ENThe aim of this paper is to analyse the semantic values of the Lithuanian perfect construction, putting them into a perspective of grammaticalization. The paper is based entirely on the data from a 2-million-word Facebook comments corpus created ad hoc for this study. The quantitative and qualitative analysis of the semantic values of the perfect tokens extracted from the corpus reveals several previously unidentified features of this Lithuanian construction. A large proportion of structures formally corresponding to the perfect should be described as copular constructions with adjectivized participles. This formal coincidence and the ambiguity generated by it in certain cases should not be seen as accidental but rather considered a likely source of the grammaticalization of the Lithuanian perfect, as the influence of its semantic features can be seen in all the perfect’s other values. Considering it as a source, it seems that the development of the Lithuanian perfect is going in two separate, but also related directions, each of which is based on a gradual abandonment of one of the two core features of the prototypical Lithuanian perfects―the subject-oriented resultatives. In the case of the transitive resultative perfects, the orientation towards the subject is lost, while in the case of the experientials, it is the resultative meaning that is lost. Of these two values, the experientials are twice as frequent, which shows that the resultative meaning is abandoned more readily than the need to express a state or a quality of the subject. However, the experiential perfects seem to present some formal differences from all the other perfect values, namely, a significantly more frequent auxiliary usage which has so far been considered accidental. Keywords: perfect, Lithuanian, Facebook, comments, resultative, experiential, subjectoriented, participles, adjectival, grammaticalization. [From the publication]