Word formation in Early Middle English: Abstract nouns in the Linguistic Atlas of Early Middle English

Anne-Christine Gardner, Ruprecht-Karls-Universität Heidelberg

Abstract

1. Introduction

For a long time, research interest in the Middle English lexicon seems to have been centred more on borrowings than on word formation. Derivations, however, are no less interesting as changes and competition within the derivational system inherited from Old English, as well as the impact of suffixes of Romance origin introduced in the aftermath of the Norman Conquest can be observed. Since the early 1990s three larger-scale studies have brought derivation in Early Middle English to the limelight, mostly with regard to semantics and productivity (Zbierska-Sawala 1993, Dalton-Puffer 1996, and Ciszek 2008; Trips 2009 concentrates on three suffixes with a larger diachronic scope), yet knowledge about the historical developments still needs to be expanded in various directions. The present paper aims to shed more light on a subset of derivations in Early Middle English, specifically synonymous abstract nouns which were derived from the same base, but with different Germanic suffixes, and which share the same meaning, for example goodness and goodhood. The suffixes in question are -dom, -hood, -ness and -ship, well known from words such as freedom and fellowship, but also the less common -lac (together with its now lost Scandinavian cognate -leikr) and -rede(n), which survive today only in very few words, like wedlock and kindred.

Synonymous derivations are particularly interesting in that they can reveal more about the options available to speakers at the time and what may have influenced their choice of one suffix over the other. The choice of suffixes for a certain base could be a conscious one, and suffixes may be interchangeable without any apparent change in meaning. This is shown, for example, in the manuscript of The Ormulum, where in many derivations -ness was erased and replaced by -leikr (cp. Laing’s comments in LAEME file #301). As a matter of style, synonyms may be used to avoid repetition. This kind of variation may well be at work in those texts which contain synonym pairs without displaying any marked preference for one or the other formation. If there are small differences between the words, they are difficult to pinpoint today on the basis of the sketchy transmissions from the time. Since only about 13% of all synonyms encountered in this study occur in such constellations, other factors must have a bearing on suffix choice. This paper will investigate which suffixes competed with each other and to what extent the choice of a suffix may be influenced by contemporary developments in frequency and productivity as well as factors such as regional and text type usage.

2. Working with the Linguistic Atlas of Early Middle English

The following analyses are based on data researched with the help of the Linguistic Atlas of Early Middle English (LAEME), which contains texts written between c. 1150 and 1350. The corpus encompasses almost 650,000 words in 167 text entries, which is several hundred thousand words more than the respective sections of the Helsinki Corpus (c. 210,000) or the Penn-Helsinki Parsed Corpus of Middle English (c. 370,000). In order to facilitate the study of changes within Early Middle English, the two centuries covered by the corpus were divided into five sub-periods of about 40 years each. To that end the texts were grouped according to their approximate date of origin published by Laing (2008) in the Index of Sources. Because of the nature of the datings, the sub-periods overlap to some degree. For example, if a text was produced in a certain year (say c. 1200), it can be allocated to the corresponding sub-period. If, however, a span of 25, 50 or 100 years is indicated for the text production, the text could form part of two sub-periods. In such cases, the texts are sorted into sub-periods which cover the majority of the time span indicated for their production. A text written sometime between the last quarter of the twelfth and the first quarter of the thirteenth century would therefore be included in sub-period II, i.e. 1190–1230.

Considering that the present lexical investigation yields a relatively small amount of data, larger geographical areas are useful to work with, rather than the more specific locations determined for the majority of texts (e.g. Ludlow, South Shropshire for one version of Ancrene Wisse, or North West Essex for the Trinity Homilies), in order to make the data more tangible. In addition to observing language use in different counties, the following larger regions are of interest which approximate and are named after traditional dialect regions yet roughly follow county borders: [1]

• the North (N): texts extant from Cumberland, Durham, Yorkshire;
• the East Midlands (EML): Cambridgeshire, Essex, Huntingdonshire, Leicestershire, Lincolnshire, Middlesex, Norfolk, Northamptonshire, Suffolk;
• the West Midlands (WML): Cheshire, Gloucestershire (parts), Herefordshire, Lancashire, Shropshire, Staffordshire, Warwickshire, Worcestershire;
• the South-West (SW): Berkshire, Devon, Dorset, Gloucestershire (parts), Hampshire, Oxfordshire, Somerset, Wiltshire;
• and the South-East (SE): Kent, Surrey, Sussex.

Since the Linguistic Atlas does not provide information on text types, the necessary details were gathered from the New Index of Middle English Verse (Boffey & Edwards 2005) and the volumes of the Manual of Middle English Writings (Severs, Hartung & Beidler 1967–), which list texts according to author or subject. From this basis larger text type groups were devised, similar to the prototypical text categories proposed by the Helsinki Corpus (Kytö 1996). The new corpus thus features

• fictional texts, containing tales and romances (21,511 words),
• historical writings (82,670 w.),
• official records (17,333 w.),
• secular learning, only represented by Ælfric’s Grammar and Glossary (15,907 w.),
• secular (and mixed) verse (14,110 w.),
• biblical writings, including translations, paraphrases and commentaries (20,368 w.),
• and religious texts (433,553 w.). All remaining texts touching on a religious, philosophical or ethical subject are assembled under this heading. Among the subcategories figure sermons and homilies, religious instruction, saint’s legends, and various smaller texts like prayers, hymns and religious verse.

Corpus texts which contain elements from differing text groups are joined in a mixed category (43,347 w.).

Owing to the limited number of surviving Early Middle English sources, the various text categories and regions are not equally well represented, especially if the period is further subdivided into smaller sections. Only historical and religious writings are relatively well attested, and there are significant gaps in the coverage of most regions, particularly of the North:

A coherent comparison of text types across time, or even across time and space, is consequently almost impossible to accomplish. Working with larger text groups circumvents this problem to some degree. For religious writings, a comparatively large group (cf. Table 3), it is for example possible to contrast texts from the East Midlands with West Midland sources across the first four sub-periods. Smaller-scale investigations on subcategories could be conducted in a new study should additional information be required.

As an online resource, LAEME greatly facilitates research on word formation as suffixes are individually tagged and searchable. After entering a selected tag in a search window (e.g. ness or hood; a complete list of relevant tag forms or so-called lexels can be found online), a list will be generated which contains all occurrences of this suffix and accounts for all possible spelling variants. However, the results of such a search only refer to the number of occurrences of a suffix per text entry, and not to the words containing the suffixes. These words either have to be culled manually from the text files linked in the results page or an additional wildcard search has to be performed for words ending in the suffix in question, after which these results need to be checked against the previous ones, so that a separate list with the required data can be drawn up. In the discussion of words found in the corpus, their tagged forms (the lexical entry or lexel) given by the corpus will be used (in this case highness) instead of one of the original Middle English forms (like hehnesse).

The results of suffix searches generally have to be examined very carefully since the fact that a suffix is tagged as such does not necessarily mean that the word containing this suffix is a derivation. To give one example, a word like unworship could either be a suffixation with -ship added to the base unworth, or be constructed of worship prefixed by un-. As a prefixation according to the Oxford English Dictionary Online (OED), unworship does not reveal any information about the current use of the suffix it contains and is therefore not included in the data. Furthermore, words like rihtdom and wo:hdom were discarded from the list because they are examples of compounding: with (‘just’ or ‘unjust’) ‘judgment’ they retain a semantic layer of the Old English lexeme dōm, which is no longer present in the suffix -dom. These examples show that the tagging in LAEME merely indicates the fact that a word contains one or more elements which assume the shape of a prefix or suffix; the tagging does not give any further etymological information which would help determine whether this word is a prefixation, a suffixation or a compound.

After all non-derivatives were weeded out, about 450 different words remained, out of which 140 lexemes forming 61 sets of synonyms could be isolated (with a total of 1,352 attestations in the corpus). Only those words are accepted as synonyms which contain the same base but a different suffix, and which have the same meaning according to the Middle English Dictionary and the Oxford English Dictionary Online. Since neither of the dictionaries lists all instances of an entry, the meaning of a number of derivations had to be individually interpreted on the basis of context. For future reference the word class and the language origin of each base was noted alongside the period of the word’s first attestation, Old or Middle English. These proceedings eliminated some ostensible synonyms in the corpus, which despite their formal similarity do not share the same meaning. For example, while softhood refers to physical softness, all instances of ness-formations relate to character traits and other extended meanings. Wisdom and worship were also not considered in this study because of the overwhelming frequency of the words inherited from Old English. The new, partially synonymous formations in -hood do not offer serious competition in either case, with one and three instances against 201 and 132, respectively.

The suffix -hood was represented in Middle English by two forms: one is the continuation of Old English -hād, later developing via [hɔd] to -hood; the other is the related -hed(e), whose precise origin is still a matter of debate and which was later replaced by -hood forms (see Dietz 2007: 134-137, Ciszek 2008: 56-58, and the entry on -head in the OED). Today the two suffixal forms are only distinguished in a few words, such as godhead and godhood or maidenhead and maidenhood. In the Linguistic Atlas both forms are tagged indiscriminately as -hood. The following analyses will refer to this more general -hood, unless stated otherwise.

3. Synonymous derivatives in Early Middle English

The first point of interest concerning synonyms is the frequency with which the different suffixes occur in the five sub-periods, displayed in Table 4.1 and Table 4.2.

Early on -ness is the most frequent suffix, yet it loses some of its stronghold in the middle of the thirteenth century (normalised frequencies drop from over 20 to below 10); -ship also occurs fairly frequently, but declines even more strongly than -ness (to a frequency of c. 1) after a short surge in sub-period III. If another suffix is somehow responsible for this, it cannot be -dom with its relatively constant, yet infrequent rate of occurrence (gravitating around 1); nor can -lac or -leikr assume this role, seeing that they also show limited usage and fall more or less out of use in the last two sub-periods. Infrequent -rede(n) is not a suitable candidate either since its usage increases only minimally (staying below a frequency of 1) and would not even reach such an unprecedented height in the last sub-period (1.73) if it was not for one text sporting fellowrede(n) (compare fellowship) 24 times. The reason may ultimately be found in -hood: after a slow start, derivations in -hood, specifically those in -hed(e), increase steeply in sub-periods IV and V (from below 1 to 3.02 and 7.99, respectively), at the same time that -ness and -ship (as well as -lac and -leikr) lose in currency. [2]

Examining which suffixes compete with each other in synonym pairs or groups will help determine whether these developments are indeed correlated. For that purpose all words were disregarded in each sub-period for which there are no potential synonymous counterparts: words inherited from Old English (like gladness), which could theoretically appear throughout the time covered by the corpus, were regarded as potential synonyms in all sub-periods even if they do not occur. In contrast, words which are of Early Middle English origin and not found until, say, sub-period IV (like sikerhood ‘security’), were not considered until their first attestation. It was then possible to isolate various rivalling suffix pairs for each sub-period, among which -hood and -ness, -hood and -ship as well as -ness and -ship figure most prominently.

The ratios of formations in -ship against those in -hood and -ness in each sub-period show that -ship loses its upper hand with respect to both suffixes in the last two sub-periods (Table 5). [3] Taking into account that in the -ship/-ness pair one word alone (wareship, meaning ‘vigilance’) with 46 instances is responsible for almost half the attestations of -ship in the third sub-period, the replacement process involving -ness could be argued to begin already some 40 years earlier.

While -ness and -hood were quickly favoured over -ship, the replacement of -ness by -hood is a more gradual process, until in the early fourteenth century both suffixes find themselves on roughly equal footing (as evidenced in Table 6). Towards the closure of Early Middle English the suffix -hood also asserts itself in its competition with -dom, [4] while -rede(n) never poses a challenge, brotherrede(n) (two attestations in sub-period II) and brotherhood (one instance in sub-period V) constituting the only example for this suffix pairing.

It appears that the multiplication of -hood derivatives is indeed to a great part responsible for the decline of synonyms formed with -ness and -ship. With regard to the remaining suffixes -ness enjoys a strong position, being generally more frequent than -dom (consider the longevity of holiness as apposed to holidom) and replacing -lac and -leikr in derivatives of, for example, true ‘faithfulness’, good and hende ‘nobility, virtue’ towards the end of the period (Table 7).

In contrast, -ship begins to lose its footing in the middle of the thirteenth century not only with regard to -hood and -ness, but also in pairs like freedom/freeship and thraldom/-ship ‘subjection’, as well as foeship/-rede(n) ‘enmity’ and even fellowrede(n)/fellowship. The latter will likely be a short-lived submission, in the same way that -ness is confronted with a small number of -rede(n) formations on the bases love and sib meaning ‘love’ and ‘kinship’ (Table 8).

The changing preferences within synonymous pairings can be suitably illustrated with derivatives of fair with the meaning ‘beauty’. In the present context, most words carried over from Old to Early Middle English, like freedom, are likely to continue to be used later on. This is also true of fairness, occurring throughout the period. As anticipated it is more frequent than the newly formed fairship and fairleikr, neither of which are attested after sub-period III, while fairhood, another new formation, makes its first appearance in sub-period IV and dominates sub-period V at the expense of fairness (see Table 9).

New derivations among synonyms formed in Early Middle English are valuable indicators for the productivity of a suffix. At the same time, the developments in productivity are closely related to those concerning the frequency of and competition between the suffixes. By the end of the period, there are only two suffixes left which display marked productivity – not surprisingly -hood and -ness (see Table 10). While the latter shows a fairly constant rate of productivity, neologisms in -hood (here again those in -hed(e)) rise steeply. Although -ship can boast as many new formations as -ness, its productivity comes to a standstill after the third sub-period, similarly so that of -lac and -leikr. Neither -dom nor -rede(n) can be considered very productive, suggesting that neither suffix appears on a varied set of bases but rather in a small number of more frequent derivatives. The previously mentioned freedom and fellowrede(n) corroborate this, accounting for 37 and as much as 83 percent of all -dom and -rede(n) attestations, respectively.

As a final point three examples will be presented which highlight the importance of investigating suffix usage across text types and/or space. Firstly, a pilot study (Gardner forthcoming) on suffixation in Old English showed that already then derivations in -lac were of limited frequency and productivity, and in Late West Saxon were mainly used in religious writings. These findings led to the conclusion that in subsequent centuries such formations are most likely to appear in religious contexts; the present study confirms that it is indeed the case in Early Middle English where it is employed predominantly in texts from the so-called Katherine Group. Secondly, regarding regional usage, the last three attestations of -leikr in sub-periods IV and V all occur in texts which show connections to the former Danelaw area, in which Scandinavian influence was particularly strong – the regions east and north of a line roughly stretching from London to Chester (see also Holman 2001: 5-6). Havelok originates from Norfolk, the Northern Homily Collection from Yorkshire, and Dame Sirith is known, according to Laing (2008), to exhibit influences from the north-eastern Midlands, all part of the former Danelaw. Lastly, linguistic features do not always develop homogeneously in religious texts, but may vary in different regions. This becomes particularly evident with regard to the suffixes -hed(e) (shown to increase strongly from sub-period IV) and -ship (decreasing after sub-period III). West Midland texts hold on to -ship formations until sub-period IV, whereas in the East Midlands there is already a marked drop after the first sub-period (see Table 11 for absolute values, Figure 1 for normalised frequencies). [5] The apparent return of derivations in -ship in this region in sub-period IV seems less significant in view of the fact that in sub-period V such words do not occur in the local texts.

Although -hed(e) forms begin to appear in the West Midlands in sub-period II and increase slowly, they remain few in number. Conversely, East Midland religious writings show a strong prevalence for such derivatives from sub-period III onwards (cf. Figure 2; absolute frequencies are given in Table 11). Consequently it can be argued that texts from the West Midlands lag behind observable suffix developments, whereas East Midland writings seem to propagate these changes. [6]

4. Conclusion

To conclude, by the end of the Early Middle English period -hood was the most frequent as well as the most productive suffix, after -ness led the group in the first 80 years. The suffix -ship enjoyed some degree of popularity until sub-period III, while the remaining forms only played minor roles. Some statements made by previous scholars need to be amended in the light of new data drawn from the Linguistic Atlas of Early Middle English. Dalton-Puffer (1996: 129), for example, claims on the basis of the smaller Helsinki Corpus that -rede(n) and -lac (this includes -leikr) ‘cannot be said to compete with anything after 1250’ as they do not occur afterwards, yet according to the present data some competition still existed after that date, even if on a small level. Furthermore she states that the doublets in general ‘occur practically exclusively with de-adjectival formations’ (Dalton-Puffer 1996:126), which is contradicted by the data offered by LAEME: over 10% of the 61 synonym sets isolated in this study contain in fact a noun base (such as brother, fellow, foe, god, knight, lord, man, thral). Discussing dialectal distribution, Ciszek recently merely noted that -hed(e) and -ship occur in all areas, reserving judgment on the scarcely represented North (Ciszek 2008: 69, 107). While not technically wrong, such statements leave some revealing developments, like those in East and West Midland religious writings, hidden from view. The results of the present study underline the merit of investigating the factors of time, space and text type. Together with the assistance of recent and improved research tools such as LAEME with its intricate tagging system and easy-to-use search function, such an approach adds detail to previous scholarship, helping uncover new insights into historical topics.

Notes

Electronic references

Helsinki Corpus of English Texts: Diachronic and Dialectal. Compiled by Matti Rissanen, Merja Kytö, Leena Kahlas-Tarkka, Matti Kilpiö, Saara Nevanlinna, Irma Taavitsainen, Terttu Nevalainen, and Helena Raumolin-Brunberg. 1991. Helsinki: Department of English, University of Helsinki. http://www.helsinki.fi/varieng/CoRD/corpora/HelsinkiCorpus/index.html.

A Linguistic Atlas of Early Middle English 1150–1325 (LAEME). Compiled by Margaret Laing and Roger Lass (Edinburgh: © 2007– The University of Edinburgh). http://www.lel.ed.ac.uk/ihd/laeme2/laeme2.html.

Middle English Dictionary (MED). http://quod.lib.umich.edu/m/med/.

Oxford English Dictionary Online (OED). Oxford: Oxford University Press. http://www.oed.com/.

Penn-Helsinki Parsed Corpus of Middle English. Compiled by Anthony Kroch and Ann Taylor, 2000. https://www.ling.upenn.edu/hist-corpora/PPCME2-RELEASE-4/index.html.

References

Boffey, Julia & Anthony S. G. Edwards. 2005. A New Index of Middle English Verse. London: The British Library.

Ciszek, Ewa. 2008. Word Derivation in Early Middle English. Frankfurt am Main: Peter Lang.

Dalton-Puffer, Christiane. 1996. The French Influence on Middle English Morphology. A Corpus-Based Study of Derivation. Berlin, New York: Mouton de Gruyter.

Dietz, Klaus. 2007. “Denominale Abstraktbildungen des Altenglischen: die Wortbildung der Abstrakta auf -dōm, -hād, -lāc, -rǣden, -sceaft, -stæf und -wist und ihrer Entsprechungen im Althochdeutschen und im Altnordischen”. Beiträge zur Morphologie: Germanisch, Baltisch, Ostseefinnisch, ed. by Hans Fix, 97-172. Odense: North-Western European Language Evolution.

Gardner, Anne. Forthcoming. “Abstract noun ‘suffixes’ and text type in Old English.” Papers from the 15th International Conference on English Historical Linguistics (15 ICEHL), München, 24-30 August 2008, ed. by Hans Sauer, Ursula Lenker & Judith Huber. Amsterdam & Philadelphia: John Benjamins.

Holman, Katherine. 2001. “Defining the Danelaw.” Vikings and the Danelaw. Select Papers from the Proceedings of the Thirteenth Viking Congress, ed. by James Graham-Campbell, Richard Hall, Judith Jesch & David N. Parsons, 1-11. Oxford: Oxbow Books.

Kytö, Merja. 1996. Manual to the Diachronic Part of the Helsinki Corpus of English Texts. Coding Conventions and Lists of Source Texts. Helsinki: Department of English, University of Helsinki. http://clu.uni.no/icame/manuals/HC/INDEX.HTM.

Laing, Margaret. 2008. A Linguistic Atlas of Early Middle English. Index of Sources. http://www.lel.ed.ac.uk/ihd/laeme1/laeme1.html.

Mills, A. D. 1998. A Dictionary of English Place-Names. Oxford: Oxford University Press.

Mossé, Fernand. 1977. Mittelenglische Kurzgrammatik. Lautlehre, Formenlehre, Syntax, translated by Herbert Pilch & Ursula Siewert. München: Max Hueber.

Severs, Jonathan Burke (Vols. 1-2), Albert E. Hartung (Vols. 3-10) & Peter G. Beidler (Vol. 11-), eds. 1967–. A Manual of the Writings in Middle English 1050–1500. Hamden, Conn.: Archon Books.

Trips, Carola. 2009. Lexical Semantics and Diachronic Morphology. The Development of -hood, -dom and -ship in the History of English. Tübingen: Max Niemeyer.

Zbierska-Sawala, Anna. 1993. Early Middle English Word Formation. Semantic Aspects of Derivational Affixation in the AB Language. Frankfurt am Main: Peter Lang.