ASPECTS OF OLD ENGLISH WORD-FORMATION: COMPOUNDS IN THE OE ELEGIES AND THE CAPITULA OF THEODULF

Abstract

The paper analyses compounds and related formations in some of the Old English elegies as well as in the ThCapA (Theodulfi Capitula, A version). Both texts (or groups of texts) show the productivity of compounding, and both show that noun + noun compounds were the most productive type of compounds. But they also show significant differences: Whereas many compounds in the elegies are due to the poetic principle of variation, some compounds in the ThCapA are or may be loan-translations based on Latin models.

Headnote

Abstract: The paper analyses compounds and related formations in some of the Old English elegies as well as in the ThCapA (Theodulfi Capitula, A version). Both texts (or groups of texts) show the productivity of compounding, and both show that noun + noun compounds were the most productive type of compounds. But they also show significant differences: Whereas many compounds in the elegies are due to the poetic principle of variation, some compounds in the ThCapA are or may be loan-translations based on Latin models.

Keywords: compounds, morphology, Old English, semantics, variation, word-formation

1. Introduction

This paper is intended as a contribution to a survey of Old English word-formation. I shall point out some principles as well as some problems of analysis and classification - of course in the scope of a paper it is not possible to be comprehensive. There have been many studies of specific aspects of OE word formation, but there has been no book-length study dealing with Old English word-formation as such; to my knowledge the most recent compact survey is still Kastovsky (1992); for a discussion of compounds (henceforth cpds1), see now also Davis-Secord (2016).

The texts analysed here illustrate on a small scale how wide the scope of OE literature is: The three OE Elegies (W = Wanderer, S = Seafarer, D = Deor) represent alliterative poetry; they show a heroic and oral society, with Christian elements appearing at the end. Moreover, the language of OE poetry was a special language and not identical with prose language or everyday language: for example, some words (including compounds) are attested only in poetry and not in prose; the principle of variation is also characteristic of OE poetry (see, e.g., Godden 1992 and 3.4.2 below). We do not know what the intended audience was - but since the manuscript was probably copied in a monastery, monks were apparently also interested in this kind of poetry.

The Capitula of Theodulf (ThCap) represent OE prose, more specifically didactic Christian prose, and they are addressed to a specific audience (at least originally), namely parish priests. From the beginning, they show a Christian and literate society, literate at least as far as the priests are concerned: reading is important, but the oral component is also there, because the priests, who can read (or at least are supposed to be able to read) have to preach the basic tenets of Christianity to the laity, the majority of whom was probably still illiterate.

Moreover, the elegies are original OE compositions (as far as we know), whereas the OE Capitula are translations from Latin; with respect to wordformation, this raises also the question which compounds are loan-formations based on Latin models (see 3.5. below). Actually, there are two independent translations, which I have called ThCapA und ThCapB. Whereas ThCapA gives a translation of the complete text, ThCapB is only preserved as a fragment, which now begins in the middle of chapter 25. Here I concentrate on the beginning of ThCapA, Prologue to chapter 11. ThCapA is also a somewhat freer translation than ThCapB.

The dating of the OE elegies is very uncertain - the fact that they depict an archaic and heroic society does not necessarily mean that they are early compositions (cf. also Davis-Secord 2016: 101); they could also be late manifestations of the tradition of secular and heroic poetry. The terminus ante quem is the manuscript, in this case the Exeter Book, which was written around 1000 A.D.

The dating of the OE ThCapA is also uncertain. The text is transmitted in a manuscript written at Exeter in the third quarter of the 11th century (now Cambridge, Corpus Christi College 201), i.e. shortly before or perhaps even shortly after the Norman Conquest. The original translation may have been made earlier, however, perhaps in the second half of the 10th century or in the early 11th century. The Latin original was written around 800; thus the OE translations must in any case have been made later.

For the OE elegies (W, S, D) I have used the edition in ASPR III, but I have also consulted the separate editions of W by Leslie 1966 and by Dunning & Bliss 1969, of S by Gordon 1960 and by Cucina 2008, of D by Malone 1933 [1977]; for the ThCapA I have used the edition by Sauer 1978. On the manuscripts, see especially Ker 1957, nos. 50, 116, 318; Gneuss and Lapidge 2014, nos. 66, 257, 608; furthermore Sauer 1978: 88 ff.; Hofstetter 1987, nos. 132 and 133.

2. The main types of Old English word-formation and some problems of their classification

The texts dealt with there show the main types of OE word-formation (hereafter wf), namely compounding, prefixation, suffixation, and derivation without a suffix (sometimes also called conversion or zero-derivation). Here I concentrate on compounds and related phenomena. Other types which are often mentioned in discussions of Modern English word-formation are absent in these texts and are generally rare or even non-existent or in any case difficult to recognize in OE, namely back-formation, sound-symbolic wf, wf with reduplication (including rhyme and ablaut), clipping, blending and acronyms.

2.1.Some problems of analysis

(1) Main types or subtypes. All types have subtypes, but with some patterns it is difficult to decide whether they should be regarded as types in their own right or as subtypes of other types. Compounds, for example, can be roughly defined as combinations of two words. Combinations with particles (e.g. ModE in-, out-, over-, under-) should therefore be regarded as compounds, because particles are also used as independent words. Some handbooks on wf label them as prefixes, however (e.g. Koziol 1972), probably because semantically they are closer to prefixes than to words; in any case particles belong to the category of grammatical words, not to the lexical words. Similar to Marchand 1969, I propose to put them into a separate category, but, for reasons of space, I do not deal with them here; examples from the texts dealt with here are æ fter-cweþende (S) 'posterity', lit. 'those speaking after'; geond-hweorfan 'visit every part (of)', lit. 'move throughout'.

Conversely, some elements (such as cyne-) do not occur independently, but are semantically closer to words, or they are some way in between words and affixes; I have put them into the category of affixoids (prefixoids and suffixoids; see ?6 below). Accordingly, in the following, I deal with compounding (compounds) and with affixoids.

(2) Recursivity. This term refers to the phenomenon that complex words, including compounds, can contain elements that are complex in themselves, i.e. that are derived, or to put it the other way round: complex words can become part of new word-formations. Here the question is whether the word-formation processes worked in sequence, or simultaneously. Even for ModE this is not always easy to answer. Both street-fight and taxpayer contain a deverbal second elements (fight noun is derived from fight verb), but street-fight is probably easier to analyse as a noun + noun cpd, 'a fight in the streets' (rather than 'fighting in the streets'), whereas taxpayer is probably easier to analyse as a simultaneous application of compounding and suffixation, i.e. 'someone who pays taxes' (rather than as 'a payer of taxes'). An additional problem is that some of the deverbal elements are also attested independently, but others are only attested as second elements of cpds. One possible solution of the problem is to provide a double classification, e.g. of taxpayer under compounds and under derivations with the suffix -er.

2.2.Compounds and compounding

(1) Subtypes. Compounds can primarily be subdivided into compound nouns (or noun compounds, or substantival compounds) and compound adjectives (adjective compounds or compound adjectives), each with further subgroups. Compound verbs are a problematic group, because they are usually derived from compound nouns etc., in OE as well as in ModE: OE cynehelmian 'to crown', for example, is not a compound consisting of ·cyne- + helmian, but derived from the compound cynehelm 'crown', lit. 'royal helmet, royal diadem' (for cyne- see 6.1 (1)below), and ModE spotlight (verb) is a conversion (or zero-derivation) from spotlight (noun). Verbal cpds are thus cpds if they are analysed from a purely synchronic point of view, but derivations, if their origin is taken into consideration, i.e. applying synchronic and diachronic criteria in this case leads to different results (cf. also Hubner 2017, chapter 4.5.2.). But the material analysed here does not contain any cpd verbs (apart from verbs the first element of which is a particle; see above), so we need not enter into a detailed discussion here. There are also a few cpd pronouns (nan-þing), adverbs (heonon-forð) and conjunctions (for-þon), but again I shall not discuss those any further.

(2) Productivity. The productivity of compounding is also shown by the fact that the same word could be used in various cpds. I list the following (without claiming completeness):

(a) Repeated first elements, especially nouns, from W and S (& ThCapA) are, e.g.: breost 'breast' (W: breost-cofa; S: breost-hord; breost-cearu); brim 'sea, water' (W: brim-fugol; S: brim-lad); eard 'earth, land, dwelling-place'(W: eardgeard, eard-stapa); gold 'gold' (W: gold-wine; S: gold-giefa); medu, meodu 'mead' (W: meodu-healle; S: medo-drinc); niht 'night' (W: niht-helm, niht-scua), sele 'hall' (W: sele-dream, sele-secg); stan 'stone' (W: stan-hlið; S: stan-clif).

(b) Repeated second elements, are, e.g., cofa 'chamber, cave' (W: breostcofa; hord-cofa); dryhten 'lord' (W: mon-dryhten; wine-dryhten); geard 'dwelling, land'(W: eard-geard; cf. also middan-geard); giefa, gyfa 'giver' (S: gold-giefa; W: maþþum-gyfa); hama, homa 'covering' (S: flæ sc-homa; ThCapA: lic-homa); lad 'way, course' (W: lagu-lad; S: brim-lad); loca 'enclosure' (W: ferð-loca; S: hreðer-loca); weg 'way' (S: flod-weg; hwæ l-weg).

(c) Words that are used as first and as second elements are, e.g., cearu, caru 'care, sorrow' (S: breost-cearu; cear-seld); cwide 'speech, word'(W: cwide-giedd; lar-cwide); lāst 'footprint, trace' (S: last-word; wræ c-last); mæ g 'kinsman' (maguþegn; wine-mæ g) wine 'friend' (NOT ModE 'wine'; W: wine-mæ g, wine-dryhten; gold-wine).

(d) Examples from the first few chapters of ThCapA are: Repeated first element: lic 'body,corpse' (lic-homa; lic-tun) and mæ sse 'mass' (mæ sse-preost; mæ sse-reaf; mæ sse-sang); repeated second element: man, mon 'man, human being' (ambiht-man; wif-mon) and weorc 'work, action' (hand-weorc; weorold-weorc). As these examples show, the use of the same element in various cpds was also common in prose, i.e. it is typical of compounding in general.

(e) Elements of cpd adjectives that occur repeatedly are -ceald, -cearig and -gifre: ceald and gifre occur twice as second elements, cearig even three times: ceald 'cold' (W: hrim-ceald; S: is-ceald); cearig 'sorrowful' (W: mod-cearig; D: sorg-cearig; W: earm-cearig); gifre 'greedy' (W: feoh-gifre; wæ l-gifru).

3. Compound nouns: noun + noun

3.1.Number and frequency

The most frequent type of compounding and one of the most productive patters of wf generally were the noun + noun cpds; apparently they have been the most productive pattern of compounding throughout the history of English. Therefore discussing them will take up most of the space; the other types of compounds are much rarer and can be dealt with quicker. There are 36 different noun + noun cpds in W, and 27 different noun+noun cpds. in S; together W & S have 59 different cpds - the sum is a little less than 63, because four cpds. occur both in W & in S (mod-sefa, niht-scua, geswinc-dagas, wine-mæ g). Deor as a very short text has only 4 noun + noun cpds from the general vocabulary, plus five names, and ThCapA has 11 different cpds in its first chapters; thus the texts analysed here have 74 different noun + noun cpds altogether, and five names.

3.2.Some problems of analysis and classification

For a number of reasons it is, however, difficult to give precise numbers. The problem of how to classify deverbal elements has been alluded to above (see 2.2 (1)); other problems are posed by obscured compounds and by prefixoids and suffixoids (on the latter see also 6 below).

(1) Obscured cpds are original cpds where one or both elements were weakened and shortened, so that their character as cpds was lost, and the words making up the compound developed differently in independent use (cf., e.g, Götz 1971; Faiss 1978). A standard example is ModE lord. It consists of one morpheme from a purely synchronic point of view, but it goes back to an OE cpd hlāf-weard lit. 'bread-guardian'; however, even in OE it was mostly used in the obscured form hlaford (as such also attested in D), where the second element was weakened to - ord and probably no longer associated with weard. Another example (from ThCapA) is weofod 'altar', an obscured form of wigbed lit. 'idol-table' ('table where a idol was put'; see, e.g., Gneuss 1955: 85; Käsmann 1961: 210-214). Formations attested in our texts which were still transparent cpds in OE, but were obscured later, are, e.g., wifmon (> ModE woman) and, from the adj + noun cpds, godspel (> ModE gospel; see below). I have not counted hlaford and weofod among the cpds here.

(2) Word into suffix. Another reason why it is difficult to give precise numbers is the phenomenon by which some words lost their status as independent words and changed into suffixes; see further ?6 below. As a consequence, some combinations (especially those with -dom and -had) were probably still compounds in OE, but have to be regarded as suffix-formations for Middle English and Modern English.

(3) Hapax legomena and poetic compounds. Fairly many of the cpds in W and S, namely 19 altogether (out of 59, i.e. roughly a third), are hapax legomena, i.e. they are attested only once (see, e.g., Dunning and Bliss 1969: 36; Cucina 2008). The hapax legomena were apparently ad-hoc formations and not really part of the common OE vocabulary, but they show the productivity of the pattern of noun + noun cpds, which was apparently almost unlimited in OE. It must have also been relatively easy to understand these formations. The nine hapax legomena among the noun + noun cpds in W are: brim-fugol, cwide-giedd, eard-stapa, gliwstæ f hæ gl-faru, maþþum-gyfa, sele-secg, weal-steal, win-sæ l; the ten hapax legomena among the noun + noun cpds. in S are: bealo-siþ, cear-seld, earfoð-hwil, hrim-gicel, hwæ l-weg, hring-þegu, last-word, medo-drinc, niht-waco, tid-deg.

Moreover, many of the cpds in W and S occur only in poetry, 19 in W and 14 in S, i.e. 33 altogether. Poetic cpds from W are (ClHall marks poetic cpds with a dagger): breost-cofa, byrn-wiga, eard-geard, ferð-loca, gold-wine, hord-cofa, lagulad, lar-cwide, magu-þegn, meodu-healle, mod-sefa, mon-dryhten, niht-helm, nihtscua, sele-dream, sinc-þege, stan-hleoþ, wine-mæ g, wræ c-last. Poetic cpds from S are: breost-hord, brim-lad, ecg-hete, flæ sc-homa, flod-weg, gold-giefa, hleo-mæ g, hreþer-loca, mere-flod, mod-sefa, niht-scua, geswinc-dæ g, wine-mæ g, wræ c-last.

The hapax legomena can, of course, also be counted among those formations that are only attested in poetry. If we combine the number of hapax legomena and of poetic compounds, they add up to 28 poetic cpds in W (out of 36 noun + noun cpds altogether), and 24 poetic cpds in S (out of 27 noun + noun cpds altogether), i.e. in W and S, 52 out of 59 cpds are poetic cpds - or to put it the other way round, apparently only a small minority of the cpds in W and S were part of the common OE vocabulary (seven out of 59). Among the cpds also attested in OE prose texts are, e.g., eorð-scræ f (W), woruld-rice (W), eorð-wela (S), stan-clif (S), and middan-geard 'earth' lit. 'middle-earth'; the morphology of the latter is difficult to analyse, however (see 3.3(3) below); The first chapters in ThCapA have only one hapax legomenon, namely ungemethleahtor. This shows perhaps that hapax legomena are on the whole more frequent in OE poetry than in OE prose.

3.3.Morphology: typical and less typical patterns

(1) Typically noun + noun compounds consist of two primary nouns, e.g. breost-cofa 'heart', lit. 'breast-cave', meodu-healle 'mead-hall', or wine-mæ g 'dear kinsman', lit. 'friend- kinsman'. But some formations pose problems, to mention just two:

(2) Compounds with a deverbal element are not quite so typical. As mentioned above (cf. 2.1(2)), it is not always easy to decide whether cpds with such elements should be synchronically classified as noun + noun cpds or as cpds with a deverbal element. In such cases, I propose to use a cross-classification and to list them among the noun + noun cpds as well as among the deverbal derivations without a suffix. A number of noun + noun cpds contain a deverbal element, e.g., from W: hæ gl-faru (in ClHall listed s.v. hagolfaru) 'hailstorm' (faru is derived from faran, strong verb VI); maþþum-gyfa 'generous chieftain', lit. 'treasure-giver' and gold-giefa 'gold-giver' (gyfa, giefa is derived from giefan, strong verb V), sinc-þegu 'receiving of treasure' (þegu is derived from þicgan 'to receive', strong verb V); the examples also show that with strong verbs, the basis for the derivation of nouns was sometimes the present stem (faran > faru; giefan > giefa, gyfa) , but sometimes also one of the stems of the past or the past participle (þegu). Geswinc in geswinc-dagum could be the present stem of the strong vb (ge)swincan 'to labour, struggle', or the deverbal noun (ge)swinc 'work, toil, effort' derived from (ge)swincan. Stapa 'someone who goes, steps' (W: eard-stapa) is derived from the strong vb VI stæ ppan 'to step, go'.

(3) Formal equivalence of noun and adjective. Often a noun and an adjective had the same basic form in OE; in such cases it is difficult to decide whether a given formation should be assigned to noun + noun cpds or to adj + noun cpds. Instances where an element could be a noun or an adj are: earfoð(e) 'hardship, suffering' or 'hard, troublesome' in earfoð-hwil; midd(e) (in middan-geard 'earth' lit. 'middle-earth' - apparently the model for Tolkien's middle earth), where midd(e) could be the adjective midd, midde 'mid, middle', or the noun midde. I have classified both earfoð-hwil and middan-geard as adj + noun cpds (cf. 4.2. below). An additional morphological complication with middan-geard is that middan is inflected according to the weak declension. Inflexion of the first element is unusual in cpds; this is perhaps also the reason why genitival cpds (see 4.1. below) are much rarer than noun + noun cpds. This does not help, however, to determine the word-class of middan, because the weak consonantal declension would be the same for the noun and the adjective.

But even if those formations are counted as adj + noun cpds and not as noun + noun cpds, the group of noun + noun cpds still remains very large, whereas the the group of adj + noun cpds remains very small; see further 4.2. below.

3.4.Semantics

Noun + noun cpds are not only the most frequent group among the cpds, they also show the most varied semantic patterns. Moreover, semantic analysis works on (at least) two levels: The literal relation between the elements can be described and classified (see 3.4.1.), but in some cases the meaning of the entire cpd is not just the sum of the meanings of its parts, but has additional (or lost) meaning components; the latter phenomenon is captured by terms such as idiomatization and lexicalization and will be dealt with a little later (? 3.4.2.).

It has been claimed that cpds have a unified meaning, but it is difficult to decide what a unified meaning is, and perhaps not all cpds have a unified meaning. The ModE translations of OE cpds are not very helpful in this respect: for example, both breost-cofa and brim-fugol are generally regarded as cpds, but breost-cofa, lit. 'breast-chamber', is often translated as 'heart', which would point to a unified meaning, whereas brim-fugol is usually translated as 'sea-bird', where the two elements and their meaning are still transparent. Modern dictionaries and glossaries are not very helpful for this kind of analysis. Often they translate an OE cpd (in our case a cpd from W, S, D) with one ModE word, sometimes with a noun preceded by an adjective, and sometimes with a compound (definitions taken from Dunning and Bliss 1969), e.g. (1) simple word: breost-cofa 'heart', eard-geard 'city', eardstapa 'wanderer', etc.; (2) noun with a preceding adjective: e.g. anhoga 'solitary man', byrn-wiga 'mailed warrior', cwide-giedd 'spoken utterance'; (3) cpd, e.g. brim-fugol 'sea-bird', but dictionaries and glossaries do not always agree: Leslie (1966) and Dunning and Bliss (1969) sometimes have the same definitions, but sometimes different ones. Leslie, for example, renders eard-geard with a cpd, 'dwelling place', but byrn-wiga, 'warrior', and cwide-giedd, 'utterance', with a simple ModE word. Dunning and Bliss do it exactly the other way round; they explain eard-geard, 'city', with a simple word, but byrn-wiga, 'mailed warrior', and cwide-giedd, 'spoken utterance' with a phrase - this shows that ModE explanations are not reliable guides as to whether OE (poetic) cpds have (or had) a unified meaning (see further 3.4.2. below). Therefore an analysis according to the concepts of idiomatization or lexicalization seems more promising. One could say that non-lexicalized cpds are self-explanatory, whereas lexicalized or idiomatized cpds are not. But, as I shall argue in 3.4.2., a distinction between meaning and reference should also be made (at least in principle).

3.4.1.The semantic relation of the elements

There have been many attempts at a semantic subclassification of noun + noun cpds, often taking the reference of one element or the relation between the elements as a starting point, but none is entirely satisfactory (cf., e.g., Sauer 1992; Hubner 2017, chapter 7.3).

One problem is the following: it is one function of cpds to condense information, but for the analysis, this means that one compound can simultaneously contain several semantic structures which have to be disentangled. Win-sæ l (winsalo) 'wine-hall' and meodu-heall 'mead-hall', for example, refer to 'the place where wine or mead is drunk' ('A is drunk in B', or 'the warriors drink A in B'), but also to the purpose of the hall, namely that its function (or one of its functions) is to drink wine or mead there ('B is intended for drinking A'), and it is difficult to say whether one of those meanings (place; purpose) is more dominant than the other. Another problem is that many words are polysemous (at least according to their dictionary definitions), and if they are a part of a compound it is often difficult to say which of their meanings is the relevant one. For example, wæ l is defined as 'slaughter, field of battle' and sleaht (slieht) as 'slaughter, murder, death, battle'; wæ l-sleaht could accordingly mean 'slaughter-slaughter' (in which case it would be a tautologic copula cpd; see also below), or 'battle-slaughter, slaughter during the battle, slaughter on the battle-field' (in which case it would be a rectional cpd). A further problem is that for a (syntactic) paraphrase of noun + noun compounds a verb has to be inserted, and even for ModE this is not always easy, e.g., should police-dog be paraphrased as 'the dog works for the police' (with dog as subject), or 'the police uses the dog' (with dog as object)?

A widely accepted distinction is between copula compounds and rectional cpds. In copula cpds, both elements have a (more or less) equal status, and in the paraphrase, 'and' or a form of 'to be' has to be inserted. In rectional cpds, the second element dominates semantically and another verb (different from 'to be') has to be inserted. The latter have many subgroups; here they are subdivided into compounds referring to: people, place, time, material, instrument, part-whole and whole-part, result, action. This classification can accommodate most cpds attested in our texts.

I) Copula compounds

Copula cpds, i.e. A and B are semantically on the same level ('A is also B', or 'A and B taken together'). Here belong probably: wine-mæ g (W and S; also in the plural wine-magas) 'friend-relative', i.e. 'a relative, kinsman who is also a friend', and wine-dryhten (W) 'friendly lord' lit. 'friend-lord', i.e. 'a lord who is also a friend', and cwide-giedd (W) 'utterance, speech' lit. 'speech-speech' or 'saying-saying' - i.e. cwide-giedd may even be tautologic. Woruld-rice 'world', lit. 'world-kingdom', is probably to be analysed as copulative 'The world is a rice [kingdom]' (perhaps as opposed to heaven); eard-geard 'city, earth' (eard 'country' etc.) lit. 'dwelling-place, country - dwelling, land' can be analysed as a copula cpd 'the land is a dwelling-place'. From S could be added mod-sefa 'heart, mind' lit. 'mind-mind'.

Other formations are more complicated: they could be analysed as copula cpds, but also as rectional cpds, e.g.: sorg-lufu (D) lit 'sorrow-love' could be a copula cpd. (and simultaneously an oxymoron, expressing semantic opposites), but also a rectional cpd, if it is analysed as 'a love that causes sorrow' - from the context, the latter interpretation seems more probable (Malone glosses it as 'distressing love', ClH as 'sad love'). Wæ l-sleaht 'battle' can be analysed as a copula cpd or as a rectional cpd, see above; weall-steal lit. 'wall-place' can be analysed as copulative 'the wall is a place', or perhaps also as rectional 'a place near the wall'. On brim-lad and lagu-lad see 2) below. An example of a copula cpd is probably also wif-mon > 'woman' (ThCapA); the original meaning must have been something like 'human being [man] who is a female [wif]'.

II) Rectional compounds

As mentioned above, rectional cpds have many subgroups, and there are several ways of classifying the subgroups. From the elegies and the ThCapA, the following subgroups emerge:

(1) People, including agents (i.e. B refers to a person). Examples from W are: byrn-wiga ('mailed warrior, corsleted warrior', i.e. 'warrior wearing a mailcoat'; in more abstract terms, 'B wears A', i.e. A referring to an instrument); eard-stapa 'wanderer', lit. 'country-stepper, someone who walks across the land' (i.e. A referring to a place; see further 3.4.2. below); gold-wine 'gold-friend' ('friend who gives gold', i.e. A refers to an object); sele-secg 'retainer' lit. 'hallman, hall-warrior' (A again referring to a place); from S: hleo-mæ g 'protecting kinsman, kinsman who provides protection' ('B provides A'), and from W and S cpds. with gifa, gyfa 'giver' as second element: maþþum-gyfa treasure-giver', goldgiefa 'gold-giver' (B gives A'), i.e. their semantic structure is similar to gold-wine, but whereas in gold-wine the element 'giver' is not expressed on the surface of the cpd and has to be supplied in the paraphrase, it is expressed on the surface of maþþum-gyfa and gold-giefa. Examples from ThCapA are ambiht-men 'servants', lit. something like 'people providing a service' and mæ sse-preost 'priest', lit. 'mass-priest', i.e. 'priest who celebrates mass'.

(2) Place (location), i.e. A indicates the place where B lives or is situated or takes place, or B indicates the place where A lives or takes place. Examples from W are: breost-cofa 'heart', lit. 'breast-cave' ('the cave is in the breast', or 'the breast contains a cave', etc.); brim-fugol 'sea-bird', eard-geard 'dwelling-place', eorð-scræ f 'grave', lit. 'earth-cave' ('a cave in the earth'); ferð-loca 'breast', lit. 'mind-enclosure', meodu-heall 'mead-hall' (see above); sele-dream 'hall-joy, hallrevelry' ('B takes place in A'). The adj + noun cpd middan-geard 'earth', lit. 'middle-land' (i.e. land between heaven above and hell below), also belongs here (see above). From S can be added: breost-hord, breost-cearu, cear-seld, eorð-wela, flod-weg, hwæ l-weg, hreðer-loca, wræ c-last. See also? 3.4.2 below. Cpds. with the same second element are brim-lad (S) and lagu-lad (W), both meaning 'water-way, sea-way' (i.e. 'the sea provides a way', or 'the sea is a way', which would make it a copula cpd.). Examples from ThCapA are: gebed-hus 'chapel' lit. 'prayer-house' ('people pray in that house'); lic-tun 'cemetery' lit. 'corpse-enclosure'.

(3) Time. Examples from W are: gear-dagas 'old days, days of yore' lit.'year-days' (both elements referring to time); niht-helm 'cover of night', lit. 'night-covering, night-protection' ('A provides B'; A referring to time); niht-scua 'shadow of night', lit. 'night-shadow' ('A provides B'; in W and S); geswinc-dagas 'days of tribulation' lit. 'tribulation-days'; from S can be added: niht-waco 'nightwatch' ('B takes place during A').

(4) Material, i.e. 'A consists of B', or 'B consists of A' (or 'B characteristically contains A'): An example from W is: stan-hleoþ 'rocky slope, stone-cliff ('B consists of A'); examples from S are: hrim-gicel 'icicle'; medodrinc 'mead-drink' ('B consists of A'); mere-flod 'sea', lit. 'sea-flood'; sealt-yþ 'sea-wave', lit. 'salt-wave' (i.e. 'B contains A')

(5) Instrument, i.e. 'B is for A, B provides A'; instruments usually also have the semantic element 'purpose'. Examples from W are: gief-stol 'gift-seat', i.e. 'seat where the lord distributes gifts to his warriors' ('seat the function of which it is to distribute gifts from it'); gliw-stæ f 'melody, joy', lit. 'pleasure-staff (perhaps 'B produces A'; but here used adverbially gliw-stafum 'joyfully'); lar-cwide 'counsel', lit. 'advice-speech' ('B provides A'); furthermore the cpds. with homa, hama 'covering' probably belong here: lic-homa (ThCapA) 'body', lit. 'bodycovering', 'B covers A', 'B is for covering A' (but homa is strictly speaking pleonastic, because lic already has the meaning 'body'), and similarly flæ sc-homa (S) 'body' lit. 'flesh-covering'; semantically in both lichoma and flæ sc-homa a part ('body-covering', 'flesh-covering') stands for the whole ('body'). Examples from ThCapA are husel-fæ t 'sacramental vessel' and mæ sse-reaf 'mass-vestments' i.e. 'vestments used for mass'.

(6) Part-Whole; Whole-Part (belonging to).This is a frequent relation in ModE, but there do not seem to be any clear examples in the material analysed here. Possible examples are flæ sc-homa 'body', lit. 'flesh-covering', and lic-hama, lic-homa 'body', lit. 'body-covering', if flæ sc and lic are taken to refer to the whole, and homa to the part.

(7) Result, effect ('B brings about, causes A'), e.g. last-word (S) 'fame after death', lit. 'footprint-word' (i.e. 'word that leaves a footprint, a trace'); geswincdagas (W and S) 'days of tribulation'; wræ c-last (S) 'path of exile, exile-footprint'.

(8) Action. It is often said that actions are typically expressed by verbs, but actions can also be expressed by nouns (often, but not always, deverbal nouns): Examples from W are: hæ gl-fare 'hail-storm', lit. 'hail-going'; sinc-þege 'receiving of treasure'; an example from S is sæ -for 'sea-voyage'. Here belong also the cpds with sið 'journey' as a second element: bealo-sið 'destruction, death', lit. 'destruction-journey' ('a journey that leads to destruction'). Examples from ThCapA are mæ sse-sang 'mass', lit. 'mass-song, mass-singing' and ungemethleahtor 'excessive laughter'.

3.4.2.Lexicalization, idiomatization, and reference

As mentioned above, a second layer of meaning and a second step in the semantic analysis is when the meaning of a cpd is not the sum of the meaning of its elements, but when the meaning of the whole is more than the meaning of its parts, i.e. when there are additional (or lost) semantic elements. If the meaning cannot be inferred (or can only be partly inferred) from the meaning of the elements, then a cpd. is labelled as idiomatized or lexicalized. But especially poetic compounds often have both a literal and a transferred meaning, and one apparently does not exclude the other: gold-giefa literally is the 'gold-giver', but it refers specifically to the lord, who gives gold to his men. Perhaps a third aspect should be added, and a distinction between meaning and reference should be made; but probably this will be easier to make in theory than in practice. The idea that cpds have a unified meaning is not very helpful here (see 3.4 above).

A curious case is eard-stapa, translated by dictionaries and glossaries as 'wanderer', lit. 'earth-stepper'. Stapa is a deverbal noun and its literal meaning is 'stepper, someone who steps', but when stapa is used (apparently very rarely) as an independent word, the dictionaries give its meaning as 'grasshopper, locust' - i.e. as a (rare) independent word, stapa has a lexicalized meaning; but as second element of the compound eard-stapa, lit. 'earth-stepper, land-stepper', it has its literal meaning 'someone who steps, wanders'.

But self-explanatoriness and lexicalization do not always exclude each other, at least not in OE poetry; this has also to do with the principle of variation. Variation is one of the techniques of OE poetry, which is also used in the OE Elegies: the same person or thing or concept is referred to with various words, including compounds - those may or may not be synonymous, but they have identical reference. Three concepts in the OE Elegies, where variation is particularly noticeable, are 'lord, ruler', 'man, warrior', and 'breast, heart, mind'.

(1)In W, there are four compounds that refer to the lord or ruler, namely gold-wine, lit. 'gold-friend' (i.e. a friend who gives gold [to his warriors]), maþþum-gyfa (W) lit. 'treasure-giver', mon-dryhten, lit. 'man-lord' ('B rules, commands A'), and wine-dryhten, lit. 'friend-lord' ('a lord who is also a friend [of his warriors]'); from S, there can be added gold-giefa, lit. 'gold-giver'. There are even more expressions referring to the lord, if phrases are included, e.g. sinces brytta 'treasure's giver' (semantically equivalent to the cpd maþþum-gyfa). All of those refer to the 'lord, ruler, chieftain', but he is actually called dryhten 'lord' in only one of them (wine-dryhten). These compounds and phrases stress various aspects of the lord, i.e. that he is a friend of his men and that he gives them treasures, including gold - this is at least the image presented in the OE elegies (cf. also Beowulf); how far this reflected social reality in Anglo-Saxon England is yet another question (see also the following section).

(2) The men and warriors are also referred to with four noun + noun cpds. in W: byrn-wiga 'mail-coat warrior, corslet warrior'; magu-þegn 'young retainer', lit. 'young man - retainer' (copulative); sele-secgas 'hall-retainers, hall-warriors' (probably referring to the image of warriors who celebrate and also sleep in the hall); wine-magas 'friends and kinsmen' (also copulative); from S, one can add hleo-mæ g, lit. 'protection-kinsman', i.e. 'kinsman who offers protection'. If other types of cpds are taken into consideration, then freo-magas 'free kinsmen' (W) also belongs here. They stress various aspects of the warriors, namely that they wear armour (byrn-wiga), that they are young (magu-þegn), that they celebrate and also sleep in the hall (sele-secg), that they are friends and even relatives (wine-magas). As these formations show, often there is hardly a distinction between the concepts 'man', 'warrior', 'kinsman', relative'. How far these compounds are poetic images and how far they reflect social reality is, of course, very difficult to tell; one of the problems is that society between the beginning of Anglo-Saxon England (ca. 450) and the time of the Exeter Book (ca. 1000) had certainly changed.

(3) The concepts of 'breast', 'heart', 'mind', 'thought' are difficult to distinguish in W; these are also expressed by four cpds, namely breost-cofa, lit. 'breast-cave, breast-chamber', ferð-loca 'mind-enclosure', hord-cofa 'treasurechamber, treasure-cave', usually taken to refer to the 'heart'; mod-sefa (W, S), lit. 'mind-mind' (copulative) or 'mind-heart'. Most of those refer to the heart or mind as a cave or enclosure, or as something that is enclosed (for a more detailed discussion with further references see, e.g., Davis-Secord 2016: 97-101).

I have mentioned above that it is often said that one of the functions of cpds is the condensation of information, but this is apparently only partly true of the poetic cpds, which often expand information and thus contribute to the poetic style. As has just been shown, this is at least partly due to the principle of variation; it can also be seen from the fact that there are relatively many copula cpds, where both elements are of equal weight - as opposed to rectional cpds, where normally the second element expresses the more general concept, which is specified by the first element (meodu-healle, e.g., is a specific kind of hall, namely a hall where mead is drunk).

Some cpds in OE poetry are kennings (kenningar; or can be analysed as kennings), i.e. they stress a relatively minor aspect of the referent which does not come first to mind when thinking about the referent (the classic study of OE kenningar is Marquardt 1938). Thus the sea is called lagu-lad in W and brim-lad and flod-weg in S, all three meaning 'water-way' - the sea is a way for its inhabitants (fish, whales) and for ships and boats, but this is perhaps not its most prominent aspect. Cucina (2008: 397-414) regards eight of the noun + noun cpds from S as kennings, namely breost-hord, brim-lad, ecg-hete, flæ sc-homa, flod-weg, gold-giefa, hreþer-loca, last-word.

3.5.Loan-formations

Whereas the cpds in the elegies seem to be original formations, with some of the cpds in ThCapA the question is whether they are loan-formations. Loanformations are imitations of foreign words with native material, often resulting in cpds; in the case of OE, the model was usually a Latin word. The assumption is that the OE word in question did not exist before. This is relatively easy to show with many words expressing Christian concepts, because it can usually be assumed that the Christian concepts were not known to the Anglo-Saxons before their conversion to Christianity; it is often more difficult to show with words belonging to other semantic fields, e.g. names for people or for certain professions; see also the examples given under (4) below. The likelihood of a cpd being a loanformation is increased if there is close morphological correspondence between the OE word and its presumed Latin model. Once they have been coined, loanformations are often used independently; they do not have to be used together with their model.

In ThCapA (Prologue to chapter 11) there are three noun+ noun cpds that are likely to be loan-formations; moreover, there are two other cpds and a syntactic group that could be loan-formations, so that there are six formations that are or might be loan-formations. The most recent monograph dealing with OE loanformations is still Gneuss 1955, who concentrates on the OE psalter glosses; but more research, taking other texts into account, is still needed - one semantic field where many OE loan-formations based on Latin models occur are the OE plant names (see, e.g., Sauer and Kubaschewski 2018).

A subdivision of loan-formations is into loan-translations, loan-renditions and loan-creations. Loan-translations show the closest correspondence and are accordingly easiest to detect; in loan-renditions usually only one element corresponds precisely to the Latin model; loan-creations were triggered off by the Latin model, but they show no formal correspondence.

Certain or possible loan-formations from ThCapA are: noun + noun: gebedhus (for domus orationis), lic-tun (for cimiterium, coemeterium), hand-weorc; adj + noun: god-spell; particle-combinations: ymbstandenda[s]; syntactic group: bisceopa gemot, lt. 'bishops' meeting'.

(1) A standard example of a loan-translation is god-spell (in ModE obscured to gospel); it is usually explained as a loan-translation of Lat. euangelium (which in its turn is a loan-word from Greek euangelion 'good news'); cf, e.g., Gneuss 1955: 94, 87.

(2) Gebed-hus 'prayer-house' is also a likely loan-translation, because it corresponds morphologically to Lat. domus orationis 'house of prayer' (if one ignores the different sequence of the elements and the fact that domus orationis is a genitival phrase, whereas gebed-hus is a cpd).

(3) Bisceopa gemot (for Lat. synodum) could be a loan-creation: it has no morphological correspondence to synodus, but it is a Christian term which the Anglo-Saxons probably did not have before their conversion to Christianity.

(4) With the other formations a decision is more difficult, one of the reasons being that they are not specifically Christian terms: lic-tun 'cemetery': cemeteries existed before the conversion, but since we have no written documents from before the conversion, we do not know whether the Anglo-Saxons used the word lic-tun before Christianization. His ymbstandendan ?those standing around him? (ThCapA 7) is, of course, not a Christian concept, but its close morphologic correspondence to the Latin qui ei circumstent makes it a possible candidate for a loan-translation. Hand-weorc for manuum operatio corresponds morphologically to the Latin phrase. Of course work with one?s hands existed before the conversion, but in the context of the ThCap it might nevertheless be a Christian concept, because (especially for priests and monks) it is opposed to more spiritual activities such as prayer, attending mass, reading Christian books.

3.6.Names

Names are a special group within the vocabulary (see, e.g., Cecily Clark 1992; Khallieva Boiche 2015). Germanic and OE names were often cpds, and, at least originally, they had a meaning, usually a positive one; how far the original meaning was preserved in later formations is another question. Five examples occur in Deor, namely: Beado-hild, Eorman-ric, ?eod-ric, M??-hild, Ni?-had; how far these refer to historic or to fictitious persons (or to historic persons who were later fictionalized) is also a question which I cannot enter into here (cf. the Glossary in Malone?s edition of Deor). There has been an extensive discussion about most of these names, and in some cases the original form and meaning might have been changed later through sound-changes or through popular etymology. Two are names of women (Beado-hild, M??-hild), and they have the same second element, hild ?war, battle?. For us it is probably strange that women were called ?war?, but the Germanic people obviously thought that ?war? or cpds with ?war? were good names for women.

(1) Beado-hild lit. ?battle-battle?, or ?battle-war? is probably also tautologic; it can be traced back to the Germanic ·Baudi-hildi (cf. Khallieva Boich? 2015: 205). M??-hild is a bit more difficult: Malone identifies the first element with ?might? (G. Macht; OE meaht, miht). Another possible explanation of the first element is OE m?? ?measure, honour, virtue, ability?, which would fit in with its latter meanings (?virtue-battle?, ?ability-battle?, i.e. perhaps ?someone [a woman] who is able to fight in a battle?; but m?g? ?maiden, girl? and m?g? ?family, tribe? also come to mind - perhaps several words had coalesced in Mae?hhild.

(2) Eorman-ric and ?eod-ric also have the same second element, which is apparently identical with r?ca ?ruler? (from r?ce ?strong, powerful?); the literal meaning of ?eod-ric is thus ?people-ruler? (where strictly speaking the first element is pleonastic, because a ruler usually rules over his people); Eormanric again is a bit more difficult. Its first element could be identical with OE eormen (yrmen, iermen ) ?mighty, vast?; the meaning of Eormanric would then be ?mighty ruler?.

(3) Ni?-had apparently consists of ni? ?enmity, hatred, war? and h?d ?rank, person?, i.e. the literal meaning apparently was ?hate-person, hateful-person, person who is full of hate?. Ni?-had thus is one of the relatively few names with a negative meaning, but it would fit the negative characterization of Ni?-had in Deor.

4.Other types of compound nouns

4.1. Noun in the genitive + noun

This is not an old or original type of compounding: the combination of noun in the genitive + noun existed (and still exists) as a syntactic group and as such is attested several times in the texts discussed here; to give just a few examples of syntactic genitival groups: sinces brytta 'treasure's giver' (W 25); dryhtnes dreamas 'the lord's [i.e. God's] joys' (S 65); hwæ les eþel 'whale's land' i.e. 'the sea' (S 60); earfoða dæ l 'amount of hardships', lit. 'hardships' amount' (D 30); bisceopa gemot (ThCapA 4.1.) 'bishops' meeting' for Lat. synod (see above); heofona rice 'heavens' kingdom, kingdom of heaven' (ThCapA 1.26; frequent). In the latter case a compound (OE heofonrice) and a syntactic group (OE heofona rice) apparently co-existed in OE, but the form heofona rice as attested in ThCap clearly represents the syntactic genitival group.

The only genitival combination that coalesced into a syntactic group is domes-dæ g 'judgement's day' (> ModE doomsday); but whether the translator of the ThCapA regarded it as a cpd is difficult to say. Grein and ClHall, for example, list the cpd. döm-dæ g, but not a cpd. · domes-dæ g. Semantically, both the cpd and the syntactic group belong to the category which expresses time ('judgement [doom] takes place on that day'; see 3.4.1, Il (3) above).

4.2. Adjective + noun

Similar to genitival groups, combinations of adj + noun occured (and still occur) as syntactic groups, and occasionally as cpds. When used as the first part of a cpd, the adjective is not inflected. But in the texts analysed here, there are very few adj + noun combinations which can be regarded with any confidence as cpds: W has just one clear example, namely freo-mæ g 'free kinsman' (see also 3.3. (3) above).

The two examples from S are ambiguous, namely earfoð-hwil, soð-gied: earfoð(e) could be used as a noun 'hardship, trouble, etc.' or as an adjective 'hard, difficult', and even as an adverb 'with difficulty'; soð was also used as a noun 'truth, justice, etc.' and as an adjective 'true, just, etc.'. Earfoð-hwil could therefore be interpreted as a noun + noun cpd 'hardship-time', i.e. 'time of hardship', or as an adj + noun cpd, 'hard time'; soð-gied could be interpreted as a noun + noun cpd. 'truth-song' ('song that tells the truth') or as 'true song'. Here I have classified them as adj + noun cpds; if they were interpreted as noun + noun cpds, that group would, of course, still be larger.

Even so, the group of adj + noun cpds is small, comprising five formations, if all the ambiguous formations are included; see also the following examples. Middan-geard 'earth', lit. 'middle earth' (i.e. between heaven and hell) is also unusual, because the first element (probably from the adjective midd 'mid, middle') is inflected; the -an probably represents the accusative singular (could also be genitive or dative singular) of the weak adjectival inflexion (which is the same for nouns and adjectives; cf. also 3.3.(3) above). The example from ThCapA, ungemethleahtor, is also ambiguous; ungemet was mostly used as a noun 'excess, immensity', and only rarely (in the form ungemete) as an adjective 'huge' (more often as an adverb 'excessively, extremely').

Semantically, adj + noun cpds normally have a copulative structure (cf. 3.4.1, I above) and can be schematically explained as 'B is A', e.g. earfoð-hwil 'the time is hard'; soð-gied 'the song is true'; middan-geard 'the earth is middle'; ungemethleahtor 'the laughter is excessive'.

4.3. Numeral + noun

Compounds with a numeral are also rare; there are three examples in our texts, namely an-haga and an-hoga from W, and an-floga from S. All three have an 'one' as their first element, and they are probably slightly lexicalized, because they express not only the numeral 'one', but also 'solitarity' or 'loneliness'. The meaning of these formations is relatively clear, namely 'solitary man, one who is alone' for both anhaga and anhoga, referring to the protagonist of W, and an-floga 'lonely flier' for a seabird. There has been a debate, however, whether anhaga and anhoga are variants of the same word or different words (see, e.g., Dunning and Bliss 1969: 37-40) - they are certainly synonyms, i.e. they have a similar or perhaps even the same meaning, but etymologically they are probably different words. The second element of an-haga is perhaps derived from haga 'hedge, enclosure', and the second element of an-hoga is perhaps derived from hogian 'to think' (i.e. lit. 'one who thinks alone'); the second element of an-floga is derived from the strong vb (II) flēogan - fleah - flugon - flogen, i.e. -haga is denominal, whereas -hoga and -floga are deverbal. The structure of an-haga is even more complex, because (originally) it does not refer to a hedge, but to someone who lives alone in a hedge. Similar formations are OE hæ gtesse 'witch' (related to G Hexe 'witch'), originally 'someone (a female) who lives in a hedge, or rides on a fence', and German Hagestolz 'bachelor' (originally also 'someone who lives alone in a hedge' (but some details of the formation of Hagestolz and Hexe are also still unclear; cf. eWdS).

4.4. Adverb + noun

This is also a rare combination, with just one example in the texts analysed here. The type is probably unusual, because adverbs normally modify verbs or adjectives or other adverbs, but not nouns. S nevertheless once has the hapax legomenon iu-wine, lit. 'formerly friend, earlier friend', which is generally regarded as a cpd, and often taken to have the literal meaning 'departed friend, dead friend', probably specifically referring to the dead lord.

5.Compound adjectives

5.1. Noun + adjective

Among the cpd adjectives, the most frequent group is formed on the pattern noun + adj. W, S, D together have 16 examples, and ThCapA (first chs.) has two more.

(1)Examples

(a)W has seven instances, namely 1) feoh-gifre 'avaricious', lit. 'moneygreedy'; 2) dom-georn 'eager for glory, ambitious', lit. 'glory-eager'; 3) hrim-ceald 'ice-cold'; 4) mod-cearig 'troubled in heart', 'troubled in thought', lit. 'mindsorrowful'; 5) sele-dreorig 'sad at the loss of the hall', lit. 'hall-sad'; 6) wæ l-gifru 'greedy for slaughter', lit. 'slaughter-greedy'; 7) winter-cearig, lit. 'wintersorrowful'.

(b) S has six examples: 1) est-eadig 'prosperous', lit. 'pleasure-prosperous, etc.; 2) is-ceald, is-cald 'icy cold', lit. 'ice-cold'; 3) mere-werig 'sea-weary'; 4) mod-wlonc 'haughty', lit. 'mind-proud'; 5) win-gal 'intoxicated with wine', lit. 'wine-proud', etc.; I have also put 6) el-þeodig 'foreign, strange' into this group (see further below).

(c) D has three relevant formations, namely 1) leoð-cræ ftig 'skilled in song', lit. 'song-skilled'; 2) sorg-cearig 'sorrowful'; 3) winter-ceald 'severe', lit. 'wintercold, as cold as winter'.

(d) The first few chapters of ThCapA have two examples, 1) lic-wyrðe 'pleasing, acceptable', lit. 'like-worthy', i.e. 'worthy to be liked'; 2) riht-wis 'righteous' (see below).

(2) Morphology

The typical morphologic shape is noun + adj, which is exhibited by, e.g., dom-georn, hrim-ceald, win-gal. But relatively many noun + adj cpds have derived second elements; four are derived adjectives with the suffix -ig (mostly derived from nouns): cearig 'sorrowful' (from cearu; ClH s.v. carig), dreorig 'sad, sorrowful' (from dreor blood'), eadig 'happy, prosperous' (from ead 'riches, prosperity'), werig 'weary' (ClHall refers to wör as the basis). The first element of lic-wyrðe is probably the stem of the verb lician 'to like, to please', and not the noun lic 'body' (see above), nor the adjective gelic 'like, similar'.

Riht-wis (> 'righteous') is problematic (cf. Sauer 1992): from its original meaning it could be interpreted as a noun + adj cpd 'wise with respect to what is right', or more probably, as a bahuvrihi adjective (formally adj +noun) 'having the right manner [wis] '.

A special case is also el-þeodig 'foreign, strange' in S: it is probably not a cpd, but is derived from el-þeod 'strange, foreign people'; moreover el- does not occur independently and, strictly speaking, has to be regarded as a prefix.

(3) Semantic structure

Some noun + adj. cpds imply a comparison; both is-ceald and hrim-ceald can be paraphrased with 'as cold as ice' (or perhaps 'cold from ice'?), probably also with an intensifying function 'very cold'. The majority of the cpd adjectives from W and S can, however, be explained using a preposition: FOR: feoh-gifre 'greedy for money'; dom-georn 'eager for glory'; wæ l-gifru 'greedy for slaughter'; IN: mod-cearig 'troubled, sorrowful in the mind, heart'; mod-wlonc 'proud in the mind'; FROM: mere-werig 'weary from the sea' (i.e. from sailing on the sea); wingal 'proud from wine' (i.e. 'proud from drinking wine'). Probably the last two should be regarded as (slightly) lexicalized. Sele-dreorig 'hall-sad' can also be regarded as lexicalized; it does not just mean 'sad at the hall or because of the hall'; it is usually taken to have a more specific meaning 'sad at [the loss of] the hall'. Lic-wyrðe has a passive meaning, 'worthy to be liked'; the problematic interpretation of riht-wis has been discussed above.

5.2.Adjective + adjective

There is just one example, which, however, occurs in W and S, namely earm-cearig 'care-worn, wretched and troubled', lit. 'poor-sorrowful' (ClH s.v. carig).

Morphologically, cearig is a derived adjective (from caru, cearu 'CARE, sorrow'); semantically earm-cearig is probably an additive formation (a copula cpd) 'poor and sorrowful' ('A + B'; see 3.4.1.Д above). Cearig as a second element occurs also in two noun + adj cpds., see 5.1 above.

5.3. Bahuvrihi adjectives

(1) Description of the type: Bahuvrihi adjectives form a special category, because formally they consist of adj +noun, i.e. the second element is a noun, but they function as adjectives. Semantically, they express a 'have' relation, i.e. 'X has AB'; they mostly refer to parts of the body or the mind. Later they died out and were replaced by extended bahuvrihi adjectives (see 5.4. below); the reason was perhaps that extended bahuvrihis adjectives are formally marked as adjectives, whereas bahuvrihi adjectives are not formally marked as adjectives.

(2) Examples: W has three, and S has four (ThCapA has none in chs. 1-11), i.e. there are seven in our material.

(a) W: collen-ferð 'ready for action, resolute, proud', lit. 'with a proud (?) mind'; dreorig-hleor 'with tear-stained cheeks, sad-faced', lit. '[having a] sorrowful face'; hat-heort 'angry, impulsive', lit. '[having a] hot heart' (heart as the seat of emotions).

(b) S: eaþ-mod 'humble-minded, friendly', lit. '[having an] easy mind'; gomel-feax 'grey-haired' (first element: gamol 'old, hoary'); isig-feðera 'with icy, frosted wings' hap.leg.; urig-feðera 'dewy-winged' - the latter two are used as nouns.

(3) Morphology: Typically, bahuvrihi adjectives consist of adj + noun, with the noun functioning as an adjective, e.g. eaþ-mod, hat-heort. But collen- and urigare apparently not attested independently in OE and are therefore difficult to explain. Urig is usually taken to mean 'damp'. Collen- in W perhaps means something like 'proud'; the literal meaning of collen-ferð would then be 'having a proud mind'. On collen- and urig see AEW. A form similar to collen- is collon- in the plant-name collon-croh 'water-lily (?), but this is also not attested independently and therefore does not help to explain collen-ferð (cf. Sauer and Kubaschewski 2018).

Three first elements are apparently derived with the suffix -ig: dreorig (from dreor 'blood'), isig (from īs 'ice'), and urig. Isig-feðera and urig- feðera are substantivized according to the weak declension, i.e. isig-feðera 'the one having icy wings'; urig-feðera 'the one having dewy wings', both referring to sea-birds.

(4) Semantics: As stated above, the meaning usually is 'having AB', e.g. dreorig-hleor 'having a sad face', and in substantivized bahuvrihis, 'X has AB'.

5.4. Extended bahuvrihi adjectives

Probably in order to show their adjectival character more clearly, bahuvrihi adjectives were even in OE sometimes extended with a suffix, in OE often -ig or -ed, later usually -ed. Not every extended bahuvrihi adjective is, however, actually an extension of an existing bahuvrihi adjective; many are new formations without a preceding bahuvrihi adjetive.

In our material, there is just one example of an extended bahuvrihi adjective, namely the hapax legomenon fea-sceaftig 'miserable, poor', which is an extension of the also attested fea-sceaft.

5.5.Numeral + adjective

This is also a rare pattem with just two examples in our material. As with the pattem num + noun, in both cases of num + adj the first element is a form of ān 'one': an-hydig 'resolute, strong-minded', lit. 'having one mind' (D); æ n-lipie 'single, solitary' (ThCapA; ClHall s.v. an-liepig); i.e. in æ n-lipie the numeral (first element) expresses once more 'solitariness', but in an-hydig it expresses strength.

6.Prefixoids and suffixoids

Prefixoids and suffixoids are elements that stand somewhere between words and prefixes or suffixes. A few, such as cyne- do not occur independently, but they can be regarded as morphologically conditioned allomorphs of words, in this case of cyning 'king'. Others, such as -dom or -had, were originally words, but either they split up into a word and a suffix (in the case of dom) or died out as independent words and became suffixes (in the case of had > ModE -hood), but both processes probably took a long time to complete, and when they began is difficult to say.

6.1. Prefixoids

(1) cyne- can be regarded as a morphologically conditioned allomorph of cyning (cf. Sauer 1992): whereas cyning is used as an independent word, cyne- is used as the first element of cpds. There is one attestation in D, namely cynerice 'kingdom'. Later cyne- died out and was generally replaced by king (e.g. ModE kingdom).

6.2. Suffixoids

6.2.1. Noun suffixoids

(1) -dom: The OE word döm had a wide range of meanings, e.g. 'judgement, law, advice, choice, etc.'. Gradually it split up into the ModE noun doom /du:m/ and the ModE suffix -dom /do m/. But when this split occurred is difficult to tell. ClHall assumes that even in OE there existed a noun döm and a suffix -döm, but the spelling was the same, and the meaning 'condition, etc.', which ClHall gives for the suffix, was also a meaning of the independent word. In our material, there are two formations with -dom, both in the ThCapA, namely recenddom 'governance', and þeowdom 'servitude, slavery'. Paraphrasing them is not easy; perhaps 'state of being a ruler, governor' for recenddom and 'state of being a slave' for þeowdom.

(2) -had: 'person, condition, state' was originally also an independent word. It died out in independent use in Early Middle English, but has survived as a suffix, ModE -hood. As with -dom, the question is when -had started to become a suffix. ClHall assumes that even in OE there were a noun hād and a suffix -hād, but the meaning 'condition, state', which ClHall gives for the suffix, was also a meaning of the noun. Therefore it is difficult to distinguish between hād noun and -hād suffix in Old English; accordingly I have put combinations with -had among the formations with suffixoids.

(3)-wara 'inhabitants' did not occur as an independent word in OE, but only as a second element of combinations. There is one attestation of burgwara 'inhabitants of a 'burg' (W), citizens, burghers'. It seems to have been synonymous with compounds where the second element is also attested as an independent word, namely burg-leod, burh-mann, both also meaning 'citizen'; therefore I have classified it here as a suffixoid.

6.2.2.Adjective suffixoids

(1) -bæ re 'bringing' is derived from the strong verb IV beran - bæ r - bæ ron, but does not occur independently as an adjectival element (see also Sauer and Waxenberger 2017: 176, ?10). It occurs in wæ stmbæ re 'fruitful', lit. 'bringing fruit' in ThCapA.

(2) -leas 'without, free from' was still used as an independent word in Old English, e.g. W 86 burgwara breahtma lease 'without the noise of the citizens', but later it became a suffix (-less) e.g. ModE friendless, bottomless etc. ClHall lists it just as a word and not as a suffix, but in view of its development, I have classified it as a suffixoid here. There are four examples in our texts, from W freond-leas and wine-leas, both meaning 'friendless, without a friend', and from D ende-leas 'endless' and grund-lease 'numberless', lit. 'groundless'. Some formations with -less have an antonymous formation in -ful (e.g. ModE helpful - helpless), but not, e.g., the formations just quoted from W and D.

7.Conclusion

The analysis of compounds in some of the OE Elegies (Wanderer, Seafarer, Deor) and in chapters 1-11 of the OE Theodulf Capitula A has yielded a number of interesting results; here I can sketch just a few of them. Some principles and problems of word-formation in general and of compounding in particular are common to poetic texts and prose texts, but the two groups also have their specific features. In poetic compounds, the principle of variation plays a role, whereas with some of the compounds attested in prose texts the question arises whether they were coined as loan-formations, based on a Latin model. In the poetic texts, moreover, many compounds are hapax legomena or confined to poetic language; nevertheless they show the productivity of compounding in Old English.

As fas as morphology is concerned, compound nouns of the type noun + noun are by far the most frequent pattern (with cca. 75 different formations), followed by compound adjectives of the type noun + adjective (with 16 different formations). All other types of compound nouns (noun in the genitive + noun; adj + noun; numeral + noun; adverb + noun) and compound adjectives attested in our material (adj + adj; bahuvrihi adjs; extended bahuvrihi adjs; numeral + adj) are much rarer. For a number of reasons, it is, however, impossible to give accurate numbers for many types. For example, in several formations, the word-class of one element (often the first element) is uncertain (e.g. noun or adjective; primary noun or derived noun - cases of the latter type can often be explained etymologically, but whether this had synchronic relevance is another question). I have classified some elements as prefixoids or suffixoids, i.e. they stand somewhere in-between prefixes or suffixes and. In the case of elements that are not primary, but the result of derivation, a cross-classification might be useful.

Especially with noun + noun compounds, there are also many semantic subgroups, e.g. copula compounds and rectional compounds, and among the latter, compounds referring to people, place, time, material, instrument, action, etc. Once more, some formations are fairly straightforward, whereas others are ambiguous and difficult to classify. Moreover, when dealing with semantics, a distinction should be made between compounds that can be understood and explained with the help of their elements, and compounds that have additional semantic features, i.e. that are lexicalized or idiomatized. Especially in poetry, one phenomenon does not exclude the other, however, and in addition to the meaning of the elements and features of lexicalization, the reference of compounds should also be taken into account. Gold-wine (a noun + noun compound), for example, has the literal meaning 'gold-friend', but it refers to the lord who gives gold to his warriors and is not only their lord, but also their friend (at least this is the image presented in the OE elegies - how far this also mirrored social reality is a different and difficult question). It can, of course, be debated whether this is a semantic phenomenon (a feature of meaning) or a matter of reference. Compounds of the type gold-wine were probably formed due to the poetic principle of variation.

Sidebar

Hans Sauer is Emeritus Professor of English (with an emphasis on the history of English and Medieval English literature) at the University of Munich (LMU), Germany; currently he teaches also at Vistula University in Warsaw. He has taught as a visiting professor at many universities and has given papers at many international conferences. He has published on Medieval English texts as well as on word-formation, glossaries and lexicography, interjections and binomials, and he has edited or co-edited several volumes of collected scholarly articles.

E-mail address: [email protected]

Footnote

1 Abbreviations used in the text: adj. = adjective; cpd. = compound; lit. = literally; sv = sub verbo / under the word; ThCapA = Theodulfi Capitula, Old English A version; wf = word formation

References

References

CHEL I = The Cambridge History of the English Language. Vol. 1: The Beginnings to 1066. Hogg, Richard M. (ed.). Cambridge: Cambridge University Press, 1992.

Clark, Cecily. 1992. "Onomastics" in Hogg, Richard M. (ed.). The Cambridge History of the English Languag. Vol. 1: The Beginnings to 1066. Cambridge: Cambridge University Press, pp. 452-489.

Davis-Secord, Jonathan. 2016. Joinings: Compound Words in Old English Literature. Toronto Anglo-Saxon Series 20. Toronto: University of Toronto Press.

Faiss, Klaus. 1978. Verdunkelte Compounds im Englischen. Tübinger Beiträge zur Linguistik 104. Tübingen: Gunter Narr.

Gneuss, Helmut. 1955. Lehnbildungen und Lehnbedeutungen im Altenglischen. Berlin: Erich Schmidt.

Gneuss, Helmut and Michael Lapidge. 2014. Anglo-Saxon Manuscripts: a Bibliographical Handlist of Manuscripts and Manuscript Fragments Written or Owned in England up to 1100. Toronto: University of Toronto Press.

Godden, Malcolm R. 1992. "Literary Language" in CHEL I, pp. 490-535.

Götz, Dieter. 1971. Studien zu den verdunkelten Komposita im Englischen. Nürnberg: Hans Carl.

Hofstetter, Walter. 1987. Winchester und der spätaltenglische Sprachgebrauch: Untersuchungen zur geographischen und zeitlichen Verbreitung altenglischer Synonyme. Munich: Wilhelm Fink.

Hubner, Julia. 2017. "We will Compound this Quarrel: Marlowe, Shakespeare, Jonson and their Use of Compounds". Doctoral dissertation, Ludwig-Maximilians-Universität Munich (LMU).

Käsmann, Hans. 1961. Studien zum kirchlichen Wortschatz des Mittelenglischen 11001350. Buchreihe der Anglia 9. Tübingen: Niemeyer.

Kastovsky, Dieter. 1992. "Semantics and Vocabulary" in CHEL I, pp. 290-408.

Ker, Neil R. 1957. Catalogue of Manuscripts Containing Anglo-Saxon. Oxford: Oxford University Press.

Khallieva Boiché, Olga. 2015. Imja et Name: Aux sources de l'anthroponymie germanique, anglo-saxonne et slave. Paris: PUPS / Presses de l'université Paris-Sorbonne.

Koziol, Herbert. 1972. Handbuch der englischen Wortbildungslehre. 2nd edition. Heidelberg: Winter.

Marchand, Hans. 1969. The Categories and Types of Present-Day English WordFormation. 2nd edition. Munich: C.H. Beck.

Marquardt, Herta. 1938. Die altenglischen Kenningar. Ein Beitrag zur Stilkunde altgermanischer Dichtung. Halle: Max Niemeyer.

Sauer, Hans. 1992. Nominalkomposita im Frühmittelenglischen. Tübingen: Niemeyer.

Sauer, Hans and Gaby Waxenberger. 2017. "Old English Dialects" in Brinton, Laurel J. and Alexander Bergs (eds). The History of English, vol. 2: Old English. 2nd edition. Berlin: de Gruyter, pp. 160-186.

Sauer, Hans and Elisabeth Kubaschewski. 2018. Planting the Seeds of Knowledge: an Inventory of Old English Plant Names. English & Beyond 8. Munich: Utz.

Dictionaries

AEW = Ferdinand Holthausen. Altenglisches etymologisches Wörterbuch. 3rd edition. Heidelberg. Winter, 1974 [the 3rded. is basically a reprint of the 2nd ed. 1963].

BT = Bosworth & Toller: An Anglo-Saxon Dictionary. Bosworth, Joseph and T. Northcote Toller (eds.). 1898; BTS = An Anglo-Saxon Dictionary: Supplement, by T. Northcote Toller; both published at Oxford: Oxford University Press.

ClHall = John R. Clark Hall. A Concise Anglo-Saxon Dictionary. 4th ed. Cambridge: Cambridge University Press,1960 [many reprints].

DOE = Dictionary of Old English. Cameron, Angus, Ashley Crandell Amos, Antonette diPaolo Healey, Roy Liuzza, and Haruko Momma (eds.).Toronto: University of Toronto [at present available A-H].

EWDS = Friedrich Kluge. Etymologisches Wörterbuch der deutschen Sprache, 22nd edition by Elmar Seebold. Berlin: de Gruyter, 1989.

Grein = Christian Wilhelm M. Grein. Sprachschatz der angelsächsischen Dichter. 2nd edition. rev. by Ferdinand Holthausen and Johann J. Köhler (1912). repr. Heidelberg, Winter 1974.

Editions

ASPR III = The Anglo-Saxon Poetic Records III: The Exeter Book. George Philip Krapp, Elliott van Kirk Dobbie (eds.). New York: Columbia University Press, 1936 [contains, among others, also W, S, D].

D = Deor. Malone, Kemp (ed.). London: Methuen, 1933 [several reprints].

S = (a) The Seafarer. Gordon, Ida L. (ed.). London: Methuen, 1960 [several reprints]; (b) Il Seafarer: La navigatio Cristiana di un poeta anglosassone. Cucina, Carla (ed.). Roma: Edizioni Kappa, 2008.

ThCap = Theodulfi Capitula in England. Sauer, Hans (ed.). TUEPh 8. Munich: Fink, 1978.

W = The Wanderer. Dunning, Thomas P. and Alan J. Bliss (eds.). London: Methuen, 1969 [several reprints].

W = The Wanderer. Roy F. Leslie (ed.). Manchester: Manchester University Press, 1966 [several reprints].