A re-classification of Old English nouns.

ABSTRACT

The standard classification of Old English nominal inflections, employing the Proto-Germanic root structure, fails to reflect synchronic features of the system. This paper follows Kastovsky (1995) and Lass (1997) in rejecting the traditional model as incompatible with synchronic data, and postulates an alternative view on the categorisation of Old English nouns. The new interpretation makes it possible to capture synchronic relationships within the Old English nominal paradigm, as well as predict developmental tendencies observable in the Middle English period.

1. Introduction

In his paper Roger Lass (1997) argued that there are strong methodological doubts as to the reality of gender/declension assignments in the case of Old English nouns. He rightly observed that the degree of gender indeterminacy for many nouns is so high that it is virtually impossible to say with any degree of certainty that such assignment is at all possible. Similarly, Dieter Kastovsky (1995) questioned the validity of the classification of nominal paradigms as found in standard textbooks devoted to Old English nominal morphology.

The aim of this paper is to continue the discussion of issues raised by these authors as well as by Krygier (1998). It will be argued here that there is no synchronic motivation for the type of declensional classification commonly accepted by students of Old English, and an alternative proposal will be suggested.

2. The critique of the traditional account

When in 1969 Alfred Reszkiewicz published his Synchronic essentials of Old English he included a very interesting statement in the preface, where he observed that "a synchronic (descriptive) presentation should precede diachronic explanations, even in historical grammars" (Reszkiewicz 1998 [1969]: 7). These words sound so commonsensical that treating them as worth repeating after over thirty years may seem odd. However, judging from the prevailing approach to Old English morphology, and nominal morphology in particular, Reszkiewicz's wishes still await fulfilment.

For no grammar of Old English can be viewed as truly deserving this title. It is standard practice to apply Proto-Germanic inflectional categories to a language, which, although its direct descendant, is at a completely different stage of development. These sentiments have been fully expressed by the author of this paper elsewhere (Krygier 1998), therefore at this point only a short synopsis of the main arguments will be given.

In Proto-Germanic the morphological structure of a noun was very transparent and consisted of three elements: a root, a stem formative, and an inflectional ending, e.g., *dag-a.z 'day'. The stem formative characterised the noun as belonging unambiguously to one of many nominal declensions.

By the time of Old English, however, stem formatives were no longer distinguishable, suffering the fate of other medial unstressed syllables. In no nominal paradigm of Old English can one find any morphological element that would unambiguously serve as its characteristic feature. Moreover, even phonetic processes accompanying the reduction and loss of some stem formatives in Preliterary Old English (henceforth POE), such as i-umlaut, did not produce alternations that could be used to identify reflexes of Proto-Germanic paradigms (as, e.g., fronted vowels were common in at least four different inflectional classes, viz, masculine ja-stems, neuter ja-stems, jo-stems, and i-stems). Therefore, there is no synchronic motivation for preserving the Proto-Germanic descriptive framework when classifying Old English nouns. And yet, this perspective is widely adopted even in very recent publications; e.g., Welna (1996) in his English historical morphology states without hesitation that:

[w]ith regard to their stem structure, Old English nouns can be classified into three groups which include words with:

- vocalic stems, ending in -a-, -o-, -i-, -u- (...);

- consonantal stems, ending in -n-, -r-, -o-, -nd-, -iz/az-;

- and root-consonant stems;

all forming the respective declension types.

(Welna 1996: ??1.2)

Welna's approach is probably the clearest example of using non-existent criteria to classify Old English nouns, but by no means an isolated one.

If one accepts the conclusion stemming from the preceding discussion, the descriptive adequacy of the standard classification of Old English nouns, here presented for reference's sake as Figure 1, may begin to seem rather doubtful.

For if the stem formatives are not there to justify such a classification, it would seem necessary to find some other evidence in its favour. However, it is rather hard to come by. Lass (1997) rightly points out that, e.g., the paradigmatic difference between the types exemplified by OE stan "stone" and OE scip "ship" is limited to two case endings only, viz. the nominative and accusative plural. If one remembers that these two cases were formally identical in the respective paradigms, the difference becomes even more tenuous.

Naturally, one can find such contrasts where the differences are much more considerable, as, e.g., between the stan and talu "tale" types. However, there is no denying the fact that inflectional endings alone cannot support the current model of the Old English nominal paradigm.

There are also other curious inconsequences. One of the major ones is customarily treating the a-stem masculines and neuters as belonging to different paradigms, while masculine and neuter n-stems are without exception collapsed into one inflectional type. Another is treating as linguistic reality such artificial constructs as i-stems or u-stems. Campbell (1959) in his description of the latter states that, e.g., "no fem. nouns are free from the influence of the o-stems, but some u-stem forms occur from the following" (Campbell 1959: [section]614), and then goes on to enumerate as many as five (!) nouns from this particular category. It is very risky to postulate the existence of a paradigmatic type on the basis of five items, none of which is free from analogical remodelling. This would be equivalent to claiming that Modem English contains a weak nominal declension, because there are three nouns that take the -en ending in the plural. And yet every student of English (and many scholars) will unflinchingly st ate that the u-stem nouns were one of the inflectional types available to the speakers of Old English.

This leads to another aspect of the OE nominal morphology which urgently needs rethinking, namely the category of grammatical gender. Most scholars implicitly assume its existence in Old English as well as its connection with paradigmatic types. The category is so deeply entrenched in linguistic tradition that two of the most important works dealing with Old English morphology, Campbell's Old English grammar (1959) and Brunner's Altenglische Grammatik (1965), do not consider it worthy of any comment, and launch straight into the presentation of the Old English nominal inflection proper. Others simply state that Old English nouns came in three different genders, and stop there.

If this kind of consensus seems to have been reached, why should one want to question it? For one thing, the connection between grammatical gender and paradigmatic type appears to be misconceived. More precisely, it is difficult to see how case markers can be made to carry gender information as well. After all, their repository is preciously small. Already in Elfredian Old English there were as few as 9 distinctive endings: -[empty set], -a, -an, -as, -e, -ena, -es, -u, and -um, in the whole of the nominal paradigm, excluding irregular nouns. In Late Old English their number fell to 6: [-[empty set]], [-??], [-??n], [-??n??], and [-??s], all of them expressing more than one inflectional category.

It would be really difficult to expect such an overburdened system not only to express case-number relationships but also to participate in grammatical gender assignment. Each of the endings would be polyfunctional, and, depending on the syntactic context, it would mark all three genders in addition to a number of case-number types. It is very hard to see how: to take a very easy example -- the [-??s] ending could be both masculine and neuter as well as genitive singular, nominative plural, and accusative plural. Even from such a simplified perspective there is something distinctly wrong with this model.

It is very interesting to observe that the only alternative approach has received virtually no attention in scholarly publications on the subject. The only major Old English grammar that actually formulates it explicitly is An Old English grammar (1957) by Quirk and Wrenn. There it is stated that: "OE nouns fall into three groups, masculine, neuter and feminine, according as they require one or other form of the demonstratives Se, paet, seo, and enforce corresponding agreement on the other demonstratives, on adjectives, and on pronouns" (Quirk and Wrenn 1957: [ss]25). Nevertheless, it seems that this statement was intended purely descriptively, without full realisation of its theoretical implications.

The notion of noun-external marking of grammatical gender as a secondary (and extraparadigmatic) property of Old English nouns has a number of clear advantages over the traditional approach. Firstly, it removes the weight of gender assignment from inflectional endings to other grammatical elements, viz. demonstrative pronouns, to which others, e.g., morphemic structure, could possibly be added. Hence, the subsequent decay of the grammatical gender could be viewed from a different perspective, independently from the loss of nominal inflections. Secondly, it correlates nicely with current interpretations of the situation in, e.g., Modern German, the nominal system of which is quite similar to that of Old English. It is often assumed that morphological assignment criteria are indirect ones there and are relevant only when semantic ones fail. Moreover, "the role of ... morphology as a gender determinant is a psycholinguistic sense will depend heavily on frequency" (Zubin and Kopcke 1981: 443); a good case in poin t here are the Old English abstract nouns such as those in -ness, which regularly have no plurals, or in -pu, which on top of that are uninflected in the singular, and yet are assigned the feminine gender. All in all, there is enough evidence to treat gender and paradigm assignment as two parallel, yet distinct processes (cf. Kilarski 2001: 47-49).

So far this paper has tried to show two things: (a) that morphological structure of the noun cannot serve as a criterion for paradigm assignment in Old English, and (b) that grammatical gender should be excluded from any attempt at finding one. On a more positive note, an approach will be now presented which will be faithful to surviving data in trying to reconstruct the truly synchronic layout of the nominal system in Old English.

If neither the shape of the nominative singular nor of the nominative plural can be used for this purpose, it seems advisable to concentrate on whole sets of similarities. In other words, it could be conceived that nominal inflections in Old English behaved similarly to Old English strong verbs (cf. Krygier 2001a for details), namely as a continuum with gravity centres of shared similarities. Major paradigms would be those formed by inflectional types similar enough to be perceived by a linguistically naive native speaker as belonging to one paradigm. Those less distinctive would be attracted by the closest gravity centre, in constant danger of disintegrating and being incorporated into the major pattern.

In order to isolate gravity centres of the Old English nominal system a similarity matrix was prepared, in which the number of shared similarities among all productive inflectional types was calculated.

The analysis of the results yielded by the similarity matrix allows one to postulate a system of Old English nominal inflection noticeably different from the traditional one. Using paradigm unity as the main criterion, three declensional types can be reconstructed. The descriptive labels are those of the genitive singular, as the distribution of this particular case-number ending aligns neatly with paradigm boundaries. However, it is important to remember they are only labels and nothing more; genitive singular itself cannot serve as the decisive criterion here any more than, e.g., nominative singular in dividing Older Germanic languages into the dagr, dags, and daeg groups.

Declension I (the es-type): traditional strong masculines (stan) and neuters (word, scip)

Declension II (the e-type): traditional strong feminines (talu, lar)

Declension III (the an-type): traditional weak masculines (hunta), feminines (tunge) and neuters (eare)

What, if any, are the advantages of this model over the traditional one? For if one chooses to question one of the basic tenets of the understanding of Old English morphology, one should have good reasons for it.

The first advantage is that the model is truly synchronic. It is based on relationships that existed in Old English and were available to native speakers of the language, rather than on unrecoverable morphological structures, lost hundreds of years earlier.

Secondly, the model is much simpler. It is a well known fact that in language processing, as in computing, a simpler algorithm has a distinct advantage over a competing complex one (cf. e.g., Pinker 1999 for a very convincing, if at times simplistic, application of this idea to the Modern English verbal system). Under the traditional model, a lexical entry for stan would look like this:

There would be eight competing, often very similar patterns, and every one of them would come with a predetermined grammatical gender. On the other hand, the new model would generate a structure like this one:

As the main lexical entry is based on the nominative, all nouns with the exception of traditional strong neuters can be underspecified with regard to plural, which results in major reduction of model complexity in contrast to the traditional interpretation; cf. Figure 6.

Thirdly, the model has clear retrodictive power. By retrodiction it is understood here the ability to correctly predict and explain the subsequent developments as they really happened. The traditional model has to mix diachronic and synchronic criteria to achieve this goal, as stem structure obviously has little to do with paradigm realignments of Late Old English and Early Middle English -- after all it had not existed for over five hundred years then. The new model, however, very clearly indicates the two main paths of development that the system was bound to follow if guided by morphological considerations alone. One of them would be the abandonment of the additional plural information in the scip-type nouns, thus forcing them to conform fully to the Declension I paradigm and accepting the plural -as, later -es ending, which, unless blocked by seimantic factors, did happen (cf., e.g., Newman 2001). The other would be for Declension II nouns to gravitate towards Declension I, as more similar than the wholly alien Declension III, which in time should produce a bipolar system. Once again, this is exactly what ultimately happened (as long as it is possible to talk about the continuation of Old English paradigms).

In conclusion, the present paper is an attempt at offering a truly synchronic interpretation of the Old English nominal system. It has a number of theoretical advantages over the traditional model, which does not in the least reflect the Old English data. Whether and to what extent later developments can confirm or disprove its validity requires further research, however, it deserves attention in shifting the focus from the diachronic to synchronic plane.

Figure 1

Old English standardised nominal system.

     SM                SN                SN                SF

Nsg  stan-[empty set]  scip-[empty set]  word-[empty set]  tal-u
Gsg  stan-es           scip-es           word-es           tal-e
Dsg  stan-e            scip-e            word-e            tal-e
Asg  stan-[empty set]  scip-[empty set]  word-[empty set]  tal-e

Npl  stan-as           scip-u            word-[empty set]  tal-a
Gpl  stan-a            scip-a            word-a            tal-a
Dpl  stan-um           scip-um           word-um           tal-um
Apl  stan-as           scip-u            word-[empty set]  tal-a

     SF               WM        WF        WN

Nsg  lar-[empty set]  hunt-a    tung-e    ear-e
Gsg  lar-e            hunt-an   tung-an   ear-an
Dsg  lar-e            hunt-an   tung-an   ear-an
Asg  lar-e            hunt-an   tung-an   ear-e

Npl  lar-a            hunt-an   tung-an   ear-an
Gpl  lar-a            hunt-ena  tung-ena  ear-ena
Dpl  lar-um           hunt-um   tung-um   ear-um
Apl  lar-a            hunt-an   tung-an   ear-an

Figure 2.

Nominal inflectional endings and their functional load in Old English

       Classical Old English                Late Old English

-[empty set]             9/64      -[empty set]        9/64
-a                       10/64     -??                 25/64
-e                       12/64
-u                       3/64

-an                      14/64     -??n                22/64
-um                      8/64

-as                      2/64      -??s                5/64
-es                      3/64

-ena                     3/64      -??n??              3/64

Figure 3.

Similarity matrix for Old English nouns - AElfredian times.

       stan  scip  word  lar  talu  hunta  tunge  eare

stan   8     6     6     4    3     1      1      1
scip   6     8     6     4    3     1      1      1
word   6     6     8     4    3     1      1      1
lar    4     4     4     8    7     1      1      2
talu   3     3     3     7    8     1      1      2
hunta  1     1     1     1    1     8      7      6
tunge  1     1     1     1    1     7      8      7
eare   1     1     1     2    2     6      7      8

REFERENCES

Andersen, Henning (ed.)

1995 Historical linguistics. Selected papers from the Eleventh International Conference on Historical Linguistics, Los Angeles, 16-20 August 1993. (Current Issues in Linguistic Theory 124). Amsterdam: Benjamins.

Campbell, Alistair

1959 Old English grammar. Oxford: Clarendon Press.

Kastovsky, Dieter

1995 "Morphological reanalysis and typology: The case of the German r-plural and why English did not develop it", in: Henning Andersen (ed.), 227-238.

Kilarski, Marcin

2001 Gender assignment of English loanwords in Danish, Swedish and Norwegian. [Unpublished Ph.D. dissertation, Adam Mickiewicz University: Poznan].

Krygier, Marcin

1998 "On a synchronic approach to Old English morphology", Folia Linguistica Historica 19: 119-128.

2001a "Reconsidering the history of the English verbal system", Studia Anglica Posnaniensia 36: 51-60.

2001b "Strong verbs in Lazamon's Brut", paper presented at the 36th International Congress on Medieval Studies, Kalamazoo, 2001.

Lass, Roger

1997 "Why house is an Old English 'masculine a-stem"'; in: Terttu Nevalainen -- Leena Kahlas-Tarkka (eds.), 101-109.

Nevalainen, Terttu -- Leena Kahlas-Tarkka (eds.)

1997 To explain the present: Studies in the changing English language in honour of Matti Rissanen. (Memoires de la Societe Neophilologique de Helsinki 52). Helsinki: Societe Neophilologique.

Newman, John G.

2001 The diffusion of the regular s-plural formative among Old and Middle English nouns (700-1400). [Unpublished Ph.D. dissertation, Warsaw University].

Peters, Robert A.

1965 "Case-number morphs of Old English nouns", Linguistics 14: 41-51.

Pinker, Steven

1999 Words and rules. The ingredients of language. HarperCollins: New York.

1961 Elementy gramatyki historycznej jezyka angielskiego. vol. I: Jezyk staroangleiski.

Reszkiewicz, Alfred

Warszawa: PWN.

1969 Synchronic essentials of Old English: West Saxon. Warszawa: PWN. [1998]

Welna, Jerzy

1996 English historical morphology. Warszawa: Wydawnictwa Uniwersytetu Warszawskiego.

Zubin, David A. -- Klaus-Michael Kopcke

1981 "Gender: A less than arbitrary grammatical category", CLS 17: 439-449.