The word dæl in Ælfric's grammar

András Cser

All dictionaries and glossaries of Old English list the word dæl and give ‘part’ as its only or primary meaning. Bosworth and Toller (1898 and then several editions), however, as well as Hall (1894 and then several editions, incorporating material from Bosworth and Toller) also indicate ‘word’ as a meaning of dæl. In this short paper I look at the origin of this surprising gloss and trace it back to a misinterpretation of an Ælfric-locus, which, in turn, exemplifies a certain terminological confusion in Ælfric’s original text.

1 Ælfric’s grammar and its background

Ælfric’s Excerptiones de arte grammatica anglice is an elementary grammar of Latin written in Old English in the 990’s (for a useful description and discussion of the work see Law 1997:200‒223, on which the short summary in this section is also based). After the Irish Auraicept it is the first European grammatical work ever to be written in a vernacular. It soon became very popular and was the dominant grammar in England until about the middle of the 12th century. It exists in many copies; there are even copies with French glosses, and a Middle English adaptation was composed in the 13th century. After that, however, it fell into oblivion, and when it was unearthed in the 16th century, it was no longer seen as useful — it is interesting to note that Robert Talbot used the Latin examples to make sense of the original Old English, in a way reversing the original purpose of the grammar (cf Buckalew 1982).

The background against which the grammar must be understood is the early Middle Ages. In this period the primary concern of education was the teaching of Latin. The material inherited from late Antiquity had been intended for native speakers of Latin and was thus not immediately suitable for use in lands where other languages were spoken by the population. Consequently the material — up to the 8th century mainly Donatus, from the 8th century on both Donatus and Priscian — was tampered with, truncated, changed, augmented, and new grammatical works were often patched together from several sources. Commentaries on them were also written and became part of the grammatical corpus transmitted throughout the Insular and the Carolingian periods.

From the 7th century on several works were written by Anglo-Saxon scholars (Aldhelm, Bede, Alcuin, Boniface, Tatwine), but only in and on Latin, not in or on English. Language seems to have played a central role in Anglo-Saxon intellectual life; note the following passage quoted from Lendinara (2001: 279) to underline this point:

Anglo-Saxon learning presents a curious paradox… In some respects, Anglo-Saxons were in the vanguard of European learning — this is particularly true in the field of grammar and related disciplines — but in others were an undistinguished backwater. The most characteristic feature of their learning, however, is their fascination with linguistic detail, and this fascination is reflected in countless ways: in the use of runic and cryptographic alphabets in manuscripts, in the pursuit of obscure, hermeneutic vocabulary, in the use of etymology as a pedagogical device, in the pervasive fondness for riddles and riddling, to name only a few.

In the late 10th century many new manuscripts arrived into England from Northern France as part of the Benedictine reform of the period emanating from Cluny; among these were, no doubt, some of Ælfric’s sources. The Excerptiones itself was written in the 990’s, in the same years as his other pedagogical works (Glossary, Colloquy on the occupations). The goal of his grammar was to teach the rudiments of Latin with the help of the vernacular, a worthy goal in a period when the knowledge of Latin had been on the decline for some time. It was not based on original Roman grammars from Antiquity, but on a later abridgement of Priscian’s Institutiones grammaticae and some other compilations. Its main source (Excerptiones de Prisciano) remained unpublished until recently (Porter 2002).

2 Structure and terminology: the question of dæl

The general structure of the grammar follows that of the Excerptiones de Prisciano very closely. After a preface in Latin and an added preface in Old English, which also spells out the parts of grammar in an introductory manner, first the minimal units of language are described (sounds, letters, syllables, types of sounds). Then an extensive discussion of the parts of speech follows, traditionally the main bulk of any grammar since Antiquity. The parts of speech are defined, their properties listed, and numerous examples are given. Two appendices conclude the manuscript, both patterned on the Latin original. The first lists the names of numbers; the second gives a summary of the main points and terms of grammar in thirty short paragraphs.

The terms that Ælfric uses (see Williams 1958 for an extensive discussion) are mostly calques, with some periphrastic expressions. The names of the parts of speech exemplify both types:

The parts of speech in Ælfric’s grammar
Latin Ælfric
nomen nama
pronomen þæs naman speliend
verbum word
adverbium wordes gefera
participium dæl nimend
coniunctio geþeodnis / feging
praepositio foresetnis
interiectio betwuxaweorpennis

In some cases terms are borrowed rather than replaced, but this is somewhat rare. For example, after listing the parts of speech Ælfric writes: feower þæra dæla synd declinabilia þæt is decliniendlice ‘four of the parts [of speech] are declinable’, with decliniendlic as the translation of declinabilis.

An interesting and hitherto not much discussed problem emerges where Ælfric describes the hierarchy of linguistic units. At first sight it looks like the traditional hierarchy given in many such grammars in line with a tradition going back to Antiquity: letter ‒ syllable ‒ word ‒ sentence. On closer scrutiny, however, an odd terminological quirk can be found. Let us look at the passage in question:The original can be seen here:

Littera is stæf on englisc. and is se læsta dæl on bócum. and untodæledlic; we todælað ða bóc to cwydum. and syððan ða cwydas to dælum. eft ða dælas to stæfgefegum. and syððan ða stæfgefegu to stafum…

The translation of this passage (minus the first third, which I have supplied) in Bosworth and Toller, s. vv. cwide and dæl, is the following:

‘[littera is letter in English and is the smallest part of books and indivisible.] We divide the book into sentences, and then the sentences into words [parts], again the words into syllables, and then the syllables into letters…’

A slightly different translation is given in Hall (2009:202):

‘We divide the book into sentences, and then the sentences into parts [of speech], then the parts into syllables, and then the syllables into letters.’

As can be seen, the main difference between the two translations hinges on the word dæl: Bosworth and Toller translate it as ‘words [parts]‘, whereas Hall translates it as ‘parts [of speech]‘. In both cases the parenthetical additions hint at the problematic nature of the precise interpretation, but the next clause, where the same word is resumed, indicates what the primary meaning is taken to be (‘words’ and ‘parts’, respectively). Thus the full hierarchy appears to be this:

boc = book
cwyde = sentence
dæl = word/part (?)
stæfgefeg = syllable
stæf = letter

The sensible thing to do is to look at Ælfric’s Latin source, the Excerptiones de Prisciano, and see what it says at the parallel locus. We find the following (with David Porter’s translation added):

Dividimus codicem in sententias, sententias in partes, partes in sillabas, sillabas in litteras.
‘We divide a book into sentences, sentences into words, words into syllables and syllables into letters.’

It is thus clear that Ælfric gives a literal translation for pars where he writes dæl (and note that Porter translates pars as ‘word’). The full hierarchy shows a word-for-word correspondence, as can be seen:

codex = boc
sententia = cwyde
pars = dæl
sillaba = stæfgefeg
littera = stæf

As I said at the beginning of the paper, both the Bosworth‒Toller dictionary and Hall’s dictionary give ‘word’ as a gloss for Old English dæl; but, significantly, their only source reference is this particular locus in Ælfric’s grammar. It is thus important to decide whether ‘word’ is a correct translation for dæl here. I think the answer is no. Although dæl here clearly refers to a word-sized unit in the hierarchy of linguistic units, it is the literal translation of Latin pars, which can mean ‘part of speech’ (a category assigned only to words in grammar before the twentieth century), but strictly speaking it cannot mean ‘word’.The two dictionaries also list ‘part of speech’ as a meaning of dæl, and Bosworth and Toller also lists ‘part of sentence’. However, the notion of part of sentence as dictinct from word class did not exist in early Medieval grammar at all. Williams (1958) interprets Ælfric’s dæl as ‘part’ as a matter of course.

The problem that manifests itself here is ultimately Latin-based: in the Latin grammatical tradition the word for ‘word’ in general (verbum) was appropriated for ‘verb’, but a new lexeme was created to match the meaning ‘word’ (in many grammars this was dictio), thus the problem was solved. Ælfric, however, while translating verbum (ie ‘verb’) as word into Old English, did not bother to find or create a suitable replacement for the general sense ‘word’, and actually uses word also in this general sense occasionally ‒ just as he uses dæl too in the general ‘part’ sense at several places.

Thus in the verbatim rendering of his Latin source Ælfric slips into a mild confusion of categories in that he uses the same term for the hierarchical unit (= word as constituent) and for the type (= word class) ‒ not to mention the general sense (‘part’). This is not really surprising since the Latin term pars orationis is itself somewhat misleading ‒ it literally means ‘part of the sentence’, but in its use in grammars it clearly refers to categories that words fall into.

3 Conclusion

Two orthogonal aspects are mixed in the way Ælfric uses the word dæl as a grammatical term: a hierarchical aspect (word as constituent, the level between sentence and syllable) and a “horizontal” aspect (word as belonging to certain types or classes). In the practical work of teaching Latin, however, this is likely not to have caused many problems. The ambiguity goes back to the use of Latin pars in grammar (specifically pars orations ‘part of speech’), but in the Latin tradition it was resolved. The upshot of this brief discussion is that Old English dictionaries (such as Bosworth and Toller as well as Hall) are wrong in listing ‘word’ as a meaning of dæl.


Bosworth, Joseph and T. Northcote Toller. 1898 (and later editions). An Anglo-Saxon dictionary. Oxford: Clarendon.

Buckalew, Ronald E. 1982. Nowell, Lambarde and Leland: The significance of Laurence Nowell’s transcription of Ælfric’s Grammar and Glossary. In: Carl T. Berkhout and Milton McC. Gatch (eds) Anglo-Saxon Scholarship: The First Three Centuries. Boston MA: G. C. Hall. 19‒50.

Hall, Clark. 1894 (and later editions). A concise Anglo-Saxon dictionary. Cambridge: CUP.

Hall, Thomas N. 2009. Ælfric as pedagogue. In: Hugh Magennis and Mary Swan (eds) A Companion to Ælfric. Leiden/Boston: Brill. 193‒216.

Law, Vivien. 1997. Grammar and Grammarians in the Early Middle Ages. London: Longman.

Lendinara, Patrizia. 2001. The world of Anglo-Saxon learning. In: Malcolm Godden and Michael Lapidge (eds) The Cambridge Companion to Old English Literature. Cambridge: CUP. 264‒281.

Porter, David W., (ed. and trans.) 2002. Excerptiones de Prisciano: The Source for Ælfric’s Latin-Old English Grammar. Cambridge: Brewer.

Williams, Edna R. 1958. Ælfric’s grammatical terminology. Publications of the Modern Languages Association 73: 453–462.