything around it. The reader can feel the tilt without always being able to name it. A third point is the discussion section that raises new ideas, concepts or literatures that did not appear in the theoretical framework , perhaps because the writer found them interesting or the data seemed to call for them. The discussion develops and extends what the thesis set up, it doesn’t start a new conversation.

Coherence problems come from the gap between the whole and the parts. The writer has been attending to each paragraph or section, and has lost track of what it is doing in the larger structure. This is easy to understand as a process problem. When you are deep in a chapter, the chapter is the world. The question “how does this chapter connect to my central argument?” requires you to step back to a level of abstraction that can be hard to maintain when you’re also trying to work out what to say next.

So what to do to get coherence? It’s helpful to develop the habit of checking the whole against the particular part at regular intervals. One strategy is to backward map.  Write a one-sentence description of what each paragraph does – not what it says, but what it does in the argument  –  and then read those sentences in sequence. If the sequence makes sense as a compressed version of the argument, the coherence is probably there. If it doesn’t, or if some sentences just describe content, the writing lacks a whole text orientation.

Revision is where coherence is usually fixed, because coherence problems can be pretty hard to see until the whole is on the page. This is one reason why writing advice is often to write before you are ready- an incoherent draft gives you something you can then examine for coherence. You can work on getting the text to be coherent.

In sum, coherence is the result of having a clear argument. If you know what you are arguing, you can ask every part of the text whether it is serving that argument. If you don’t yet know what your argument is and are writing to sort it out, the text will reflect your uncertainty, however polished the individual components are. If you know your argument isn’t sorted, it’s sensible to look next for coherence of the whole text, rather than continuing to play with parts.