This review discusses several kinds of linguistic generalizations that pose a challenge for the constructionist approach to linguistic generalizations advocated by Adele Goldberg. It is difficult to see, for example, how such an account can explain the wide-ranging linguistic phenomena governed by structural properties, such as c-command, or semantic properties, such as downward entailment. We also argue against Goldberg's rejection of formal semantics in favour of an account of meaning based primarily on information structure and discourse function.