The Position of the Adjective in Old English Introduction

Gyldenne wingeard trumlicne and fæstlicne (accus. sg. masc. strong adjs)

(Alex. 1.107)
golden vinyard durable and firm

Nor is it clear why we should get postnominal adjectives when there are no prenominal ones at all. After all, in that case, there is no reason to opt for a

[place table 3 about here]
“significantly different” word order. Such instances are, however, highly frequent in my data (see Table 3, row 1), especially when the postnominal adjective is a quantifier, or an adjective in -weard:⁹

(9)a þæt hi hyra æhta ealle beceapedon (Ælet1 1.40)

that they their possessions all sold

2. 
   þa wilnode ic indeum innanwearde to geseonne (Alex. l.129)
   

then wanted I India inner to see
(c) Petrus hæfde wif ful sođ (Æletw. l.17)
Peter had (a) wife full true
(d) Gersuman unateallendlice (ChronE 2, 1086.59)
treasures uncountable

A question that arises in this connection is, why are postnominal adjectives practically always strong (there are only two exceptions in the data in Table 3)? This is so even when a determiner precedes the noun, as Mitchell (§126) indicates,

(10) in þissum life ondwardum (Mitchell §126, Ch 1508)

in this life present

The postnominal adjective is only weak when the determiner is repeated too. Most cases of this recorded in Mitchell §126 are superlatives, which tend to be weak adjectives even when predicative (for an explanation of this see Brunner 1962: 54, quoted above).

A last point that Spamer’s hypothesis does not satisfactorily account for is the frequent occurrence of two prenominal weak adjectives connected by and, where there should be no need for and (because the weak adjective is recursive). At a quick glance through the DOE I found many of these. Set against the few examples of weak adjectives in a series (all discussed below), this is, to say the least, remarkable,

(11)a Se æresða stream þære clænan and hlutran burnan

(DOE, GDPref 1 2,18)
The first stream of-the clean and clear spring
(b) For þære micclan and stiðan drohtnunge (DOE, LS7(Euphr)233)
Because-of the intense and severe way-of-life
There is something, however, that very strongly pleads in favour of Spamer’s analysis of weak adjectives as adjuncts, but which he does not discuss himself, and this is the complete absence of constructions such as,¹⁰

(12) *se/his swiþe ealda man

the/his very old man

Adjunctival adjectives (as Spamer qualifies the weak adjectives, see note 7), being of a nominal character cannot normally be modified with adverbs, cf. * a very stone wall. I have checked the complete DOE, and have not been able to find a single example with a weak adjective, neither with swiþe nor with ful. What we do find quite regularly is strong adjectives modified by swiþe, both when they are used predicatively, as in he wæs swiþe eald ‘he was very old’, in indefinite NPs, swiþe eald man ‘a very old man’, as well as when they are found postnominally as in,

(13)a and þær ofslogan anne brittiscne man swiþe æþelne

(DOE, ChronC(Rositzke) 501.1)
and there killed a Briton (man) very noble
(b) nu cweðað oft preostas þæt petrus hæfde wif ful soð (cf. 9c above)
now say often priests that Peter had wife full true
‘now priests often say that Peter had a very true wife’

The only exception is,

14. 
    His ful leof fæder (BlHom3 1.14, p.101-2)
    

his very dear father

Where ful may originally have been an adjective like the micel examples which I will discuss below.

What does all this suggest? Spamer seems to be on to something when he associates weak adjectives with adjuncts (which makes the adjective + noun a kind of compound, cf. note 7), but he is wrong about strong adjectives being determiners. He is on to something when he mentions non-recursiveness of adjectives but this seems to be true for both strong and weak ones (quantifiers excepted).¹¹ However, he has missed the significance of why adjectives may be positioned after the headword. Postnominal position is not an automatic consequence of the impossibility of serial strong adjectives, rather it seems to involve a choice, and it seems to make a semantic distinction. This distinction may well be linked up with strong- and weakness of adjectives, or with indefinite-/definiteness, with which these adjectives were traditionally linked.
I now want to go back to the ideas I mentioned at the beginning: the link between adjective position and definiteness, between adjective position and contrast, and between adjective position and linear iconicity. What I would like to suggest is (but this is still tentative, more research is clearly needed) that all strong adjectives in Old English are or started off as predicative adjectives and that weak adjectives tend to be adjunctive (in the sense of Spamer). The natural position for predicative adjectives would be after an NP followed by a copula, but they also came to be used straight after nouns in so-called secondary predicates (traditionally: subject or object complements), conveying in both cases additional and new information. The natural position for weak or adjunctive adjectives, on the other hand, would be prehead, conveying already known (and therefore definite) information (it would explain at the very least why postnominal adjectives are always strong (for some exceptions see Fischer 1999). This would link up with Bolinger’s idea that when there is a choice in position, then the prenominal adjective qualifies the noun and turns it into a different category. The postnominal adjective, on the other hand, gives extra information about the noun; the quality it expresses is not interpreted as inherent in the noun itself. The postnominal adjective in this sense has more of a verbal character (as said above, it functions like a secondary predicate), and can therefore also be modified by elements that can modify verbs, such as adverbs and prepositional phrases. This would then explain why in Old English degree adverbs like swiþe can only modify verbs and strong adjectives, as we have seen. It also explains why only strong adjectives can be followed by a PP (as far as I could tell from my data; and compare the remark made by Sørensen above)), as in,

(15)a God ða forð ateah of ðære moldan ælces cynnes treow fæger on gesyhðe

(Æl. Old Test.1 l.2.9)
God then forth out-pulled from the earth of-each kind tree beautiful in sight
‘God then pulled forth from the earth every kind of tree beautiful to see’
(b) swiðe geleafull wer welig on æhtum (Ælet4, l.737)

1. 
   very faithful man wealthy in goods
   

(c) bollan fulne hates wines (Læceboc 2.14.2)
cup full of-hot wine

Similarly, it would explain why in some cases the strong postnominal adjective itself is comparable to a PP, as in,

(16) and þær sint swiðe micle meras fersce (Mitchell § 126, Oros. 19.5)

and there are very big (many?) lakes fresh
‘there are very big fresh-water lakes’

The adjective fresc here does not qualify the lake itself but adds information about what is in the lake, and is thus comparable to the Dutch example een broodje gezond, discussed in section 2. Compare this also to (17b) below, where fresc modifies “water” and does stand before the headword. (Fischer 1999 goes deeper into the nominal and verbal character of the weak and strong adjectives respectively.)

The nature of the Old English adjectives, i.e. predicative and adjunctive, also explains why adjectives cannot really modify each other. Predicative adjectives are always parallel and not hierarchically ordered. Thus, one can say a beautiful young woman, but not the woman was beautiful young. This would explain why the strong adjective is not recursive (cf. Spamer’s claim but made here on different grounds) and why adjectives used postnominally must be strung together by means of and in Old English. Adjunctive adjectives are also more difficult to string together, because of their compound characteristic.
We saw above that we sometimes do find a series of strong adjectives without and prenominally in Old English (in my data, see Table 3, I have found very few examples of a postnominal series of adjectives without and). How are these then to be explained? First of all, the most usual string is that of two quantifiers following each other. A possible explanation for these strings is that the first quantifier acts like an indefinite determiner (which accounts for the fact that the adjective or quantifier following always has the strong form), while the second may be more adjectival in nature.
Concerning two “true” adjectives in a series prenominally, I have found eight examples in my data of adjective-adjective noun, where both adjectives are strong; one example of three strong adjectives and a noun, but these adjectives are all the same, so a different case altogether (halig halig halig Drihten, Ælet4, 1.1154; moreover a translation from the Latin sanctus sanctus sanctus); and two examples of a quantifier followed by two strong adjectives (see Table 2, rows 6, 8 and 7 resp.). Noteworthy about these examples is that the second adjective refers in six cases either to a material (æscenne ‘of ashwood’ (Læceboc1 13.10.1), beren ‘of barley’ (Læceboc1 19.4.1), rigenre ‘of rye’ (Læceboc3 159.1.1), sylfrenum (Documents4, 2a, 1.21)) or to a nation (brettisc (ChronA Early 1.501.1), englisc (Documents4, 2a 1.41)); in other words they are denominal adjectives. These adjectives (cf. Quirk and Greenbaum 1973, §13.40) stand always close to the noun in Modern English because they are the least adjective-like. It may be for this reason that this series is possible because the second adjective is so noun-like. If we are witnessing a change from purely parallel use of adjectives to serial use, this is where we might expect it to start. It is interesting to note, by the way, that adnominal adjectives are characterized by Spamer as typical examples of weak adjectives (adjuncts), but they are strong here. The other examples are more difficult to account for,

(17)a Swa beorht scinende steorra (Bede3.1.13.428.18)

Such (a) bright shining star
(b) Swetum ferscum wæterum (Læceboc2, 1.16.1.4)
with-sweet fresh water
(c) God hluttor eala (Læceboc3 1.30.1.1)
good clear ale
(d) Mid ofermæte unclæne luste (Mart1 1.150)
with excessive unclean lust

The example from Bede may have been influenced by the Latin text, and the fact that one adjective is participial may play a role too. Possibly the other three concern adjective-noun combinations which may well be idiomatic and can therefore be seen almost as a compound, thus allowing an other adjective in front of it. It could also be said that god and ofermæte express degree and could begin to function like adverbs modifying the next adjective, just as the micel-examples discussed below.

Concerning a series of adjectives between a determiner and a headword (so weak adjectives), I have found ten instances of adjective-adjective noun (Table 2, row 13), and one of three adjectives plus a noun (row 14). The examples are similar to the previous group in that in five cases a locality/nation is present (suðwesterna (Apollonius 1.11.11), frencisce (ChronE2.1.1087.1), romaniscan (Greg. Dial.2, ms H, B.5.8.2), macedonisc (Marvels 12.10, 2x)). Four are similar again in that the adjective may form an idiomatic unit with the noun (godan willan (Cura Past.3 1.53.417.12), synfulla man ‘sinner’ (WHom1.1.43), soða God ‘true God’ (WHom8.l.25), eadige fæmne ‘Mary’ (Marg. 1.350)), and there are two I cannot immediately account for,

(18)a þæt ofstandene þicce slipige horh (Læceboc2 1.16.1.14)

the remaining thick slimy mucus
(b) on þam æftemestan mæran freolsdæge (West Saxon G. 1.7.37)
on the last known (?) festival-day

It could be that ‘slimy mucus’ is an idiomatic unity too (note that the adjective and noun express more or less the same thing), and that participial adjectives behave somewhat differently, and that in the second example æftemestan functions like a kind of ordinal numeral.

A rather different group of the determiner-adjective-adjective-noun type concerns cases where the first adjective is agen ‘own’ (Table 2, row 15: three instances) or ylca/self (row 16: thirteen instances). To my mind these adjectives are different because they seem to form a unity with the determiner rather than with the noun. Note indeed that in ME the ilke often comes to be written as thilke.
Yet another special group (row 17 in Table 2) concerns the following type,

(19) On þære micclan niwan gecyðnysse (Ælet4 1.925)

In that big new testimony/ in that very new testimony

Here it is not entirely clear whether micclan functions as an adjective to the noun or as an adverb to niwan. Both interpretations would make sense. Note that the ending in

–an in micclan functions both as a weak ending and as an adverb according to the traditional Old English grammars, but the adverbial usage of micclan may well have started in these constructions and have developed from an original adjective because