SELIM03.pdf

7
Roger Lass, Selim 3 (1993): 7-25

INTERPRETING vs. DISAPPEARING:

ON TEXTS AS HISTORICAL OBJECTS1

1 POLEMICAL SETTING

Much of the factual material in this paper will be familiar to readers of

SELIM, though the implications I draw may be novel. My intention is to re -

open debate on certain issues generally taken as settled a long time ago. For

this reason (as well as because this is the kind of thing I seem to write) the

paper is both idiosyncratic and waspishly polemical. The rhetorical stance is

one of treating complex issues with a kind of pseudo-naive innocence, seeing

what would happen if one asked Daddy about the Emperor’s clothing. It

never hurts to look at even the most obvious things again, and remind our-

selves of what we once had to learn with difficulty and now accept as un-

proble matic.

As long as people have been seriously editing older texts, it has been

customary to ‘repair’ the often intractable messes presented by older MSS, to

make them useful, palatable, easily appreciated by readers, etc. Such tinkering

comes in both (rela tively) benign and ma lignant types, but the latter far

outnumber the former. To set the cat properly among the pigeons, I suggest

first that regardless of the qualifications of the editor involved, any emenda-

tion of older texts (or even worse, the filling of lacunae left by destroyed text)

1 This is an expanded and more technical version of material to appear in Lass
(Forthcoming: ch. 2), here aimed at a specifically Anglicist and medievalist audience.

is a destructive and ultimately indefensible procedure. And second, that

‘normalization’ of teaching texts as usually practiced is also harmful, and

even though it appears to serve pedagogical goals, serves the wrong ones.

This is so because of the peculiar kind of object an early ‘text’ is, and an im-

portant class of insights any such object fails to afford if it is treated in a cer-

tain way.

If the priggishness underlying these remarks were to embed itself in edi-

torial practice, it would make older texts much harder to read, and greatly com-

plicate the task of ‘literary appreciation’ for students, making the texts -as-

objects an even more alien life -form than they are already. This is pre cisely as

it should be; it’s only proper that the in terpretation of an Old English text, for

example, should in every way be more difficult, more tentative, more full of

gaps and puzzles, than that of a modern one.

This is not to suggest that we oughtn’t to have editions with rich textual

apparatus as well as explanatory and ‘intertextual’ notes; far from it. What I

would argue for is an enrichment of these guides to appreciation, and a radi-

cal impoverishment of the tampering actually done with texts as they appear

on the page. To take an extreme case, if a piece of text is garbage, then the

garbage ought to stay in the version we read, as an inseparable part of it, and

be (conjecturally) sorted out only in notes. What underlies this claim is a pair

of positions either pro foundly reactionary or radical, depending on your

point of view.

First, a text (or any other heritage) from an older his torical period is an

autonomous object, and our duty toward it is (a) to accept it as a contin gent

surviror from the past, with all its problems, and (b) to try and experience it as

closely as possible to the way it might have been experienced in its own time.

I’m not silly enough to think this can be done; but one can get a lot nearer

this ideal by confronting a clean text, with lots of background material--in its

proper place. At least we can do our best to look at the object as it was orig -

inally pres ented, in precisely the language it was written in.

My rationale is not dissimilar to that behind the movement for ‘authentic’

performances of early music (even as late as Beethoven or Berlioz). If a lis -

tener of the day could not have heard the sounds we make on modern instr-

uments, then it could be argued that we are not actually listening to the music

in ques tion if we hear it that way. (This view is sometimes dis missed by

critics as Klangfarbmaterialismus, but there is a lot to be said for it.) Are we

for instance really ‘hearing Mozart’ if we have string-playing with continuous

vibrato, valved in stead of natural horns and trumpets, no harpsichord con-

tinuo, pitch a half-tone too high, timpani with padded sticks instead of hard

ones? In such a case we are creating an acoustic object that the composer

could not have imagined nor his audience heard; the object is remanufactured

‘on the basis of’ the orig inal, but more a translation than a faithful copy. It

provides a very different experience from the kind it could originally have

induced.

So I suggest in the first instance that it might be a good idea to rid our-

selves of the romantic desire to ‘remake’ past art objects and other things in

our own image (to deprive ourselves of the undisciplined luxury of decon-

struction, and start respecting artifacts as artifacts); if all old things have to

be modernized, what’s the point of the past at all? At least from one perspec-

tive, the delight of the old is in its old ness, not what we ‘make of it’; our job is

to attempt, however ineptly, to remake our own consciousness on the model

of older ones, not to remake those on a new model. Without this attempt we

fail to amplify our experience, and confine ourselves to the familiar and self-

generated.

The second point is lin guistic, and more important to my mind than this

merely aesthetic one. My main concern here is the falsification of the linguis -

tic past produced by over-editing. The primary value of older texts is as

‘samp les’ or ‘representations’ of the languages they are written in; they lose

a certain vital historical import when seen through too many sets of

spectacles. From the lin guistic point of view, there is much to be said for

leaving texts, garbage and all, the way they have come down to us, and little

if anything to say for editorial tampering. I will illustrate this point (which has

a serious as well as a cranky dimension) with a familiar and particularly loath-

some example, and then draw some more general conclusions.

2 WHAT’s IN A WORD?

Let us look at some strategies for dealing with an intractable problem.

Older text corpora may contain lexical items that appear only once (hapax

legomena), and apparently have no descendants. Interpreting such forms

may be near impossible; they are often simplex and therefore compositionally

opaque (unlike say a unique compound both of whose elements are indepen-

dently attested); they may have no apparent near or distant cognates either.

To what extent are such items really ‘historical data’, and how should we treat

them? Should we give them SOME interpretation, no matter how conjectural?

Should we just leave them there flagged as cruxes? Or should we dis appear

them? (The reason I use this stage-magician’s verb will become clear.)

In particularly horrid cas es an item of this kind may have multiple wit -

nesses in a textual tradition. The one to be dis cussed here has four: is this

buried history or folie àquatre?

The Anglo -Saxon Chronicle for the year 937 contains a linguistic object

commonly taken to be a poem and usually called ‘The Battle of Brunanburh’.1

On this interpretation (see remarks in the next section), what are pre sented as

lines 10b-13a read as follows (Corpus Christi MS 173: text from Dobbie 1942:

17):

Hettend crungun

1 Or, to anticipate a kind of problem taken up immediately below, maybe it should be
Brunnanburh. The Corpus Christi MS 173 has brunanburh with a second <n> added
above the line after the <u> in a later hand, and Cotton Tiberius B. iv has brun nanburh
(Dobbie 1942: 16).

Sceotta leoda and scipflotan
fæge feollan, feld dænnede
secga swate …

which can be translated near-literally as

(the) enemies died,

Men of the Scots and Ship -floaters [= Vikings]

doomed, fell, (the) field ?-ed

(with) men’s blood …

The gloss ‘? -ed’ indicates that while the other lexical items in the text are

known from elsewhere in OE, dænnede is a hapax. So for that matter are the

readings of the other witnesses, MSS Cotton Tiberius A. vi, Cotton Tiberius

B. i (dennade), and Cotton Tiberius B. iv (dennode). Most student editio ns

read dennode, but Dobbie uses Corpus 173 as his copy-text and re tains the

reading (see below).

These three forms represent the same verb, apparently; at least nobody

has (for good parsimonious reasons, I suppose) suggested two unknown

forms, and the consensus is that these are variants of one item (the -ede/-

ade/-ode variation is more or less expectable.) Three of the witnesses agree

on the nuclear vowel (<e> as against <æ>), and a medial geminate <nn>.1 The

morphology is also transparent: whatever this/these verb(s) is/are, the forms

are all on syntactic grounds 3rd person singulars, and on morphological

grounds preterites belonging to the class II weak conjugation. This is clear

because a thematic weak preterite (-a-de, -o-de) ending after a heavy root

1 Corpus, aside from its idiosyncratic <æ>, seems to hesitate about the medial consonant.
At least the second <n> is added above the line, in another (apparently contemporary)
hand (Dobbie, 147). If this is, as seems likely, the work of a corrector, then whatever
this verb is it must have been known to somebody besides the first scribe, since there
was clearly a ‘wrong' version, and the ‘right' one had medial /-nn-/.

syllable must be class II.1 This class of verbs standardly has the infinitive

ending -ian (as opposed to normal class I -an); hence one could say that

dennode, dennade are preterite 3 singular word -forms of a lexeme whose dic -

tionary headword wo uld be the infinitive dennian (so Hulbert 1959, Clark Hall

1960, Holthausen 1968).

Since none of these forms occur anywhere except in versions of Brunan-

burh 12b, how can we find a gloss? There are different sorts of attempts to

deal with it. Hulbert (1959: 223) suggests dennode = ‘became slip pery’, pre-

sumably following the unsupported conjecture of Bosworth-Toller;

Holthausen 1968 s.v. dennian gives ‘fließen (?)’, with Skr dhanvati ‘it flows’

as a possible cognate. Clark Hall (1960) gives ‘to stre am (?)’, Pope (1966: 158)

has ‘dennian, wk.v.II. become wet, flow?’, but says properly ‘meaning

doubtful’ (the standard euphemism for ‘unknown’). This word (or ‘word’) ex-

emplifies a common problem. Is it a ‘real’ word, or is it a mistake, copied in -

accurately from some (perhaps corrupt) original, and then perpetuated, or

what?

All attempts at glossing (and even Holthausen’s stab at a possible etymo -

logical connection) are ultimately based on conjecture from context. After a

battle, what would a fie ld DO ‘with blood’? Well, given the characteris tic

rhetoric of Old Germanic battle poetry, it certainly might run or flow or stream

or become slippery. A non-word (in the sense that only a form exists, with no

extra-contextual backup for a meaning) is being interpreted simply by trying

to make sense of it; this is little more than low-level cleverness, and except for

Holthausen’s attempt involves no real argument, and is not intellectually

respectable.

1 In class I (the commonest), preterites are athematic after a heavy root (deµm -de ‘he
judged'), and thematic after a light (frem -e-de ‘he performed'). There is a small class of
verbs in /-r/ which belong to class I historically, but for reasons not relevant here have
thematic preterites after light roots (herian ‘praise', nerian ‘save', preterites her-e-de,
ner-e-de); they also, again irrelevantly, have an infinitive ending that looks like class
II.

Holthausen appears to apply this kind of strategy: if there’s no attestation

in the language in question, look for a putative cognate, i.e. a form whose

phonological shape could be reasonably related, and which has something

like what you think might be the right semantics. Apparently there is nothing

in attested Germanic, but there is one Sanskrit shape that sort of fits: den-

could = IE */dhan-/ or */dhen-/, since */e, a/ merge in Sanskrit. A PIE shape

*/dhen-/ would give PGmc */pen-/, which would give OE /den-/. The gemi -

nate is still to be accounted for (Holthausen does not deal with this); they are

rare in class II except as the result of assimilation (e.g. OE blissian ‘rejoice’, cf

blÈµÍe ’blithe’, and the noun bliss with its alternant bÈµps, and OS

blÈµ?sea).

So much for attempts at interpretation. They are of course all (more or

less) hopeless, if in Holthausen’s case with some sort of min imal support. But

scholars (rightly in principle, if often dreadfully wrongly in practice) abhor

vacuums as much as nature does, and sometimes do naughty things to fill

them. Since the late 19th century many have attempted to make something

out of this crux; and the bulk of them have not tried to find an etymology, but

rather assumed an error, and replaced the offending object with a familiar one

that is at least similar, and has a (more or less) solid meaning. That is, rather

than interpreting the (multiply attested) form, or abandoning it as hopeless,

they disappear it by emending the text, and tell us ‘what must have been

meant’. This is only one step less criminal than another common text -editor’s

ploy: ‘restoring’ lacunae resulting from destruction of text by inventing mate-

rial that ought to have been there. But it is of interest, since it dis plays the

same strategies as the other attempts, but with less justification.

Here is a sample of the suggested replacements for dennode and its

friends (the individual arguments are not really to the point here; for full dis -

cussion and references see Dobbie, 147):

(a) dengode, dengede ‘manured’ < dengian, a variant of dyngian ‘to

dung’ (cf. G düngen). The line then reads ‘the field was manured with the

blood of men’, which seems to me at least a most un-OE kind of conceit. The

MS readings are then based on an original error of substituting an <n> for a

<g>. But if so, why did the Corpus corrector bother to put in another <n>?

This suggests that indeed there was ‘such a word’, and that it had been mis -

spelled, and the only thing the corrector was unhappy about was the single

<n>. If the <æ> is original, the <e>-variant is Kentish, and the language of the

MSS is not; but there is evidence that at least two of them have connections

with Canterbury (Dobbie, xxxvf).

(b) deµanian ‘to steam’. The root (not to mention the verb) is itself unat-

tested, but could be connected with Gothic dauns ‘odor’. Verbs of this class

are often denominal (Lass 1993), so this is not out in principle. But the mean-

ing of the Gothic form is being stretched, and there are no traces of Gmc

*/Íaun-/ in Old English. This falls fo ul as well of the ‘correction’ of <n> to

<nn>, and the <nn> in the other witnesses.

like an Old Germanic metaphor than the manure above. (Though cf. Genesis

4.10, ‘uox sanguinis fratris tui clamat ad me de terra’, which the author would

surely have known.) The vocalism is manageable: at the time of the sources

(10th-11th c.) confusion of <a> and <æ> is not unlikely, and that of <e> and

<æ> possible but less so. And a writing of <e> for <y> would be just possible

on the grounds of the Kentish connection ((a) above), since /y(:)/ had gone

to /e(:)/ in Kentish, as had /æ(:)/.

(e) Íaµnode ‘became wet’. From an attested class II verb, and cf.

paµn ’moist’. This is palaeographically reasonable, since in most OE hands

the letters <d> and <Í> differ only in the latter having a cross-stroke. But we

still need a different vowel, and the geminate <nn>.

(f) dunnode ‘became dark’. Also attested class II. This, like dynnede,

would require only an initial error in the vowel; but in this case the Corpus

dænnede is much more difficult, because there is no likelihood whatever of a

‘Kenticizing’ of an orig inal /u/ <u> to <e> or <æ>.

Dobbie remarks, after a nice discussion, that ‘either dunnode or …

Íaµnode is defensible, but in view of other possibilities the text has not been

emended’. Good for him. At least one of the other possibilities, in my opin ion

the only correct one, is to leave the text as it is, and re frain from any sugges -

tion at all except one that would give the existing lemma an etymology, and

take that with a good deal of salt.

But why should one not be allowed to emend texts, and take the result of

the emendation as a second-order witness? Surely production of second-or-

der ((re)constructed) witnesses is a vital part of his tory -making (cf. Lass 1980:

ch. 2, Appendix, Lass 1986, Forthcoming, ch. 1). Isn’t this is in fact precisely

what we do when we reconstruct, on the basis of attested witnesses, a proto-

form, e.g. IE */dhan-/ or */dhen-/ on the basis of dennian and Skr dhanvati,

etc.?

One obvious motivation for not emending is that emendation (unlike re -

construction) is destructive; it obliterates part of the record, and substitutes

for it an invention of another time, place and culture. Rather than filling an

epistemic gap, as (properly constrained) linguis tic reconstruction or interpre -

tation of documents do, it falsifies the record, and produces corrupt second-

order witnesses through a pro cedure that is not principle -driven, but depen-

dent only on weak (usually flabby and aesthetic rather than methodologically

controlled) argument.

But the real motivation is largely a matter of constraints, i.e. limitations on

what kind of past objects we can be allowed to make; or to put it another way,

the deductive consequences of our abductive guesses. In a case like this,

without the power of ‘hard’ methods, we have very little procedural help to

tell us when we are producing nonsense, and almost nothing in the way of a

critical armoury for testing our guesses. We can talk to some extent about

likelihoods of particular kinds of scribal error, and suggest palaeographic

grounds for (putative) misreadings. But we do not have anywhere near as se-

cure an array of reconstructive methodologies and critical protocols for this

as we do for say phonological reconstruction, e.g. compara tive method, etc.1

(There are of course no methodologies for filling lacunae in texts!)

There is another difference too, and probably a more important one. In

reconstructing say a proto-IE segment, we are reconstructing a TYPE, not a

TOKEN , i.e. we operate on ‘system’ level (in the sense of Lyons 1977: ch. 1).

In trying to emend a text, we are dealing with a token, in the sense that the

text is (a representation of) a particular, unique utterance. There is no recon-

structive technique that can pro ject at utterance level; we may take texts as

utterances, and look for the kind of variation that is typical of certain kinds of

utterances in certain kinds of situations, i.e. treat a text as a ‘transcription’ in

some sense (cf. Lass 1992), subject it to variationist analysis, etc. But we still

come out with constructs at type level.

An attempt however to reconstruct an individual’s word -choice is as fatu-

ous as producing texts in a proto-language, or reading poems aloud in a re -

constructed pro nunciation with ‘expression’, supras egmentals and all. Not

that this stops people from doing it, but they shouldn’t. As Herbert Penzl

very properly puts it (1991: 62), ‘Our methods permit reconstruction of pho-

nemes, morphemes, phrases, syntactical rules, vocabulary forms, but not of

speakers and their language acts, such as e.g. the creation of written texts’.2

1 There are working principles for editing that often give good results, e.g. difficilor
lectio praeferenda est; but in this case the ‘more difficult' reading is in fact the unknown
original, so that on these grounds too it should stand.

2 This holds a fortiori for ‘texts' written in protolanguages (whether real ones like Indo -
European, or increasingly dubious ones like ‘Nostratic', ‘Amerind', ‘Proto-World', or
whatever), as well as for ‘reconstruct ion' of texts in their supposed original dialects,
when the witnesses come down to us in something else. As Penzl says of such efforts
(1991: 61), ‘their scholarly value is zero, even if there is no question of the learning
and competence' of their perpetrators.

3 AND WHAT, FOR THAT MATTER, IS A ‘TEXT’ ANYHOW?

The last section was basically about ‘words’. Despite the familiar theoreti-

cal problems of defining ‘word’ as a grammatical category, most of us have

fairly useful intuitions about what they are, good enough for practical work

anyhow. Etymological dictionaries, gloss aries, commentaries, etc. generally

seem to assume that whatever phonological level spelling might access (pho-

nemic, morphophonemic, on occasion allophonic), the morphological input is

not in ques tion: orthographies spell words. And if texts utilize such systems,

this implies in a way that texts are made of words too, since the main sources

of our words are texts. This however is oversimple; there are traditions that

do not always write in the ‘normal’ way, but take different-sized elements as

inputs.

Even modern English sometimes does this: ‘contractions’ like don’t,

weren’t are not really morphological syntagms, but pieces of syntax: not

‘words’ but constructions of auxiliary and negator. The lack of space and the

apostrophe are signs of phonological cliticization. If we take an ‘orthographic

word’ as a piece of an orthographic utterance surrounded by space, then

these are in fact words in the textual sense, and there is not necessarily a

direct ma pping (with re spect to boundaries) from morphology onto orthogra -

phy.

Old English does this kind of thing to a much greater extent; and this may

give us some evidence about supra segmentals, or at least about writers’ intu-

itions about suprasegmental constituents like tone-groups, etc. In many Old

English texts, for instance, light elements like prepositions, conjunctions and

determiners are often writ ten as if part of the following major category item.

The Beowulf manuscript (Zupitza 1959) has forms like the following (second

OE citation with conventional editorial word-divisions, first English gloss

literal, following ‘word’-divisions in the original):

(i) Îa com ofmore = Îa com of more (710)

then came frommoor ‘then came from the moor’

(ii) insele pam hean = in sele pam hean (713)

inhall the high ‘in the high hall’

(iii) com pa torecede = com pa to recede (720)

came then tobuilding ‘then he came to the building’

(iv) huseman scaÍa = hu se man-scaÍa (737)

howthewickedness injurer ‘how the wicked-injurer’

(v) Íawæs onuhtan = Ía wæs on uhtan (126)

thenwas inmorning ‘then there was in the morning’

The word -divisions in the second versions (those that appear in printed

editions, and to some extent even in Zupitza’s transcription of the MS) are

presumably based on independent occurrences of the forms with spaces

around them in other texts (as well as modern survivals and cognates). The

reading of man-scaÍa and mæl-ceare as compounds depends on prosodic ev-

idence; they alliterate on /m/, and according to the OE rules as we understand

them the first syllables should there fore bear primary stress.1

The scribe then takes what looks like potential phonological cliticization

and writes it; on the other hand he normally separates what ought to be first

1 Reading the sequence <man-> in man scaÍa as representing the lexeme maµn
‘wickedness' rather than man(n) ‘man' is in fact a secondary inference, based
apparently on metrical argument; but if man(n) indeed ends in a geminate, as it
probably does, then mann and mãn are met rically equivalent, and the compound could
mean ‘harmer of men'. Even what lexeme is being represented is not entirely clear.

elements of compounds from seconds, even affixes from stems (see below).

Perhaps even more distressing, there is virtually no punctuation, and what

there is, while usually coinciding with major syntactic boundaries, often does

not, but may signal line-ends or caesuras, or even, as far as we can tell, noth-

ing much in particular. One might add of course that even though the text is

in verse, and conventionally printed as such, it is written throughout as

prose, with little indication of line breaks. Here for instance are lines 99ff in

the standard edition ((Klaeber 1950):1

Swaµ Íaµ drihtguman dreµamum lifdon,

eµadiglÈµce, oÍ Íæt aµn ongan

fyrene frem(m)an feµond on helle;

wæs se grimma gst Grendel haµten,

mæµre mearcstapa, seµ pe moµras heµold,

fen ond fæsten …

[‘so the noble retainers lived in joys/blessedly, until one began/to commit

crimes, a fiend from hell;/the grim spirit was named Grendel,/famous march-

stepper, who held the moors,/ fen and fastness …’]

The manuscript reads:

SwaÍa driht guman dreamum lifdon

eadig lice oÍÍæt an ongan fyrene frem

1 Klaeber, like many editors, marks length as well, making the text even less like the
original, and falsifying the orthographic system. Length is never consistently marked in
OE, though it is sporadically indicated in some traditions by vowel-graph-doubling or
accents (see e.g. Krapp & Dobbie 1936: xxiv on the Exeter Book). There is however a
good reason for doing this at an early stage in the teaching of Old English, since
etymological identification is an essential part of what the student of the history of
English ought to learn, and the visual repetition of macrons is a useful way of fixing this
information.

man feond on helle was segrim ma gæst

grendel haten mære mearc stapa

se pe moras heold fen 7fæsten

Literally, more or less

Sothe noble men in -joys lived

blessed ly tothat one began crime t o-

do fiend in hell was the grim m spirit

grendel called famous march stepper

who that moors held fen &fastness

There are two matters of interest here. First, it would seem reasonable to

assume that the scribe was (generally) writing ‘by ear’ (since such conven-

tions are not universal in OE), and that therefore he perceived certain adverbs

and prepositions as clitics, and contrariwise certain (what we consider) affixes

as independent ‘words’. He also seems to have expected the reader to be able

to track the structure of the text by the alliteration, and to construe it without

the help of punctuation or sentence-division.

Second, this says something sobering (for those not familiar with the ac-

tual properties of certain historical sources) about the kind of artifact a mod-

ern ‘text’ is, with respect to what it purports to represent. This makes any

modern edition problematic: especially considering the kind of fairly loose,

rather paratactic syntax of this kind of Old English.1 The modern editor t hat is

has to make choices in a reading text which are frequently not the only

possible ones; there is a rich (and often not very edifying) literature concern -

ing alternative punctuations of texts from traditions of this kind. The only

thing that’s clear is that (on uniformitarian grounds: there are no languages

1 For an interesting treatment of some of the syntactic problems raised by unpointed or
minimally pointed OE texts, see Dunning & Bliss 1969: §§I,1, II.

that typically produce utterances made of continuous unbounded syntactic

strings) we can’t possibly assume that the syntactic structure of the text is

mirrored in the writing in the same way that the segmental-phonological and

(at times) the suprasegmental are. Texts like this then tell us things we are

very interested in knowing, and fail to tell us bigger things that we’d like to

know, and in many cases there’s no way of finding out.

Aside from decisions that essentially amount to parsing for the reader,

modern editions even make special claims about genre; the lineation as

‘verse’ is essentially a second-order construct (based to be sure on a long

tradition that is essentially well founded, if problematic in details); but the vi-

sual experience for the modern reader (no matter how well versed in the lan-

guage) is essentially different from that of the contemporary reader, since in

the editions even the page layout itself defines a genre. For Old English any-

how we don’t actually know what constituted a ‘poem’ as opposed to a piece

of highly alliterative prose, for instance; lineation of poetry to show what it is

comes later in the English tradition.1 As, one might add, does giving poems

titles (none of the poems in the Exeter Book or the Junius Manuscript, for in -

stance, have original titles).2 So ‘texts’, as we find them in editions that stu-

dents (and all too often scholars and historians of the language) read, are al-

ready very different kinds of objects from what was pres ented to their con-

temporary readers.

They may also fail, and this is much more serious, to represent anything

like the state of the language in which they are actually written. If as is often

the case, texts are ‘normalized’ (or as Penzl puts it ‘redialectalized’), this irons

out variation. And this may produce a kind of object that could not in fact

1 The distinction between verse and prose was certainly clear to Old English speakers
(though there appears to be no special word for the latter); the Exeter Book for
instance is described in a list of donations to Exeter Cathedral as being ‘on leoÍwisan
geworht', literally ‘in song-manner made' (Krapp & Dobbie 1936: ix).

2 Interestingly, two later hands have added titles to the Junius MS poem we now know as
Genesis, ‘Genesis in lingua Saxania', and ‘Genesis in Anglico' (Krapp 1931: xviii).

have been written by anybody in the tradition the text comes from. Let us re -

turn for instance to Brunanburh: this text is not written in the kind of lan-

guage one learns from the grammars (indeed practically no texts are, a point I

will return to in a moment). In any pedagogical grammar one might use for

begin ning students, it would become clear at an early stage that ‘the preterite

indicative plural ending’ is -on; this is what appears in the paradigms (though

other forms may be given in notes), and indeed this is what appears in the

‘normalized’ edition in Pope (1966). But this is not actually the case for any

real witness. In Dobbie’s copy-text for instance, there are 4 occurrences of -

on, as opposed to 7 of -un and 10 of -an. That is, the ‘standard’ ending

amounts to only 18% of the tokens in the text, and the one that is

(confusingly, but importantly) a homograph (and homophone?) of the infini-

tive ending and the bulk of the oblique forms of weak nouns, -an, constitutes

45%. Nearly half the tokens of this category are not what the student has

been taught should occur, and if it were not for the syntax could be taken as

infinitives or noun forms.1

But it is an important fact about the language of this text (or nearly any

other OE or other Old Germanic one) that the orthography is not codified, and

that it may tell us things about the language that are of interest and im-

portance for historical study. ‘Literary appreciation’ is not so primary that

texts should be presented in forms totally alien to the way they were origi-

nally experienced: and, much worse, in essentially ‘different languages’. (This

is surely the case of those in Pope 1966, which were the first ones I read; it

was only later that I discovered the much more interesting texts in the various

avatars of Sweet’s Reader, with their textual apparatus that was itself an

education in the nature o f variation in Old English.)

1 One might also note that one dative plural in this text (lafan 6b) also has -an instead
of expected -u m , and another (wundun, 43b) has -un. For a dative plural in -on one has
to turn to another witness (Íeodon, MS Cotton Tiberius B. iv, 23b); all of which helps
make the point.

This view has some pedagogical implications. Students must of course

start some where, with a few handles to get hold of. The standard grammars

(which are in themselves based on pseudo-codified versions of languages

which were not codified in that way by their speakers) are a useful, perhaps

necessary begin ning: provided variability and gaps in knowledge are made

clear at the outset, and students are warned that in the first texts they read

they will find ‘exceptions’ to the rules they’ve learned. So normalizing

grammars with (preferably rather late) West Saxon lemmata have their role to

play at the outset, and indeed knowledge of this tradition is an essential part

of the whole business of getting acculturated to Old English studies. But the

first real texts the student encounters should be pres ented unnormalized, or

at worst with original and edited texts (lightly normalised, with vowel-length

marked, and maybe even palatalization) on facing pages. (I know of no

student grammar that does this, but I could be wrong.) this way they come to

experience the language with all its warts from the beginning. and in fact get

some experience of what they are actually writ ten in.1

There are some good arguments, I think, even for getting students of Old

English and similar languages, at a quite early stage in their careers, to read

texts in MS or literal unpunctuated, uncapitalized transliteration.2 This will

1 Some normalizers are more horrendous than others. F.P. Magoun (1960) uses the
‘early West Saxon' forms in Holthausen for his lemmata, and includes them in his texts,
so that poems of Exeter Book age have <oe>, etc. This is a case of ‘in affecting
Holthausen he writ no language'. One learns to use Holthausen only after learning the
history of the language, and for very special purposes.

2 An exemplary presentation of texts (even for early learners) with minimal interference
is Bravo et al. (1992), which punctuates minimally as well, and dispenses with editorial
macrons in the texts themselves. (It may sound odd, but to my mind this textbook by
Spanish scholars comes much closer in the way it punctuates to what OE syntax really
feels like than most editions by English or German speakers.) I might add that at some
point in a serious student's education there ough t to be some contact with actual MSS
and some palaeographic training, even little excercises in ‘editing', so that they can get
some feel for what editors have actually done. It's very interesting to see what students
make out of a text when they have not only to turn it into modern letter-shapes, but

help disabuse them of the idea that the literary (or at least scrib al) culture of

the writers was as close to their own as tarted-up editions might suggest, or

even that the languages of the texts were the kind of codified standards that

most of them will be speakers of. Much of our ‘textual’ data (as the previous

section has suggested) is also second-order, not first-order, built on the

shakiest of foundations, and less trustworthy than the results of standard lin -

guistic reconstruction.

In short, emending, normalizing, and other kinds of over-editing falsify

and traduce under the guise of making accessible, and commit the worst kind

of anachronism while pretending to ‘represent’ past objects. Worse, such

pro cedures produce pseudo-data, and prejudice (or even disenable) our

reading of historical monuments. If we really hate Mozart played like Mozart,

and would rather hear it played like Tchaikowsky, why not listen to Tscha-

ikowsky, and stop pretending we’re hearing Mozart?

Roger Lass
University of Cape Town

REFERENCES

BRAVO, A., García, F., González, S. 1992. Old English Anthology. Oviedo:

Universidad de Oviedo.

CLARK HALL, J. R. 1960. A concise Anglo-Saxon dictionary. Cambridge:

Cambridge University Press.

DOBBIE, E.V.K. 1942. The Anglo-Saxon minor poems. New York: Columbia

University Press.

DUNNING, T. P. & Bliss, A. J. 1969. The Wanderer. London: Methuen.

gloss, translate, and punctuate. I can't think of a better way of get ting to grips with
what Old English syntax is really like.

HOLTHAUSEN, F. 1963. Altenglisches etymologisches Wörterbuch.2 Hei-

delberg: Winter.

HULBERT, J. R. 1959. Bright’s Anglo-Saxon reader, revised and enlarged.
New York: Holt.

KLAEBER, Fr. 1950. Beowulf and the Fight at Finnsburg .3 Boston: Heath.

KRAPP, G.P. 1931. The Junius manuscript. New York: Columbia University
Press.

KRAPP, G.P. & Dobbie, E. V. K. 1936. The Exeter Book . New York: Columbia
University Press.

LASS, R. 1980. On explaining language change. Cambridge: Cambridge
University Press.

LASS, R. 1986.Conventionalism invention and ‘historical reality’: some re -
flections on method. Diachronica 3,1.15-42.

LASS, R. 1992. Front rounded vowels in Old English. In F. Colman, Evidence
for Old English: material and theoretical bases for reconstruction (=
Edinburgh Studies in the English Language, 2), 88-116. Edinburgh:
John Donald.

LASS, R. 1993. Old English -ian: inflectional or derivational? Vienna English
Working Papers 2,1.26-34.

LASS, R. Forthcoming. Language change and historical linguistics. Cam-
bridge: Cambridge University Press.

LYONS, J. 1977. Semantics. 2 vols. Cambridge: Cambridge Unviersity Press.

MAGOUN, F. P. 1960. The Anglo-Saxon poems in Bright’s Anglo-Saxon
reader. Cambridge, Mass.: Harvard University, Department of En glish.

PENZL, H. 1991. Did Henric van Veldeken (1145-ca. 1200) write in a ‘German’
dialect? In T.F. Shannon & J.P. Snapper, The Berkeley conference on

Dutch linguistics 1989. Issues and controversies, old and new, 57-66.
Lanham: University Press of America.

POPE, J. C. 1966. Seven Old English poems. Indianapolis: Bobbs-Merrill.

ZUPITZA, J. 1959. Beowulf. Reproduced in facsimile from the unique
manuscript British Museum MS. Cotton Vitellius A. xv.2 EETS 245.
London: Oxford University Press.

* * *