The aim of this graduate thesis is to examine (mor)phonotactic features in early child's vocabulary from Beats and Binding phonology, Net Auditory Distance (NAD) principle and natural morphology perspective. All the words with consonant clusters in initial and medial position were extracted from the transcripts using the program for language analysis – CLAN (MacWhinney, 2000). Data have been collected from girl´s language corpus from the Croatian Corpus of Child Language (Kovačević, 2003) during 1;3 – 2;8 years of age. It was being observed whether the results would correspond to the findings from the literature. The data were analyzed longitudinally and in the total percentage in two phases of language development - premorphological and protomorphological. The preferability of clusters was determined using a NAD phonotactic calculator (Dziubalska-Kołaczyk, Pietrala and Aperliński, 2014). It was also determined whether the clusters are morphonotactic or phonotactic. The results are consistent with the literature: preferred medial clusters predominate in the girl's corpus and were first adopted, while the assumptions about morphotactics are partially confirmed - morphonotactic clusters first appeared in the corpus, but preferred clusters appeared in larger percentage. This paper also includes a diagram showing the distribution of clusters. The aim of the diagram is better understanding of the hierarchy of preferences and representation of cluster spacing with respect of sonority distances between segments.