The article discusses the design and application of dependency treebanks for Biblical Hebrew, focusing on their potential for linguistic research. It highlights the importance of such treebanks in studying historical languages, where native speaker intuition is unavailable. The study compares dependency grammar frameworks, such as Universal Dependencies and Prague Dependencies, examining their suitability for different research goals, including syntax-semantics interface, word order analysis, and phonological-syntactic relationships. Specific criticisms of Universal Dependencies, particularly its hybrid nature prioritising semantic over surface-syntactic relations, are addressed alongside alternatives like the multilayered Prague Dependencies. The article emphasises the need for research-driven design, recommending adaptations based on the intended linguistic applications and underlying theoretical assumptions.