The term ‘alignment’ is used to describe the patterns according to which the formal properties of subjects and objects interact with syntactic rules making reference to certain core arguments to the exclusion of others. This book gives a full survey of constructions involving variations in subject and object behaviour and marking and attempts to explain them in a typological and genetic perspective. Accusative alignment is known to be a stable feature in most ancient Indo-European languages. In Anatolian languages, however, the subject of an intransitive clause is treated differently from a transitive subject. A thorough examination of the data shows that subjects obey verbal selection: subjects of transitive verbs are restricted by rule to animate agents only. In transitive constructions inanimates agents are treated as oblique instrumentals. A second major restriction is that intransitive verbs are controlled by animate subjects only. As a result, if animate and inanimate subjects are identically encoded, they never share the same behavioural properties, while animate and inanimate subjects exhibit the same properties only when they are differently encoded. Such syntactical data are compatible with the organization of reconstructed Indo-European nominal infl ection and may explain its structure. (Text in French language)