The aim of this paper is to show how, starting only from forms belonging to various domains (morphology, syntax, and prosody), it is possible to compute grammatical relations and information structure constructions in Kabyle spon- taneous speech. The study is based on recordings made in the field, transcribed, translated, and annotated with Praat and Elan-CorpA. The methodology consists of systematically retrieving the sequences containing a verb and looking for the presence of a noun (and its inflection) within the prosodic group of the utterance, or outside, as well as studying the linear orders involved. This non-aprioristic methodology reveals the close interaction between grammatical relations and information structure in Kabyle. The study provides evidence to support the claim that the encoding of grammatical relations on nouns is a by-product of informa- tion structure constraints in Kabyle.