The Position of the Adjective in Old English Introduction

3. Adjective position in Old English

As I mentioned in the introduction, not much detailed work has been done on the position of the adjective in Old English. Mitchell is a fairly reliable guide on what has been done up to 1985. His own remarks on the topic are vague. In §172, he notes that the reason for postposition may be Latin influence, a desire for emphasis, rhythmic and stylistic variation, metre etc. In §160, he writes that it is not always clear whether with an “attributive adjective in post-position … we have to do with an attributive, predicative or appositional, use.” It seems that Mitchell himself believes that these adjectives are still attributive (witness his remark in §168 that adjectives after the noun “may seem predicative to some readers”, emphasis added). Mitchell as usual describes all the various possibilities but does not go into any semantic and/or syntactic differences that may account for these possibilities. He is a taxonomist (a very good one!) and not really interested in the “why” of variation. So clearly, for a linguist interested in an explanation for the variation, there is indeed room for more work.

What other possible explanations have been offered? In §171, Mitchell refers to Sørensen (1956), who writes: “Anyhow it is tempting to assume that the widespread use of the construction adjective + substantive + adjective in OE. was supported by, if not a direct outcome of, [a] general tendency towards looseness in construction” (1956: 262-263). This looseness in construction -- which is also visible in other “split constructions” (cf. Reszkiewicz 1966 for a list of them), e.g. split subjects, objects, genitive phrases etc.-- is according to Sørensen probably due to a certain immaturity in the writing style, which is manifested also in the more frequent use of anacolutha in Old English. This leads him to the conclusion that postnominal adjectives are possibly cases of “afterthought” (p. 262). Although I think it is too easy to explain these cases simply as immaturity and/or afterthought, there is a remark in Sørensen’s study that is certainly worth considering, but which he does not follow up himself. He refers on page 262 to Bøgholm’s (1939) English Speech from an Historical Point of View, who has pointed out that in Old English “the rule is for parallel words to be kept apart”. This links up with a remark made by Spamer (1979) that in Old English adjectives were not recursive. I will come back to this below. Some further interesting and very pertinent observations made by Sørensen with reference to modern postnominal adjectives preceded by and are:

• 
  That this “adjective may be interpreted as a predicative in its own right, as it were not very closely connected with the preceding substantive” (p. 264)
  
• 
  That the postnominal and + adjective construction “may be caused by the need for linking up the second adjective with what follows” (p. 264), where he offers this example from Somerset Maugham, It was a pleasant sight and grateful to the sensibility …
  

• 
  That the postnominal and + adjective construction may be used to express contrast where the qualities expressed by the first and second adjective do not refer to “one primary” (p. 264) . I.e. in black spirits and grey, grey may refer to another category of spirits.
  

I will show later that all these factors may well be at work in Old English, and that they are all linked.
Spamer (1979) proposes that there are two kinds of adjectives in Old English, which correspond with the morphological weak/strong distinction. The weak adjectives according to him are “adjuncts” (Quirk et al. 1973 term these denominal adjectives, e.g. of the type a stone wall, a Shakespearian critic), while the strong ones function as determiners and are on a par with OE se, seo, þæt. In his study Spamer uses this distinction in order to explain the development of the article system.⁷ It takes too long to go into all his arguments here, but he clearly suggests that there may be a link between type of adjective and (in)definiteness as suggested also by Stavrou for Modern Greek. He notes that Old English adjunctival (i.e. weak) adjectives are recursive, while strong adjectives, being determiners are not. He then notes that this explains why one may have two weak adjectives following one another prenominally, why one may have a strong followed by a weak adjective prenominally, but not two strong adjectives in a row (p. 245). When two strong adjectives modify the same noun, the word order in Old English is “significantly different” from that of Modern English (Spamer 1979: 244). As an example of this different order he quotes the following, from Mitchell (1968: 65),

(3) Þæt hi næfre ær swa clæne gold, ne swa read ne gesawon

that he never before such pure gold, nor such red not saw

whereby he reads read as a modifier of an understood (deleted because co-referential) noun gold. Further, he remarks that “[i]t is a striking feature of Old English syntax that a series of adjectives [strong ones, presumably, OF] almost invariably contains and between every single adjective and the one following”; this is necessary according to Spamer because the “adjectives which they conjoin are non-recursive” (p. 244). So in fact, he suggests that the use of and and the use of postnominal position are as it were “forced” by the non-recursiveness of strong adjectives in Old English. This hypothesis explains quite nicely why indeed quantifiers (which are almost always strong) do not normally co-occur with the demonstrative pronoun (determiner) in OE. In my prose data I did not find a single example.⁸ It also explains why, when there are two strong adjectives and no determiner, front placement of both is rare (I found only 8 examples in my data, see Table 2, row 6), the usual stratagem being to place and before the second adjective, or to place (and +) adjective or both adjectives connected by and after the noun. There are, however, quite a few problems with the data as presented by Spamer, and more specifically with the data ignored by Spamer.

[place Table 2 about here]

First of all the type consisting of two quantifiers followed by a noun, where both quantifiers are typically strong, should not be possible (because according to Spamer 1979: 243 the strong adjective is non-recursive). However, this type is extremely frequent in my data (see Table 2, row 3: 77 examples). In addition, I have found no examples of a first quantifier weak and a second strong. Some examples:

(4)a Ænig oðer sceat (nom. sg. masc., both strong ) (Æl. Hom. Supplem. II 1.39)

any other tribute
(b) Sumere oðre stowe (dat. sg. fem., both strong ) (Æl. Hom. Supplem II 1.197)
in some other place
(c) Sume feawe dagas (acc. pl., both strong) (DOE, LS9(Giles)822)
some few days

Secondly when we have a strong quantifier (which therefore must function as a determiner) followed by an adjective, we would expect the adjective to be weak according to Spamer’s theory. However, both quantifiers and adjectives are usually strong in my data (Mitchell 1985: §125 confirms this):

(5)a Ælc halig fæder (nom. sg. masc., both strong) (Ælet4 1.203)

each holy patriarch
(b) On sumne blindne seað (acc. sg. masc., both strong ) (Ælet4 1.167)
‘Into some blind pit’

This is also true when the first adjective is not a quantifier but in fact a true adjective, in that case both adjectives are also strong,

(6)a mid ofermæte unclæne luste (instr. sg. masc. strong adjs) (Mart1. 1.150)

with excessive unclean lust
(b) of surre rigenre grut (dat. sg. fem. strong adjs) (Læceboc 3. 1.59.1.1)
from sour rye(-made) groats

Note that in the last example the typical adjunctival (adnominal) adjective rigen is in fact declined like a strong adjective, whereas Spamer (1979:245) asserts that only the weak adjectives are adjuncts.

Spamer’s theory also does not explain why there are cases of postnominal adjectives when the “need” does not arise (cf. the data in Table 3, below), as in

(7) Þone ilcan ceaddan iungne (acc. sg. masc. str.) (Chad.1.184)

the same Chad young

Here iung could easily have preceded the noun syntactically speaking (semantically it is a different matter altogether), because it would have been weak after the determiner (like the other adjective, ilcan), and two weak adjectives are allowed in his theory. Also the hypothesis does not explain why and is not always inserted (see [8] and also Table 3, below) before the first postnominal strong adjective, as in the example he quotes from Mitchell (see (3) above).

8.