Guy Bertrand's « Capsules linguistiques » are broadcast daily on Radio-Canada. Each « Capsule » contains a French word or expression Bertrand considers erroneous or incorrectly used. Reviewing the literature, I found that although Bertrand is a public figure, studies of his work remain scarce. Furthermore, the transcriptions of the Capsules in published form were neither organized by theme nor by type of judgment, and contain a general bibliography rather than systematic citations. In this study, I sought to address these problems by reorganizing the prescriptive and descriptive information contained in Bertrand's work. The corpus consisted of 19 expressions employing the preposition « à », in order to analyze both lexical « errors » and syntactic « errors ». Next, I compared this information to findings in dictionaries (notably, Le Petit Robert 2014 and Le Dictionnaire québécois-français by Lionel Meney) as well as to other linguistic publications in order to also include the following: origin, etymology, semantic changes over time, other related expressions and any judgments on acceptability or register. To synthesize my findings, I established categories to answer the question of what types of errors were recurrent in our corpus. My synthesis regrouped these categories into four themes: judgments on structure; vocabulary; prescriptive or descriptive approaches; language registers. While I had predicted the presence of judgments on the influence of English and archaic terms, I had not expected the rejection of specialized terms extended to general use. Furthermore, Bertrand accepts regionalisms but generally relegates them to the familiar or popular registers. I also observed that Bertrand rejected changes to idiomatic expressions. These trends reflect a censorship of figures of speech in the standard register. For future study, it would be interesting to see which categories remain prevalent in an extended corpus of Capsules. As a point of comparison, Dire, ne pas dire… from the Académie Française website would provide insight from France. Finally, an acceptability or grammaticality judgment study with native speakers of French would provide insight as to whether the proposed alternatives reflect popular norms.