Minding the Gap : what we can learn from gaps in the surviving records for Middle English and Anglo-Norman

by Philip Durkin
Deputy Chief Editor, Oxford English Dictionary

Lecture in memory of David Trotter (1957-2015) given at the University of Aberystwyth
20 October 2017

It is a very great honour to be asked to give this lecture today in memory of David Trotter, one of the wisest, most learned, and kindest people I have ever come across in the academic world. The debt I and my colleagues owe to him will become very evident during the course of my talk today.

My topic is what we can learn about language contact between two of the languages of medieval Britain, English and Anglo-Norman, in the centuries after the Norman Conquest by looking closely at the dictionary to which David devoted so much time and energy, the Anglo-Norman Dictionary. As I hope to show, we can learn some surprisingly important things even from considering words that do not appear in this meticulously researched dictionary; if a word that we would expect to find in the AND isn’t there, and (I hasten to say) isn’t in any of our surviving Anglo-Norman records, what might that reveal to us about the nature of what survives, and how that measures up to the realities of language contact in rather remote times?

First, though, I would like to set the scene with a few more general comments about the impact of Anglo-Norman on the lexicon of English. By any measure, it is deep and wideranging. Well over 10,000 words recorded by the Oxford English Dictionary were borrowed from French, and principally from the French of the British Isles, Anglo-Norman, during the Middle English period. In the decades either side of 1400, the rate of borrowing of new words from French reached its highest ever point – the only later period that comes close is the nineteenth century, when new terminology coined by French scientists swelled the fairly constant flow of new words resulting from the enduring cultural prestige of all things French.

You can see this from this graph, based on detailed information from the 40% or so of the OED so far covered by its ongoing comprehensive revision:

Fig. 1: Loanwords from French in published parts of OED3 (Source: Durkin Borrowed Words O.U.P. 2014.)

If we then consider what proportion of all of the new words entering Middle English were of French origin, we find that in the mid to late 1300s this reached over 40%, outweighing compounding or derivation as the commonest mode of origin of new English words. By comparison, in the 1800s, when the raw numbers of loanwords are high, they make up less than 10% of all of the new words recorded by the OED, being far outweighed by words formed within English by compounding, derivation, etc. Now, the OED does not record every single compound evidenced from the Middle English period (not even the Middle English Dictionary does that), so we can’t take this as completely exhaustive data on the lexicon of English, but nonetheless the difference is striking and far outweighs the impact of dictionary selection criteria.

We get another important perspective from thinking about the sorts of words that were borrowed into English in this period. We can learn something just from a bare list of words: consider, for instance,  able, account, age, agree, air, amount, announce, apply, army, art, avoid, award – even before we get out of the ‘As’, we have identified words without which basic everyday discourse on innumerable topics becomes difficult.

We can get a bit more of an empirical grasp on this if we think about the origins of the words that are used most frequently in contemporary English: in other words, how has borrowing in the past shaped the vocabulary that is used most often today. Today there are numerous very large corpora of contemporary written English available for research; note, this is, largely, written language we are talking about, but casting the net as widely as possible to pick up different text types, genres, etc. If we look at the 1,000 most frequent words in such a corpus, we find that just over 50% of these words have been borrowed from other languages or are straightforward compounds or derivatives of borrowed words; of these, over 90% have come from French or Latin or both – I group these together because often we cannot tell which language is the primary input to English, and often there are good grounds for thinking that both have made a major contribution to a word’s history in English. Thus, in the British National Corpus we find words such as able, account, age, – in fact, all of those ‘A words’ I mentioned earlier fall in the 1000 most frequent words in the BNC, as do in total 220 words of French origin and 209 that probably owe something to input from both French and Latin.

What is more, nearly all of those words made their first appearance in English before 1500: the high-frequency vocabulary of English has been impacted relatively little by borrowing that has occurred since 1500 (that also applies to borrowing from Scandinavian languages or indeed from Celtic). In particular, a graph shows that, just like the general vocabulary, the high-frequency vocabulary has been affected particularly strongly by borrowing that appears to have taken place in the fourteenth and fifteenth centuries:

Fig. 2: Origins of 1000 most frequent words in the British National Corpus (Source: Durkin Borrowed Words O.U.P. 2014.)

Fig. 3: Breakdown of high-frequency French and Latin borrowings by period (Source: Durkin Borrowed Words O.U.P. 2014.)

This is especially the case with words that seem to show some input from both Anglo-Norman and Latin, and I would argue that there is a very good reason for this, namely that this is the period in which English takes on increasing numbers of broadly ‘official’ functions from Anglo-Norman and Latin, whether that involves legal or parliamentary pleading, governmental records, business records, or indeed many aspects of religious and intellectual life. I won’t go into this in more detail now, but I believe that understanding these sorts of practical contexts of language contact, and of gradual change in the linguistic practice of multilingual individuals, are essential to understanding the history of English in this period – and few people did more to demonstrate that than today’s honorand.

Just before I leave graphs behind, I’d like to show one more, this time focussing on basic vocabulary, rather than high-frequency vocabulary:

Fig. 4: Main sources of loanwords in a very large basic meaning list (Source: Durkin Borrowed Words O.U.P. 2014.)

Now, basic vocabulary is a tricky concept to nail down, even though we all have intuitions about some things that are learned early and form part of everyday discourse, such as family relationships, parts of the body, some plants, animals, and foodstuffs, emotions, prominent features of the landscape, and so on. For this graph, I used a list of 1,460 basic meanings used as the basis of the cross-linguistic World Loanwords Database project. I won’t go into this in any detail, but we see a broadly similar outline – certainly, it’s clear that borrowing from French in the post-Conquest period is the stand-out event, the main difference from the highfrequency vocabulary being that the words that may be from French or Latin – key to modern educated written discourse – plays much less of a part.

At this point, though, I can imagine David beginning to get a little impatient with my focus on numbers and the big picture stuff  – that’s important in order to get an overview, but what all of the numbers, totals, and graphs roll together are individual word histories, and it’s often the details of those that can be most revealing.

So I would like to look next at a word that is long obsolete and was probably never very common, pople, a type of fur, probably the summer fur of a squirrel. Here is how it looks in the Anglo-Norman Dictionary. I’m pleased to use this example because I discussed it a few years ago with David and – of course – it turned out that the late medieval trade in squirrel fur was one of the innumerable topics to which he had already given some thought and about which he genuinely knew a great deal. I also think it’s a very useful example of how the effects of multilingualism typically look to us in our surviving records, and what difficult questions that poses both theoretically and practically.

pople (‘a type of fur, probably the summer fur of a squirrel’) probably originates as a (dissimilated) variant of French porpre ‘fur or cloth of a dark red colour’, and it was probably from Anglo-Norman that it entered both Middle English and Latin as used in records in England. However, most of the instances of pople are as a vernacular word in Latin-matrix documents, and difficult to assign with complete confidence to either Anglo-Norman or Middle English. The Anglo-Norman Dictionary confines itself here, very reasonably, to a couple of examples occurring in sentences clearly written in Anglo-Norman. Things get a bit trickier for a lexicographer of English. Most of the quotation evidence contained in the main body of OED’s original entry of 1907 (and likewise in the corresponding entry in the Middle English Dictionary) has been moved to the etymology section in the revised OED entry of 2006, leaving just an antiquarian reference from the seventeenth century, and a single Middle English example of 1493 where the compound with fur probably tips the balance for analysis as English. Many of these quotations are also difficult from a bibliographical and textual perspective: for instance, two examples are from H. T. Riley’s Memorials of London (1868), which is both calendarized – that is to say, a running summary of a Latin document with occasional words cited directly from the original – and, as past experience has shown, it also sometimes modernizes or otherwise adjusts even those original word forms that it cites. The last of these quotations, from the Will of William Norton, appears to have been taken in the first edition of the OED directly from consultation of documents in the Public Record Office (at least, this is what the bibliographical style adopted suggests). My own checking of an image of this document kindly supplied by the P.R.O. shows that there is very clearly a bar through the second l. This could be an otiose mark, but it could equally well be a mark of suspension for Latin (ablative) popello – hence, this particular quotation may simply show French pople borrowed into Latin, and not a single-word switch to a vernacular word at all. Accordingly, it has now disappeared from the OED entry entirely. (The ability to revisit cases like this is one of the virtues of online publication.)

This particular example primarily concerns the world of trade, but other – typically more complicated – examples could easily be provided from law, business, religion, the intellectual world in general. I hope it helps highlight one key point, which underlies all of the practical difficulties for lexicographers – namely, that the vast majority of the content of the Anglo-Norman dictionary was produced by people who formed a subset of the Middle English speaker community, and most of them had English as their first language. They may have been a minority of the overall English-speaking population, but, until probably as late as the late 1300s, they included nearly everyone who was literate, and hence a large proportion of the people who had power and influence in their society. When we are talking about transfer of words from one language to another, we are in the first instance talking about people who knew a word in Anglo-Norman and used it when they were writing or speaking Anglo-Norman, starting to use that same word when they were writing or speaking English. When we are considering non-core, content-rich vocabulary like pople, we may even wonder how far words like this can said to belong categorically to one language or another in a context where all three languages are used in different functions by the same individuals, who nearly all have English as their first language.

But for the remainder of my talk, I would like to look in some detail at words that are by no means on the remote fringes of the vocabulary, namely the words for ‘grandfather’ and ‘grandmother’ in medieval English and medieval French, especially Anglo-Norman.

In modern English the usual words for grandfather and grandmother are hybrids, with the native English words father and mother prefixed by the French borrowing grand-. The adjective grand is found in English from the second half of the fourteenth century, which, as we have already seen, is a key period for transfer of vocabulary from both French and Latin into English. However, it seems not to have been particularly common even in late Middle English, and became more common only after the end of the Middle Ages; it appears an unproblematic borrowing from French, probably specifically Anglo-Norman, as is suggested by its contexts of use and by the occurrence of forms such as graund.

The kinship terms involving grand– are, however, found much earlier than this in English:

  • grandame meaning ‘grandmother’ is found from c1225 (this is a manuscript date, the composition was probably a little earlier),
  • grandsire meaning ‘grandfather’ from c1300,
  • grandfather itself is found from 1424, grandmother also from 1424 (in the same document, a will).

In continental French:

  • grandame is first found in the fourteenth century,
  • grantsire in the twelfth century,
  • grant pere in the mid fifteenth century (or late fourteenth in the meaning ‘male ancestor’),
  • grant mere in the thirteenth century.,

So we have a date in French earlier than the English evidence only for grandsire and grandmother. For grandame (the earliest of all these terms in English) and grandfather, all French evidence is later. Perhaps most interesting of all, none of these terms are found in the Anglo-Norman Dictionary, and none occur in any of the searchable texts on the AND Hub (I say this with more confidence than would otherwise have been the case because David checked this for me himself several years ago). So comparing the dictionary record for each language reveals a startling absence: we would expect the kinship terms in Middle English to be borrowed from Anglo-Norman, not imported from across the Channel, and anyway, even the continental French evidence presents startling disparities with the English evidence.

How can we probe this puzzling situation a little further, and maybe even tentatively work our way towards an explanation?

It certainly is not the case that Anglo-Norman texts do not ever speak about grandfathers or grandmothers. They do, but they use the earlier, Latin-derived terms, aieul and aieule (ultimately from a derivative of Latin avus), and these are found in Anglo-Norman (as ael and aele; the masculine form persisting long in legal discourse in the writ of ael); a little later, they are also borrowed into English (although they were apparently never common in English). In continental French, these older words for grandfather and grandmother were gradually replaced by the newer phrasal types using grand. Very likely, the same thing was going on in Anglo-Norman, but hasn’t made it into our surviving texts. Why not? Is this evidence perhaps telling us something about register, that the kinship terms were excluded from formal, literary registers in Anglo-Norman, but appear much earlier in English, probably as a result of borrowing in close familial settings, with, perhaps, the prestige of a French borrowing outweighing any sense that these were informal terms? This seems to me the most probable explanation, but there’s probably more to it than this. After all, why should French have replaced what appear to have been perfectly good and long-established single words meaning grandfather and grandmother with these new-fangled phrasal expressions meaning literally ‘big father’ and ‘big mother’? This, of course, brings us investigating the linguistic choices that individual speakers were making many centuries ago, and that is very difficult and dangerous territory, but I think at least worth a look. I think that we find an interesting connection if we turn back to English for a moment, and if we also consider its continental neighbours.

English, of course, also had existing ways of referring to a grandfather or grandmother. An excellent way into exploring the various different ways of referring to the same thing that have existed in the history of English is The Historical Thesaurus of the Oxford English Dictionary, the fruit of more than 40 years of painstaking research work at the University of Glasgow. (David grasped the importance of this in an instant, and did pioneering work on implementing similar structures for the Anglo-Norman Dictionary.) For the meanings ‘grandfather’ and ‘grandmother’ this alerts us to the Old English terms ealdfæder and ealdmōdor (literally ‘old father’ and ‘old mother’) which survived well into Middle English; there are also remodelled forms in Middle English, oldfather and oldmother. As the Dictionary of Old English shows, some of the extended meanings of eald ‘old’, such as ‘superior, greater’, show points of contact with those of French grand, a point to which I will return.

In some other older West Germanic languages we find similar expressions meaning grandfather’ and ‘grandmother’, for example Old Frisian aldafeder grandfather, Old Saxon aldfadar forefather, Old High German altfater grandfather, forefather, (Church) patriarch. The fact that they do not invariably mean ‘grandfather’ is interesting. In Old English too ealdfæder (and some closely related forms) can also mean ‘forefather’ or ‘ancestor’, but the meaning ‘grandfather’ seems core. In Middle English, things are less clear; in particular,  oldfather, eldfather, oldmother, and eldmother are found with the meaning ‘father-in-law’ or ‘mother-in-law’ almost as often as ‘grandfather’ or ‘grandmother’. This variability between two different specific family relationships suggests some real fluidity or even instability in terminology. We do not find this sort of variation in meaning for the grand– words ((grandfather, grandsire, etc.) in either English or French. But if we cast our net a bit further, we do find some interesting similarities.

If we return for a moment to the Historical Thesaurus data for ‘grandfather’, we find that other terms for ‘grandfather’ in Middle English are belfather and belsire – which may puzzle French specialists, who will recall that in Middle French beau-père is ‘father-in-law’, belle-mère is earliest recorded meaning ‘stepmother’, while belsire is mostly found as a respectful form of address – although the great Französisches etymologisches Wörterbuch records that formations in bel– are indeed sometimes found meaning grandfather/grandmother in particular regional varieties of Middle French and in later regional use. And, if I can introduce just one more set of words at this point, in English in the fifteenth century, mostly in Scottish sources, we get gooddame ‘grandmother’ and goodsire ‘grandfather’, but also goodfather ‘father-in-law’ and goodmother ‘mother-in-law’. I would suggest that what all of this data has in common is that new ways are being found of expressing relationships that are like that of a father or mother but at one remove (whether in generation, or as ‘–in-law’ or ‘step-’ parent). In English we can be certain that grand– has shown a decisive shift to ‘one degree removed in ascent or descent’ when we find grandchild, grandson, and granddaughter, where the ‘big’ or ‘more honoured’ metaphors make no sense, but these are first attested well after the end of the medieval period, grandchild in 1569, grandson in 1573, and granddaughter in 1608.

Some of you may by now be thinking ‘when is he going to mention the other big changes that occurred in English kinship terms’? Well, the answer to that is now. It has long been known that in the wake of the Norman Conquest the kinship terminology of English showed some big changes, not in the most basic terminology of father, mother, brother, sister, son, daughter, but when we come to consider uncles and aunts there were really dramatic changes. In Old English, father and mother were fæder and mōdor, essentially as today. But a paternal uncle was a fædera while a maternal uncle was an ēam, and likewise a paternal aunt was a faðu while a maternal aunt was a mōdrige. So, after the Conquest, not only did Anglo-Norman uncle and aunt replace native terms, there was also a very significant restructuring of the system –paternal and maternal uncles and aunts became terminologically equivalent, no longer distinguished conceptually by different terms, new uncle subsumed both fædera and ēam, without the specificity of either. Of course, it remained perfectly possible to refer to a ‘paternal uncle’, just as it is today, but not by using a single, distinctive word. Similar things happened eventually in other West Germanic languages as well. Historians and historical anthropologists as well as linguists have looked at this phenomenon, and other restructurings (for instance in the words for nephews and nieces), and linked this with a shift from a conceptualization of the family in essentially clan terms, where a wife married into a husband’s extended clan, to a focus more on the marital family unit of mother, father, and children – which many have seen as typifying the conceptualization of the family in the Latin and hence Romance world, in Roman Law, and in the Roman Church. In particular, in Anglo-Saxon culture a particular resonance seems to attach to the relationship between ēam, maternal uncle, nephew, or sister-son, sweostor sunu, which many have seen as reflecting the role of the maternal uncle as an important guarantor of a child’s welfare from another clan, that which the mother had originated from, and left at her marriage. With the transition to the new kinship terminology, this appears to be lost.

In this context, grandfather and grandmother have attracted little attention, because the change is much less dramatic: in structural terms, nothing has changed, because Old English had identical terms for paternal and maternal grandparents, and even the new terms grandfather and grandmother (and indeed even grandsire and grandame) show a strong similarity to the old ones ealdfæder and ealdmōdor – the grandfather is named with a word based on the word for ‘father’ plus a modifier, and likewise the grandmother.

But how does all of this context help us understand the absence of any of these grand– terms from Anglo-Norman, and their early appearance in English? Well, one thing we have established is that there was a lot of instability and change in the kinship terminology in English and among its Germanic siblings, extending to how the whole set of relationships were conceptualized, rather than just change in the labels occupying certain slots in the system. I mentioned earlier that Old Frisian, Old Saxon, and Old High German all had formations of the ‘old father’ type. In the ‘Middle’ periods of most West Germanic languages we find a shift to a new ‘big father’ type: so we find Middle Dutch grotevader, Middle Low German grotevader, Middle High German grossvater, and likewise with words for ‘grandmother’. These mostly show up in the historical record rather late, and it has normally been assumed that they show the influence of the French grandsire, grandpère, etc. – but, as we have seen, the French evidence is also rather late (and not there at all in Anglo-Norman), and in Middle Low German grotevader and grotemoeder go right back to the 12th century, the earliest of any of the ‘big father’ or ‘big mother’ types in any language, Germanic or Romance. Which brings me to my hypothesis about what is possibly going on here…

Naming ‘grandfather’ and ‘grandmother’ with words that show some modification of the words for ‘father’ and ‘mother’ is clearly ancient in Germanic languages, ‘old father’ and ‘old mother’ being the earliest types. Contrary to earlier assumptions, the ‘big father’ type (i.e. in French terms grandsire or grandpère) may well also be of Germanic origin, probably somewhere on the Rhine – the appearance of grandsire, grandpère, etc. in French may show influence from Middle Low German grotevader, etc., and not vice versa. We may see some echoes of Germanic influence also in the semantic variability of the bel- types in French too – this is certainly very comparable to the instability of the old eldfather type in Middle English, which comes to mean ‘father-in-law’ or just ‘ancestor’ as often as it means ‘grandfather’. There is certainly a general tendency across West Germanic languages and French for using modified forms of the words for ‘father’ and ‘mother’ for relationships at one remove from father and mother – and that appears to be Germanic in its origin.

So where do Anglo-Norman and Middle English grandfathers and grandmothers fit into these cross-currents of modification of the kinship system to a Romance model, alongside probable Germanic influences on the actual terms deployed in French?

I think that Middle English grandsire, grandame, grandfather, and grandmother tell us that kinship terms in grand– must have existed in Anglo-Norman – it’s implausible that we have cross-Channel importation of such fundamental kinship terms into Middle English while Anglo-Norman invariably retained aieul and aieule. And the terms must have come from French of some description, because the adjective grand doesn’t otherwise exist in English until centuries later. Why don’t the grand– types show up at all in Anglo-Norman sources? Perhaps a bit of chance, perhaps more to do with register and the nature of our surviving texts – which may also explain the disparity in dates between Middle English and the Continental French evidence. Plus, English was probably very receptive to the ‘modified father/mother’ terms – it had long had the ‘old father’ type (where the connotations may have been slightly more ‘honoured, revered’ anyway), and it simply replaced this with a new, Romance ‘big father’ type (also with connotations of honour, reverence, etc.).

Which is a very long way about the houses to say that the scholarship in the Anglo-Norman Dictionary matters tremendously, for what it tells us about Anglo-Norman, Middle English, French on the continent, and many further linguistic relationships. It also shows why the Anglo-Norman Dictionary needs to be a full dictionary of a language variety – something David Trotter fought hard to achieve – and not just a list of differences from the continental variety. Because when the work has been done as thoroughly as is now the case, we can learn a great deal even from what isn’t there.