Australian Linguistic Society
Loss of the postnominal genitive in English
Cynthia L. Allen, Dept. of Linguistics, Arts Faculty, Australian National University, Canberra, ACT 0200, Australia Cindy.Allen@anu.edu.au
One of the many changes which took place in English as case-marking distinctions were lost and constituent order because less flexible was that the position of adnominal genitives became fixed. In Old English (OE) a genitive modifier of a noun could come either before or after the head of its phrase:
(1) Gabriel wæs thissa brydthinga ærendwreca Gabriel was these(G) nuptials(G) messenger 'Gabriel was the messenger of these nuptials'  BlHom I, 13 (2) Seo geofu wæs broht for the grace was brought for thære synne thæs ærestan wifes the(D) sin(D) the(G) first(G) woman(G) 'The grace was brought for the sin of the first woman' BlHom I, 21
By Early Middle English (EME), however, the postnominal genitive had disappeared or was moribund; examples disappear in the 13thC. The postnominal genitive was to a large extent replaced by the of-genitive, although matters are more complicated than a simple replacement of one construction by another. I will not be discussing the of-genitive here but will instead focus on how the postnominal genitive became impossible. There is general agreement that the loss of the postnominal genitive was linked to the reduction of case marking in English in some way. However, not very much attention has been paid to exactly how this disappearance took place. We might hypothesise, for example, that the postnominal genitive became impossible because the loss of case-marking distinctions somehow made it impossible for language-learners to continue constructing grammars in which nouns could assign case either to the right or to the left, as it was in OE, and a new generation of language-learners constructed grammars in which nouns could still assign genitive case, but only to the left.
In this paper, I will present some details concerning the loss of the postnominal genitive which add to our knowledge of how the loss of case marking interacts with syntactic change. I will argue that the stages which are discernible in the loss of the postnominal genitive cannot be accounted for by assuming that the loss of case-marking distinctions took place in one stage, e.g. a change in parameter-settings which made it impossible for nouns to assign case to the right. The loss of the postnominal genitive was indeed ultimately triggered by the loss of agreement within the noun phrase, but the facts suggest that the postnominal genitive was lost in two stages, as one type of postnominal genitive disappears before the other does.
We can first note that evidence from different dialects supports the notion that there is a link between loss of case-marking forms, in particular loss of agreement within the NP, and the loss of the postnominal genitive. There is certainly a correlation between the retention of the postnominal genitive and the retention of agreement within the NP; in those dialects in which agreement is retained the longest, the posthead genitive generally remains the longest. This fact was noticed as early as (1931) by Russell Thomas in a pioneering dissertation on the development of the adnominal periphrastic genitive in English.
But Thomas was concentrating on the rise of the periphrastic or prepositional genitive, and so it is no surprise that he did not look terribly closely at the nature of the postnominal genitives which remain in the EME period in different dialects. In fact, although Thomas argued for a correlation of the sort I have just mentioned, he was forced to the conclusion that the direct correlation was not evident for every pair of texts. He observes (p. 121) that it is not the case that every text which has more reduced inflection than another also shows fewer postnominal genitives, and suggests that other factors such as differences between scribes may be at work.
However, my own investigations into the loss of the postnominal genitive indicate that the apparent counterexamples to the correlation between the loss of agreement and the loss of the postnominal genitive disappear when we pay sufficient attention to the nature of the posthead genitives which remained in the EME period.
First, in all dialects we find that all the posthead genitives of the EME period begin with a word which is clearly inflected for genitive case. This is a fact which was noted by Thomas, who observes (p. 107) that examples like the following are not found:
(3) *the sune the mannes 'the son of the man' (4) *the word halige (ge)write 'the words of holy scripture'
My own findings confirm Thomas'. As Thomas notes, we can explain the general correlation between the loss of agreement and the loss of the postnominal genitive when we realise that a prenominal genitive usually had clear case marking before the head noun, since what preceded the head noun would be the possessor noun, which normally was inflected for genitive case. On the other hand, the posthead genitive would normally begin with a determiner or an adjective, which were no longer clearly inflected for genitive case. This was particularly true because in OE, as in Modern English, there was an extremely strong tendency for proper nouns and pronouns, which constituted the largest number of genitives, to be placed in the prenominal position.
One thing which Thomas does not explicitly mention is that the posthead partitive genitive is still in use in the EME period, even in dialects which generally no longer use other types of posthead genitives:
(5) & fela othre godre cnihte and many other(GP) good(GP) knights(G) 'And many other good knights' PC 1124
These partitive genitives always conform to the generalisation that in EME, regardless of the dialect, the posthead genitive was only used when it was introduced by an element with genitive case marking. Of course, the partitive genitive often consisted of a single element in the genitive case:
(6) hwuch ure is kempe which our is victor 'which of us is the victor' St.Kat. 31.201
This restriction of the postnominal genitive to ones which began with genitive case has an interesting consequence. Let's consider how to account for the inability of genitive NPs to be postnominal from the ME period on. One approach we might want to take is to say that the reduction of agreement somehow caused language-learners to construct grammars in which nouns could assign case to the left, but not to the right. Presumably, the loss of the ability of the noun to assign case to the right would result from a greatly reduced frequency of speakers using the option of the postnominal genitive, for reasons which we have already discussed. A new generation of language learners, hearing very few examples of postnominal genitives, would assume that their language did not allow them.
It is certainly reasonable to assume that a greatly reduced frequency in the use of postnominal genitives lead to the ultimate ungrammaticality of the construction in English. We might attempt to account for the changes in the following way: we can assume that as agreement was lost, speakers tended to use more prenominal genitives or to replace the postnominal genitive with an of genitive. Once the frequency of agreeing forms dipped below a certain level, the incidence of postnominal genitives would have been so low that language-learners assumed that case assignment to the right by a noun was impossible and no longer constructed grammars which did this. This sort of scenario predicts that postnominal genitives would have disappeared very shortly in a given dialect once agreement within the NP had been lost in that dialect, since speakers would have been unable to produce the old posthead genitive.
However, the facts do not support this scenario because it is simply not the case that all posthead genitives disappeared shortly after agreement was lost in a given dialect. It is true that most kinds of posthead genitives disappear earliest in the dialects which lost agreement the earliest. But in fact one type of posthead genitive is found in all English texts into the 13thC, regardless of the state of case marking and agreement. This is the partitive genitive. Examples of the partitive genitive are still found in texts like the Katherine Group of the first quarter of the 13thC (see ex. (6) above) although other posthead genitives are no longer to be found in this West Midlands dialect in which agreement had pretty much been lost by this period. Similar examples are not uncommon in the north midlands Ormulum of c.1200. This text shows even more radically reduced case marking than that of the Katherine Group and only occasional agreement:
(7) Thatt nowwther theyyre nohht ne layy that neither their not not lay I nane depe sinness in no deep sins 'That neither of them lay in any deep sins' Orm 12872
In neither of these texts is any other posthead genitive to be found, except for a few fixed expressions which are to be found in the Katherine Group. The facts indicate that at this period, we don't have a grammar which simply does not allow case to be assigned to the right by nouns, but one which allows nouns of a particular type (quantifiers, if we want to call them that) to assign such case. Also, in these texts the lack of any nonpartitive genitives extends to genitives which are themselves introduced by genitives, which are still found in some other texts, as in (10) below. This suggests that it was not simply a matter of needing a genitive-marked constituent as the first constituent of a postnominal genitive.
It therefore will not do simply to say that the postnominal genitive was lost as a grammatical possibility once agreement had been lost. However, if we make a distinction between lexical and structural case assignment, the facts are more easily described. I suggest that in Old English case was both structurally assigned and lexically assigned. That genitive case was sometimes lexically assigned is apparent from the fact that certain adjectives required genitive case on their complements:
(8) For▀on we sceolan beon for-that we shall be gemyndige Godes beboda mindful God's commands-gen.pl 'because we should be mindful of God's commands' HomS8 202
But within the NP, we can say that genitive case was generally structurally assigned: genitive was the case automatically assigned to an NP that was inside another NP. In the case of the partitive genitive, however, we can say that certain heads (whether we want to call them nouns or quantifiers) required genitive complements. These complements could either precede or follow the head.
By the early 12thC, the use of post-head genitives was clearly on the decline in all regions, although the decline is most severe in the areas in which inflection of modifiers had decayed the most. We unfortunately only have a few texts which were composed  in the 12thC, but the couple we have from a southern area where case marking and agreement were well preserved show that the postnominal genitive, while not terribly common, was certainly still a grammatical option for all types of genitives, including the true possessive and the 'objective' genitive:
(9) Hwa is wyrhte ▀ære synne Who is author the(G) sin(G) 'Who is the author of sin?' Honorius 28 (10) for ▀ære sicornysse godes rices for the(D) certainty(D) god's kingdom(G) 'for the certainty of God's kingdom' Honorius 143.17
In contrast, in the Peterborough Chronicle continuations of the mid 12thC, we find that the few examples of the postnominal genitive are all partitives.
The facts suggest that the post-head genitives to which case had been lexically assigned were the ones which were affected last. It seems that within the NP, structural marking of genitive case in the post-head position was lost before lexical case marking was. In fact, the old post-head structural case marking was essentially replaced by another type of structural case marking-the use of a prepositional phrase, which was encouraged by contact with French. Of course, the structural assignment of genitive case still remained but was confined to the pre-head position. Assignment of genitive case by specific lexical items such as quantifiers remained longer. It seems to have disappeared at the same time in the pre-head and post-head positions.It furthermore appears that the lexically-assigned partitive genitive happened about the same time in all dialects. We find that by 1200, even in the southerly Vices and Virtues, in which the lexical assignment of case to the objects of verbs was still very much alive, the partitive genitive, either prenominal or postnominal, is very rare.
These findings concerning the of genitive case increase our knowledge of how case marking changed through the history of English. My research has suggested the following sequences in different dialects:
Expanding on this table, I note first that my earlier investigations into the decline of case marking provided evidence that lexical case marking of objects was lost before the lexical case marking of subjects. I assume that the preverbal experiencers of the 'impersonal' verbs such as me thinks were subjects which got lexical case marking. My investigation into the loss of the postnominal genitive indicates that we must distinguish between lexical and structural case assignment within the NP also.
The second point to be made is that the loss of the structurally marked postnominal genitive seems to have begun fairly early in OE before agreement had begun to decay seriously, as there is a steady reduction in the frequency of the postnominal genitive already in OE, as Thomas demonstrated. I think that the best way to view the situation is to view the use of the different genitive constructions as having been regulated in OE by factors such as topicality, heaviness of the genitive, etc. As OE wore on, the factors which would bring on the postnominal genitive became more and more restricted, especially as the new prepositional genitive became an alternative to the postnominal genitive. The loss of agreement signalled the end of the postnominal genitive because speakers now did not use it even when discourse factors would have allowed it unless the genitive began with a clearly genitive-marked element. This, combined with the growing use of the prepositional genitive, severely reduced the frequency of the structurally-marked postnominal genitive to the point where language learners constructed grammars in which case was not assigned structurally by nouns to the right.
It is interesting and perhaps surprising to find that the dialect difference in the used of the postnominal genitive which is found in the mid-12thC seems to have disappeared by the early 13thC; we find that structurally marked postnominal genitive has disappeared from the Vices and Virtues of c.1200 even though there is still abundant agreement within the NP in this text. It appears that while loss of agreement led more rapidly to the loss of the structurally-marked postnominal construction, the construction in all dialects was unable to withstand the onslaught of prepositional genitives. It appears that in the late 12thC, the prepositional genitive became so popular that it essentially replaced the postnominal genitive even in those dialects where the postnominal genitive would still have been viable because of the retention of inflection.
It appears that the loss of lexical case marking within the NP was not directly related to the loss of lexical case marking in other parts of the sentence, and also that the loss of agreement within the NP did not play a role, at least not a direct one, in the loss of the partitive genitive. The loss of agreement does not explain the loss of the partitive genitive in prenominal position. The loss of this genitive appears to have begun also in the OE period, as an alternative using the preposition of and the dative case was already used at that time. Why the partitive genitive should have hung on for so long in the face of competition by this other construction is a matter that requires further investigation.
For the sake of typographic convenience, thorns and eths have both been represented with "th" and yoghs have been replaced with "y". (Return to footnote in text.)
There are quite a few texts from this period which are copies of earlier texts, but they are unreliable witnesses to the use of the genitive in this period. (Return to footnote in text.)