Editing is Hell, and normalization is an illusion

As a procrastinatory excursion, here are some thoughts about editing historical texts. Rather than an insightful comment on editorial philosophy, the following stems from practical matters and contains nitty-gritty details, and is not written in conversation with other editors (sorry). I’m sure everything I say here has been said before, but repetitio etc.

1. Why normalization is an illusion

Years back, I found I had a problem. I was considering the expansion of abbreviated words, and one of the words in question was merchant. This word was very frequent in the text I was working on, and occurred in various forms, some of them abbreviated: “marchant”, “mrchnt”, etc. I thought it would be a straightforward matter to settle on an expanded form. I decided I would normalize according to most attested usage. That is to say, editorial expansions would reproduce the most frequent forms fully spelled out.

In other words, if for instance “marchant” was the most frequent unabbreviated spelling of merchant, then “mrchnt” would be expanded as “marchant”. (With editorial expansions indicated, e.g. “marchant”.)

My next step was to establish what was the authorial preferred spelling of the word merchant. This is what I found (word stem forms only):

mrchant* 1
mrchnt* 31
mrcht* 1
marchand* 1
marchant* 31
marchnt* 21
marshant* 1

Out of 87 hits of merchant*, only 33 were spelled out fully, i.e. 60% were abbreviated forms. And the abbreviated form “mrchnt*” was as frequent as the fully spelled out form “marchant*”.

This made me stop and think: if the fully-spelled-out form of a (stem of a) word is much rarer than abbreviated forms, can we really claim it represents authorial preference? If we chart the possible spellings of merchant as a sequence of graphs, we get the following:

Note that this image shows both the actual realized spellings in the texts, and also possible spellings which do not occur (such as “mrchand*”). (Note also that other possible contemporary spellings are not given, such as “mrchāt”, or indeed “merchant”).

I consequently completely abandoned the idea that we can ‘reconstruct’ authorial spelling, and also determined to make sure to explicitly indicate editorial intervention at all times. A spelling like “mrchnt” should never be expanded as “marchant”. Instead, modern forms should be used: the only correct expansion here is “merchant”.

Having said that, using modern spellings is of course not an option in dead language varieties, such as Middle English. And further, even in Early Modern English there are graphemic and orthographical features without counterpart in Present-Day English(es). What to do with obsolete suffixes, like in “hath”, or “didst”? And what about graphs which were already obsolete but replaced with other contemporary ones, such as <y> for thorn <þ>, as in “ye” ‘the’ and “yt” ‘that’? And what about graphs and special characters which were common in the period but which we no longer use?

2. Editing is Hell

Here is one example of how just such a special character – a brevigraph – can cause serious problems when deciding on best editorial practice.

The image below is from a document from 1599, written in a late Elizabethan cursive (aka English secretary hand), with a slightly old-fashioned ductus. It reads: “her matꝭ servyce” (or “ſervyce”, if we want to retain the long <s>1). The second word is an abbreviation of majesty’s.

The <e>-shaped graph in the abbreviated second word is what I usually call an -es-graph. Originally a medieval brevigraph2 for (usually word-final) “-is”, “-es”, and “-ys”, in Early Modern English cursive hands <ꝭ> is also used for (word-final) “-s”. Some sources claim it indicates a plural or possessive, but I have found it used in proper names, prepositions, verbs and adverbs too. (Possibly it is more frequent in one rather than another, but I expect that in its distribution scribal preference is more significant than part of speech).

So the question for today’s workshop is, what to do with the abbreviation “matꝭ”?

What indeed.

Depending on your editorial principles, I can come up with 23 different possible outcomes, based on choices regarding i) spelling, ii) contractions (abbreviations), iii) superscripts, and iv) brevigraphs (special characters).

The outcomes – and the choices – are shown in the massive table below. Let me help you read it. The top three rows indicate options when retaining original spelling. The blue section in the middle indicate choices and outcomes when normalizing spelling (since regardless of my personal opinion as stated above, normalization is a common editorial practice), and the bottom ten rows the same for modernized spelling.

The numbers show whether editorial intervention has been indicated or not. “1” indicates that the editorial intervention is marked (e.g. “majesty’s”), and “0” that it is unmarked (e.g. “majesty’s”). And “(1)” means that the feature is marked, but it is so in the original text (i.e. the superscript and the brevigraph).

Some of the features are not strictly speaking divisible – that is to say, the contraction is marked by the superscript, and hence you can’t show that you’ve changed one without doing the same for the other: expanding the contraction and lowering the superscript are in essence the same thing. Thus “(1)ss” means that contraction is marked since what has been done to superscripts is marked, and “(1)c” means that superscript is marked since what has been done to contraction is marked.

Some notes are in order:

* Arguably since the contraction (and superscript) is not marked, this does not qualify as a representation of the original spelling.

† When lowering of the superscript or expansion of the contraction are indicated, indication of the normalization or modernization of the brevigraph gets subsumed – when using italics to mark editorial interventions. If other methods are used – apostrophes, parentheses – it is possible to make this distinction, as seen in the rightmost column.

‡ Some of the possible outcomes are unlikely to be chosen by the editor, because editorial intervention can also make the word more difficult to parse, rather than less. So for instance the edited forms “mates” and “mates” seem to me undesirable outcomes – “mates” is scarcely better, whereas “ma’tes” (or “ma’tes”) at least indicates that the word is an abbreviated form.

3. Um, argh?

Well, yes exactly. And this is without going into the jungle of brackets and other symbols used by editors to distinguish between different kinds of things in the text, such as interlineal insertions, deletions, damage, etc etc (a good discussion of which, with clear examples, can be found in the appendix to Michael Hunter’s Editing Early Modern Texts (2007)).

And also, other cases – other graphs, other methods of abbreviation, other hands – produce different problems, so I’m certain that the above table does not suffice for all editorial problems.3

What can the editor do, then?

I think this question can be answered: The editor can do whatever the hell they please. What they should do, in any case, however, is to make sure that all editorial interventions in the text are visible and the original form is recoverable. How they do this is another matter, as is the extent of their meddling. But there should in any case be a chapter or document setting out very clearly the editorial principles and practices followed in the edition.


1)  Who would want to do that!, you exclaim. Well, I would, for one. Our knowledge of Early Modern English handwriting is ridiculously limited, and in particular quantifiable information is scarce. So you need geeks like me to count them long <s>s.

2)  The -es-graph in Unicode – <ꝭ> – may be okay for medieval texts, but it is quite unlike the <e>-shaped -es-graphs in Early Modern English secretary hands. For a type facsimile edition, I would need to find a better character. And indeed I have done so – as seen by the -es-graph used in the table (which I got from the Electronic Text Edition of Depositions 1560–1760, available on the CD accompanying Merja Kytö, Peter J. Grund & Terry Walker, Testifying to Language and Life in Early Modern England (Benjamins, 2011)).

3) Something to avoid are what might be called hybrid forms. For instance, combining a normalized expanded stem with the word-final brevigraph: “maiestꝭ ”. Or then expanding the contraction and modernizing the stem of the word, but normalizing the expansion of the brevigraph: “majestes”. Expanding and modernizing the contraction but leaving the superscript just looks silly: “majesty’s”.

Leave a comment

Your email address will not be published. Required fields are marked *