Abstract
This paper presents some results of an investigation into the syntax of the NP/DP in Old English (OE). Specifically, it presents some new data on the syntax of possessive pronouns and determiners and offers an LFG analysis to account for their distribution and interaction. This is an area where we still lack the empirical base necessary for evaluating hypotheses about such important questions as the relationship between the reduction of inflection in Early Middle English and changes which took place in the syntax of the English NP/DP.
The paper focuses on the status of possessive pronouns in OE. It is traditionally assumed that possessive pronouns, some of which inflected to agree with the head noun, were adjectives at the OE stage; e.g. Campbell (1959). This assumption has been carried over into recent formal treatments; for example, Demske (2001), working within an HPSG framework, argues that possessive pronouns were adjectives at the OE stage but became reanalyzed as part of the determiner system in later English. The most convincing syntactic argument which Demske makes for this analysis is that in OE, a possessive pronoun could follow a determiner, just like an adjective, as in (1):
| (1) | be | þære | his | eadgan | gemynde |
| about | that (FDS) | his | holy | memory (FDS) | |
| 'About the holy memory of him' Blickling Homilies 197.3 | |||||
I refer to the construction in (1) as the 'Det Poss' construction. This characteristics of this characteristics and occurrence of the Det Poss construction in OE and Middle English texts are discussed in Allen (2004).
Any attempt to treat OE possessives as ordinary adjectives must have difficulty explaining a second construction in which a possessive and a determiner co-occur, but the possessive precedes the determiner:
| (2) & | heo | is | þin | seo | clæneste | fæmne | |
| and | she | is | thy | that (FNS) | cleanest | virgin | |
| 'and she is your purest virgin' Blick. 157 2 | |||||||
The construction in (2), which I refer to as the Poss Det construction, is discussed by Mitchell (1985: para. 104-106) but has been ignored by most linguists working in the history of English. When it is mentioned at all, the existence of Poss Det as well as Det Poss is usually attributed to the freer word order of OE and the two orders are essentially treated as variants of the same construction in which the possessive and the demonstrative co-occur (for example Fischer 1992: 213).
My own investigation, however, has led me to the conclusion that we are dealing with two quite distinct constructions here; in Poss Det (by far the more common construction) there is always an adjective, as in (2). In this construction, the possessive pronoun appears in a position in which adjectives are not normally found, although this possessive has the same adjective-like inflection as in the more widely discussed Det Poss construction. Further more, the determiner used in this construction is invariably a form of se 'that, the', while both se and þes 'this' are found in the Det Poss construction. A construction in which a possessive precedes a determiner seems to by typologically unusual, at least in European languages (Manzelli 1990), and the restriction of this construction to nouns which are modified by adjectives seems particularly odd. The OE situation is best explained as a result of the history of the 'weak' adjectives in Germanic; These adjectives were originally nominalizations (Prokosch (1960), Curme (1910) (1910) etc.), and we can explain the presence of determiners with these adjectives if the adjectives still had some nominal characteristics in OE.
I will offer an analysis of the Det Poss and Poss Det constructions which accounts both for the syntax of these constructions and the morphology of the possessive pronouns. Formally, we can account for the association of the Poss Det construction with adjectives by assuming that the AP in OE had a position for a determiner (cf. Giusti (1997) , who argues that modifiers to nouns may have their own extended projections ). This effect can be achieved by assuming that AP could be the complement of a D head. This D was limited to the article-like determiner se .
The most important conclusions of this study are: 1. OE was already a highly endocentric language. The Poss Det and Det Poss constructions are not an example of free word order, but are hierarchically organized. 2. Despite their morphology, possessive pronouns did not have the syntax of adjectives in OE, but already had the determiner-like characteristics associated with possessives (both pronominal and nominal) of Present Day English.
This improved understanding of the structure of the OE construction can serve as a starting point for research into an explanation for why the Poss Det and Det Poss constructions later disappeared.