Absolute or (VSO) word order. Those who assume

Absolute Subject ConditionSubjects are universally the easiest position on the (AH) to relativise. That is, if a language can relativise any (NP) position on the (AH) it can relativise subjects, but this is not true of any other position of the (AH). (Anderson 1977: 244)The obligatory gap in subject position in Semitic languages (SL) such as Hebrew (see Borer (1984)) and Arabic (see Plunkett (1990)) and Non-Semitic Languages (N-SL) such as Modern Irish (see McCloskey 1979) has been the focus of research within the frame of Government and Binding theory (hence, (GB)). The basic issue is whether the underlying sentence structure in relative clauses is (SVO) or (VSO) word order. Those who assume (SVO) word order have to establish the syntactic status of this gap: that is whether the subject gap is just another instance of the pro-drop phenomenon or a trace of movement. The most prominent attitude held in (GB) literature to account for subject gap, if movement is assumed, is that the (NP, IP (3)) subject position is governed and hence licensed by Inflection (hence, (I)). An opposite attitude claims that a trace must be head governed and that (I), which is often rendered as improper governor, is too low to govern the subject in the appropriate way. Rizzi (1990) claims that the possibilities of subject extraction vary across languages and he posits three major strategies for subject traces:i. The relative Complementiser (C) is identified as a governor in virtue of its agreement with the head noun by predication relation and hence licenses subject trace.ii. Subject trace is eliminated by resumptive pronoun insertion.iii. Subject is extracted from post-verbal position.Rizzi (ibid.) proposes that the last strategy (iii) applies to pro-drop languages such as Arabic and Italian. Subject extraction can be, from a position in which proper government requirement is met, namely, post-verbal position. Rizzi supports his hypothesis with evidence from null-subject languages and concludes that post-verbal subject is adjoined to (VP) and is properly governed by (I). This attitude is reminiscent of the traditional Arabic stipulation that only post-verbal subject (NPs), are classified as agents of verbs. I assume that a subject Determiner Phrase (DP) acquires subject agreement inflection, which is base-generated in (I), by (spect-head) subject-verb agreement. Unlike object (DP), these features are not spelled out as a resumptive pronoun, but as inflectional agreement affixation on the verb. Thus, null subject in Arabic is licensed by (I) and identified by d-structure subject agreement inflectional phi-features base-generated in (I). (See Fassi Fehri 1993)