Close and Extended Relative Clauses—A Critical Account
Analytical approaches to relative clauses have by and large incorporated the growing body of evidence regarding biological constraints on embedding. Labeling higher-ranked relatives as mothers, for example, sits well with our understanding that mother-child is the closest relative bond there is. Laboratory research on mice confirms that naturally embedded offspring are regularly found within their mothers, not their fathers.1
Labels attached to offspring clauses follow suit on female-bound terminology, less for the sake of consistency than for the sake of recursion. Motherhood is a heritable trait, although only daughters may in turn generate children, once they rank-shift to matrix (not patrix) status.2 The terminology is accurate in the additional sense that only females can ever know how many relative children they have: uncontrolled offspring proliferation is as damaging to universal syntactic explanatory power as to global warming.
Mother-daughter relationships hold for restrictive as well as non-restrictive relatives, given that different degrees of active bounding arise from different locality conditions.3 Restrictive bounding of one’s relatives is the domain of INFLamed delivery, the surface representation of deep INFLexible constituency: adjacent constituents are consequently uttered on a single prosodic phrase that condones no interruption. In contrast, the easy-going nature of non-restrictive relatives is captured within the framework of control theory, which licenses PRO-drop by allowing potential prosodic breaks to leisurely drop in here and there between constituents.
Close relative behavior of mice and women is thus well accounted for. Among extended relatives, however, the picture that emerges is mixed. C-command, for instance, governing directives issued by cousins in laboratory cages as well as in syntactic trees, has naturally found limited applicability amid accounts of relative well-formedness, the reason being that cousins represent minor kindred bonds, often artificial: the label cousin may designate individuals who share no deep or surface biological bonds whatsoever.
The label Comp, in turn, leaves much to be desired on several counts. To its merit, it does represent a rare example of multilingual linguistic nomenclature, in that compadre is not a native English word (bahuvrihi is the only additional example that comes to mind), but Comp’s drawbacks far outnumber this commendable terminological diversification.
First, the label is clearly male-bound, an observation that, on closer look, turns out to be irrelevant: the point is that compadres, as well as comadres, cherished though they may be, remain spurious relatives whose nodes project from outside core trees. Second, mice were found to not bond with their allotted compadres at all, despite enticing food and money rewards lavished on them to do so. Third, Comp dominance of relative constructions is said to be unavoidable on the evidence of syntactic manipulation. Comp remains Comp, the argument goes, in extended relative constructions such as My compadre, who is also your compadre, is a syntactic intruder. versus Is my compadre, who is also your compadre, a syntactic intruder? The argument falls flat on the counter-evidence that If I’m your sister, my mother is your mother. versus The bond my sister and the compadre our mother assigned her engaged in need not extend to me for that.4 Fourth, ERT (Extended Relative Theory) posits that relatives that undergo Move-Out, through relinquishment of the family fold or expulsion from it, should be recoverable from surface organization of genealogical structure. Where such barred relatives appear in a tree, however, compadres emerge as a mere backdoor stratagem to reinstate them, in that barred nodes are consistently represented as expanding into Comp sisters of themselves.
Sisterhood is in fact the last, and the most vexing issue that plagues the assumedly innate underpinnings of relative clause architecture. Etymology reveals that the label Comp refers to co-parental, and hence sister relationships, whereby widespread representations of Comp nodes as governing extended relative clauses that contain mothers and their daughters is a travesty of biological facts. As the adage pithily has it, “No compadre ranks higher than a madre.”5
1 Generalization of this finding across mammals, including the sapiens kind, is as scientifically legitimate as generalizations about, say, mammalian sugar metabolism, from the observed results of feeding lab mice pure glucose from birth.
2 As lab mice behavior likewise confirms.
3 Relative bounding only affects close relative clauses and should therefore be carefully distinguished from relative binding, which concerns itself with antecedents, that is, overall ancestry.
4 On the thorny issue of compadre center embedding, which is not found among natural mice, I refer to M.Adam’s discussion, and references therein.
5 See note 2.