The aim of this thesis is to test two hypotheses. The first one is about the nature of comprehension impairment in Broca&#039 / s aphasia, namely the Trace Deletion Hypothesis (TDH), and the other is about the similarity between child and aphasic language, namely the Regression Hypothesis. According to TDH, agrammatic patients are impaired in comprehension of certain structures that are formed by movement transformation whereas they show normal comprehension in canonical structures. TDH proposes that patients use a default strategy, which assigns the first NP the agent role. As for the Regression hypothesis, children follow a hierarchy while acquiring a language, which is also followed by aphasic patients in the reversed order. That is, what is learnt last is lost first as a result of brain damage. In order to test these two hypotheses, we designed a psycholinguistic test in which the comprehension of Turkish relative clauses is tested via sentence-picture matching task. The same test is applied to children, agrammatic patients, and normal control subjects. We expected that Broca&#039 / s patients would be impaired in their comprehension of subject relative clauses in order for TDH to be confirmed and that children would also exhibit the same impairment as the agrammatic patients in order for the Regression Hypothesis to be supported. The results we obtained were not compatible with the TDH hypothesis in that the patients did not show comprehension deficit in subject relatives but they had problems in object relative clauses, which led us to conclude that TDH did not offer a cross-linguistic explanation for the nature of comprehension deficit in agrammatism and that even if the traces were deleted, the default strategy applied in each language was a parametric feature. We propose that Turkish patients assign the agent role to the first NP that is in the pre-verbal position, which is called the pre-verbal strategy. Also, we hypothesize that the difficulty the Turkish agrammatic patients present might be due to an impairment in another feature, which is agreement morphology rather than the deletion of traces. This study is compatible with the Regression Hypothesis in that both the patients and children exhibited similar impairment in the comprehension of object relative clauses.