What does it mean when a supervisor says a literature review is too descriptive?

The feedback is accurate almost every time it is given, and almost never actionable as phrased. What the supervisor is reacting to is a chapter where they cannot find your voice. They can see that you read the literature. They cannot see what you think about it, and the chapter therefore gives them no reason to accept your research question.

There is a quick diagnostic. Go through the chapter and highlight every sentence that could not have been written by someone who had merely skimmed the abstracts. Those are your analytical sentences. If the highlighting is sparse and clustered at the end of sections, you have a description problem, and you now know exactly where the gaps are.

The other symptom to look for is the missing "so what". Descriptive paragraphs end when the description ends. Analytical paragraphs end with a consequence: what this means for the question, what it leaves unresolved, why it justifies the next section. Adding a final sentence to each paragraph fixes more descriptive chapters than any other single change.

Published guidance names the same failure. Pautasso calls the article-by-article chapter "stamp collecting" and says a review should discuss the literature critically, identify methodological problems, and point to research gaps, so that the reader finishes with a sense of the field's major achievements, areas of debate, and unanswered questions (Pautasso, 2013, p. 3). Lingard states the requirement spatially: the literature you cite should surround and emphasize a meaningful white space, and if you describe existing knowledge so fully that nothing appears left to investigate, you have not established why your study is necessary (Lingard, 2018, Part I, p. 47). The follow-through Lingard asks for is to map the gap and then show that your methodology suits it (Lingard, 2018, Part I, p. 49).

One warning. Do not respond to this feedback by adding evaluative adverbs. A chapter where every study is "notably", "importantly", or "crucially" something reads as unearned confidence, and the underlying problem is untouched.

How do I add critique without inserting personal opinion?

Anchor every judgment to something a reader can check in the paper: sample size, design, measure, follow-up length, replication, who funded it. "The effect rests on one cross-sectional sample of forty undergraduates" is critique. "This study is weak" is an opinion nobody has to accept.

This is also what keeps critique civil. You are not attacking the researchers, you are naming the limits of the evidence, which the authors themselves almost certainly acknowledged in their own limitations section. Reading that section is often the fastest route to a fair, specific criticism.

Critique also has an upper bound, and the two sources set it from opposite sides. Pautasso warns that a review should be neither a promotional brochure nor an exercise in excessive self-denial (Pautasso, 2013, p. 3), while Lingard rejects the idea that being objective means presenting facts without taking a stance, and asks instead that you judge the quality, source, and reliability of each knowledge claim (Lingard, 2018, Part II, p. 134). The working example is at the level of the verb. If an author offered tentative reflections, write that the author "suggests" or "explored", not that the author "argues" (Lingard, 2018, Part II, p. 134). Getting that verb right is critique, because it reports how much certainty the evidence actually carries.

The same anchoring works for praise. "The only study to follow participants past five years" tells the reader why a work matters. "A seminal contribution" does not.

Which sentence patterns signal analysis to a reader?

Sentences that compare, weigh, or qualify. Phrases such as "the stronger evidence comes from", "this holds only where", "both accounts rest on the same measure", and "what has not been tested is" all announce that you are judging rather than reporting.

The most useful of those is the one about shared assumptions. Noticing that two apparently competing studies rely on the same instrument, or the same population, or the same unexamined definition, is the kind of observation that only comes from reading across a field rather than through it. It is also frequently where a gap is hiding.

What you leave out is also a judgment. Lingard says that in a well-studied area such as professionalism in medical education you have to make a strong and explicit case that a gap remains, because readers may feel the topic has been studied enough, so you select the knowledge claims that map that gap rather than listing every relevant publication (Lingard, 2018, Part I, p. 48). The volume behind that advice is Pautasso's: compared with 1991, the number of Web of Science papers indexed in 2008 was three times higher for malaria, eight times higher for obesity, and forty times higher for biodiversity (Pautasso, 2013, p. 1). Where several reviews of your topic already exist, Pautasso recommends examining their approaches, limitations, and conclusions, finding an angle they did not cover adequately, and bringing in newer material (Pautasso, 2013, p. 1).

Aim for one such sentence in every paragraph. Not more, since a chapter that argues in every line is exhausting, and not fewer, since a paragraph with none is a paragraph a reader could have got from the abstracts.