The expression and articulation of scientific assessment are key instruments in professional academic communication, and the text of peer-review exhibits a distinctly institutionalized nature, which making it a highly relevant subject for studies of expert discursive practices. The focus of this study is the relationship between the tonality of the vocabulary of the peer-review text and the general evaluative background, expressed in the reviewer’s conclusion. Based on the research corpus of peer-reviews and the Russian sentiment lexicon (RuSentiLex), the article analyzes the reviewer’s use of words expressing a qualification assessment (adjectives, adverbs) and their distribution across subcorpora of positive and negative reviews. The analysis revealed a list of evaluative lexemes evenly distributed across both subcorpora, which represents the basic evaluative vocabulary of a peer-review, determined by the journal editorial standards (a review template). The uniformity of the distribution of the basic vocabulary is maintained by changing the tonality to the opposite one using grammatical and lexical means of negation and gradation of the evaluation. Thus, a balance is established between the used sentiment lexicon and the reviewer’s conclusion. Additional sentiment lexicon indicates that the repertoire of positive assessments is very limited, while the more represented and diverse means of explicit negative assessment indicate the categorical directness – a hallmark of Russian peer-reviews. The studied material shows a positive correlation between the choice of sentiment lexicon and the reviewers’ final verdicts, in negative contexts, the semantization and contextualization of neutrally colored vocabulary induces a corresponding assessment, which strengthens the reviewer’s final verdict.