Three scales, three different cues
Signposting operates at the document, section, and sentence level, and confusing them produces either an over-signposted paper or one a reader cannot navigate. At document level you need a roadmap in the introduction and a brief orientation at each major section boundary. That is all — a paper that announces its own structure every few pages reads as anxious.
At section level, the work is done by topic sentences. A reader should be able to skim the first line of each paragraph and reconstruct the argument. This is not a hypothetical reading style: examiners, reviewers, and anyone deciding whether to cite you all read this way at least once.
At sentence level, cohesion comes from how consecutive sentences share information. English prose reads most smoothly when each sentence starts with something already familiar and ends with what is new, so the new item is then available to open the following sentence. Most prose that feels choppy is violating this pattern rather than lacking connectives.
Transitions that state a relationship
Generic connectives are the most common form of empty signposting. “Furthermore, the second experiment used a larger sample” tells a reader that another sentence has arrived, which they could see. Compare: “Because the first experiment was underpowered, the second used a larger sample.” Same information, and now the reader knows why the second experiment exists.
The same repair applies across the whole family. “In addition, participants were interviewed” becomes “To capture reasons the survey could not, participants were also interviewed.” “However, the results differed” becomes “The results differed in the older cohort, where the effect reversed.” In each case a word that marked a gap is replaced by a phrase that explains it.
This is why a long list of transition words is poor advice. The problem is almost never a missing connective; it is a missing relationship. If you cannot say how two paragraphs relate, adding “Moreover” papers over a structural problem instead of solving it.
- “Furthermore” → “Because X, we then …”
- “In contrast” → “Where X holds, Y does not, because …”
- “Additionally” → “The same pattern appears in …, which suggests …”
The topic sentence does most of the work
If you only fix one thing, fix your paragraph openings. A topic sentence that states a claim commits the paragraph to supporting exactly that, which disciplines the writing as much as it helps the reader. Paragraphs that open with a topic instead of a claim tend to wander, because nothing in them has been promised.
A test worth running on a finished draft: for each paragraph, ask whether the opening sentence and the final sentence are about the same thing. Where they are not, the paragraph drifted, and either it needs splitting or the opening needs rewriting to match what the paragraph actually ended up arguing.
The related check is against your outline. Every topic sentence should correspond to a point you planned to make. Sentences with no counterpart in the outline are usually the paragraphs that grew during drafting and never got a place in the argument.
How much signposting is too much
Over-signposting is a real failure, and it is common in theses, where writers are anxious about being followed. Prose in which every section opens with “In this section, I will …” and closes with “In this section, I have …” spends a large share of its length describing itself rather than arguing. Examiners find it tiring.
The proportion that works: heavy at the top of the document and at major boundaries, light inside a well-structured section, and none at all where the known-new link already carries the reader. If two paragraphs are clearly consecutive steps in an argument, they need no connective, and adding one interrupts rather than assists.
A quick diagnostic: count how many sentences in a section are about the paper rather than about the subject. More than a couple per section and you are narrating your own structure. Cut the ones that only restate what the heading already said.
