Traditionally Linguistics has concentrated mainly on more basic language structures (e.g. words or sentences), while discourse has been unfairly ignored until 1970s, when the interest for discourse micro- and macrostructure arose. A so called multi-level approach, which implies the interdependence of conceptual structures, content of thought, and processing linguistic units, gains more popularity in recent discourse research, but it is still rarely used in language development studies. This approach is the main framework of this project proposal. In the field of psycholinguistics and applied linguistics, discourse analyses have been considered a valuable tool for assessing language knowledge (Schneider et al., 2006; Botting, 2002). The elicitation of discourse samples is an ecologically valid assessment tool since it reflects conversation in everyday natural environment and provides a clear insight into language knowledge. This is especially important for assessing language abilities of bilingual speakers and speakers with language impairment. Despite the fact that narrative discourse has been proven a valid diagnostic tool, research of its structure and changes during language development are relatively rare in Croatian language, with some exceptions (i.e., Arapović et al., 2010; Medved Krajnović, 2001 and Kuvač, 2004). Most of the theories, approaches and discourse analyses measures are based on English language so there is a need for their validation in typologically different languages. Various approaches to language acquisition have changed based on new insights from different languages. In this project new resources will be developed and some existing updated (i.e., lexical database and different corpora). Their analyses will lead to new findings, such as frequency and types of language structures in Croatian in a particular period during language development and in different populations, as well as to development of new automatized corpus measures. These findings will be crucial for development of the new discourse assessment material. Material will be based on picture stimuli in a computer-based application. Application will allow for the automatization of the recording procedure, transcription support and automatic calculation of discourse measures. As the material is grounded in research findings and validated, it will be ready for use as language assessment tool. Validity studies will provide reliable data on discourse development for different groups of speakers (monolingual, bilingual, atypical language development, different SES).