Normalisation 3: grammar
examples of the levelling of anomalies within eME paradigms
The following is background information. It is not required reading for these who wish to learn Early Middle English. If you're looking for a description of Early Middle English grammar, see eME language.
The sections below examine the application of principles 3 & 4 which are outlined at Normalisation 1: principles. Frequent use is made of the abbreviations and definitions listed there.
Principles 3 & 4 are as follows:
- stem alternation and other complexities or anomalies within a given paradigm (noun, verb etc), are excluded from the core eME grammar when evidence of levelling exists in that paradigm in a text dated before 1250;
- where widespread levelling is observed within the paradigms of a distinct group of words with shared characteristics, the same levelling can be applied to all remaining paradigms in that group.
The upshot is that within certain paradigms inherited from OE, items are either merged or jettisoned and anomalies are levelled. The result is a significant simplification of the core eME grammar. You'll find examples below.
But first - levelling in late OE
There are two main threads that run through the story of Middle English (ME) - the simplification of the grammar inherited from Old English (OE) and the addition of vocabulary borrowed from French and Latin.
These two developments were not simultaneous. Much of the simplification of OE's grammar had taken place, in the Northumbrian dialect, and in the eastern part of the Mercian dialect area, before the end of the OE era. The influx of French and Latin words came later - a trickle at first before the mid-thirteenth century, and then a flood in the fourteeth and fifteenth centuries.
The grammatical innovations that leap from the pages of PC2 and Orm were mainly in the realm of nouns, adjectives and demonstratives. Three grammatical genders had become one. Ten classes of noun declensions (not counting sub-groups) had become one. Four noun cases had been reduced to two - the nominative (subject) and the genitive (possessive) and eight separate noun case forms had been whittled down to two, since the two remaining plural case endings were now identical to the genitive singular - -es.
Simplification to the verb system wasn't quite as dramatic by the end of the OE period, but was evident none the less.
The simplification of OE grammar, by the time it reaches eME, is essentially levelling within paradigms, and is threefold:
- a reduction in the number of items within paradigms (mainly noun and adjective);
- a reduction in the number of classes of paradigms (to a single class for both nouns and weak verbs);
- a reduction in anomalies within paradigms.
The levelling that had taken place in late OE, is evident in both PC2 and Orm and is consistently applied. It is not dealt with here. For more detail, see simplification of noun paradigms and simplification of verb paradigms.
Well begun is half done
Or rather - in the case of levelling and the core grammar of eME, well begun is enough.
Some levelling within paradigms is underway in eME. This is the levelling discussed below. In short, any levelling encountered in the early Middle English period, is incorporated in the core grammar of eME in this site, even if inconsistently applied. By the same token any elements of OE grammar that are retained sporadically, pass into the optional grammar of eME.
To illustrate, the language of Orm is in transition. Traces of dual stems in some nouns and adjectives, can still be seen. Traces of dual stems are also evident in the present tense forms of certain verbs. However, and this is key - those traces are occasional and inconsistent. For every pair of tre trewwess Orm has tre tres (tree trees), and for every habbenn hafet`t` Orm has hafenn hafet`t` (have has). The anomalies trewes, trewe, habbe, habben are cut from the core grammar but pass into the optional grammar of eME.
Here's how the two pairs of paradigms line up:
In Normalisation 1: principles I stated that "the Early Middle English used in englesaxe should be as simple and practical as possible for the learner" and "should reflect the broad patterns of development which led from Old English to Modern English". On that basis I have taken any reasonable opportunity to level within paradigms, provided those simplifications are supported by pre-1250 usage.
The optional grammatical features mentioned above and outlined at optional grammar, may appear in normalised versions of OE and ME texts, and in the weblog of this site, but do not appear in book 1 of Anglo-Saxon without tears (the first 800 words).
Keep in mind that the grammar of the normalised eME in this site is based on the East Midland dialect alone. Generally then, only PC2, Orm, SO and Ch forms are considered. West Midland and Southern texts lend no weight in deciding whether or not a conservative feature, inherited from OE, shoud be retained in the normalised eME of this site, though they may add weight to linguistic innovation which draws eME closer to ModE.
While Normalisation 2: additional eME forms deals with dictionary entry forms - the nominative of nouns, the infinitive of verbs etc, this page focuses on the other forms, i.e. changes to forms within noun, adjective and verb paradigms, due to case, number, strong/weak status, person, tense and mood i.a.
Some additional eME forms permit a simplification of verb conjugations, since it is the infinitive which is the anomalous form. An example is sec`g`an, saeg`t` > say%en, say%t` (to say, says). The same is true of certain nouns descended from OE wo and wa stems where the nominative is the anomalous form without a <w>, e.g. sinu, sinwa > sinwe, sinwes (sinew, sinews). See below as well as simplification of verb paradigms and simplification of noun paradigms for more detail.
Applying principles 3 & 4
The key, particularly in regard to the early stages of learning a language, is to offer the simplest route. This is where principles 3 & 4 come into play. Where pre-1250 examples permit, stem levelling can be applied throughout a paradigm, and then throughout the entire group of nouns or verbs which share the characteristics of that paradigm.
Two examples - simplifying paradigms with dual stems
OE wa- and wo-stems have a dual stem. Inflected forms have a <w> at the end of the stem, while the nom sg (and acc sg) does not. We can apply principles 3 & 4 to the reflexes of these nouns to remove that stem alternation in the core grammar of eME. The approach is as follows:
- we start with a noun paradigm that exhibits stem alternation, e.g. bale nom sg balwes gen/pl or scade nom sg scadwes gen/pl;
- in a pre-1250 text we find a form which removes the stem alternation; this could be either an inflected form which drops the <w> or a nom sg which includes the <w>; e.g. balwes pl > bales pl and scade nom sg > scadwe nom sg;
- via principle 3 the stem levelling of one form (e.g. pl) is extended to the remaining form (e.g. gen sg) within the paradigm; so bales pl is extended to bales gen sg;
- courtesy of principle 4, the paradigm levelling of the majority of wa- and wo- stems can be extended to the remaining words in that group; e.g. the bale bales pattern is extended to mele meles and the scadwe scadwes pattern to sinwe sinwes;
Thus we have:
- pre-1250 forms which show stem levelling:
- inflected forms which drop <w>: bales pl, smeres gen, tres pl, cnes pl, ter dat, leohe dat;
nom sg forms which add <w>: barewe, t`eww, schadwe, mae_dwa m
- principle 3 - extend stem levelling throughout paradigm:
- bale bales, smere smeres, tre__ tre__s, cne__ cne__s, tere teres, le__ le__s, barwe barwes, t`ew t`ewes, scadwe scadwes, maedwe maedwes
- principle 4 - extend pattern of simplified paradigm to entire group:
- mele meles, sinwe sinwes, straw strawes, laeswe laeswes, sarwe sarwess, badwe badwess
The upshot is that the reflexes of OE wa- and wo-stems have a single stem in the core grammar of eME.
OE masculine a-stem nouns ending in <h> are a similar group with a dual stem. The nom sg and acc sg forms have a final <h>, while all other forms lose that <h> when the inflectional ending is added. We can apply principles 3 & 4 to the reflexes of these nouns to remove that stem alternation in the core grammar of eME. The approach is as follows:
- we start with a noun paradigm that exhibits stem alternation, e.g. walh nom sg wales gen/pl or wo__h nom sg wo__s gen/pl;
- in a pre-1250 text we find a form which removes the stem alternation; this could be either a nom/acc sg which drops the <h> or an inflected form which includes the <h> (or <g> etc); e.g. walh nom sg > wale nom sg and wo__ dat > wo__ge dat;
- via principle 3 the stem levelling of one inflected form (e.g. dat) is extended to the remaining inflected form(s) (i.e. gen/pl) within the paradigm; so wo__ge dat is extended to wo__ges gen/pl;
- courtesy of principle 4, the paradigm levelling of the majority of a-stems with final <h> can be extended to the remaining words in that group; e.g. the wale wales pattern is extended to sele seles and the wo__h wo__ges pattern to farh farges;
Thus we have:
- pre-1250 forms which show stem levelling:
- nom/acc sg forms which drop <h>: wale, mare, fere, feo, sho;
inflected forms which add <h> (or <g> etc): slo_ges gen, wo_ges gen, wo_ge dat, salhas pl, saligum dat pl, eolces gen, woge dat, wohe dat etc
- principle 3 - extend stem levelling throughout paradigm:
- wale wales, mare mares, fere feres, fe__ fe__s, sco__ sco__s, slo__h slo__ges, wo__h wo__ges, elk elkes, salh salges
- principle 4 - extend pattern of simplified paradigm to entire group of nouns:
- sele seles, halh halges, farh fargess, alh algess, eh egess, horh horgess
The upshot is that the reflexes of OE a-stems ending in <h> have a single stem in the core grammar of eME.
See levelling in all pre-1250 sources (below) for more detail.
Examples of levelling underway in eME
The following grammatical features, were subject to levelling in the eME period:
- vowel alternation a/ae in preterite forms of strong verbs, classes 4 & 5 - y%af y%aefen
- stem alternation (with or without <w>) in reflexes of wa- and wo_-stems - bare barwes
- stem alternation in reflexes of a-stems ending in h - walh wales
- dative case marker -e for nouns ending in a consonant - tu__n tu__ne
- accusative form for indefinite article - a__n aenne
- dropping of unstressed penultimate syllable in genitive and plural - de__fel de__fles
- genitive forms without the -(e)s marker
The ensuing case studies provide some detail around the levelling of each feature. They are grouped according to the principle that comes into play, to remove the feature from the core grammar of eME.
Levelling underway in PC2 and Orm
This includes levelling found in either PC2 or Orm but not the other.
vowel alternation in the preterite forms of strong verbs, classes 4 & 5
This is a feature of OE. It is observed consistently in Orm, which has a for sg 1/3 and ae for the plural, e.g. sahh saey%henn, satt saetenn, spacc spaekenn, y%aff y%aefenn, but not in PC2 which usually displays the same vowel a for plural preterite forms, e.g. forbaren, namen, stali[n], drapen, iauen.
Core eME: single vowel - a in preterite of strong verbs, classes 4 & 5 - bar baren, y%af y%afen
dative case marker -e for nouns ending in a consonant
This is observed inconsistently in PC2, which has e.g. both in tune and in t`e hus.
Core eME: no dative marker for nouns - in tu__n
accusative form for indefinite article
This is observed inconsistently in Orm, which has both an and aenne for the accusative, e.g. Illc mann an peninng (m acc) y%aefe but Y`e shulenn findenn aenne child (nt acc). PC2 displays only an for the accusative.
Core eME: no separate accusative form for a__n
dropping of unstressed penultimate syllable in inflected forms of nouns and adjectives
This is a feature of OE. Unstressed penultimate syllables were often dropped in the inflected forms of nouns and adjectives before n, r and l, e.g. OE dryhten/dryhtnes, mic`el/micle. According to Wright 2, the unstressed vowel "regularly disappeared in open syllables when the stem syllable was long", though fugol, with short stem vowel, was one of a group declined like engel engles 3. The syncope can still be found in Orm - deofell/deofless, mikkell/miccle, and in PC2 - od`er/od`re, castel/castles, minster/minstre, gysles. On the other hand, we see PC1 fugeles, PC2 drihtines, micele, AW deoueles, fuheles, and Lmn & Owl fuy%eles, to give just a few examples.
Core eME: the unstressed penultimate syllable is retained in inflected forms of nouns and adjectives - casteles, de__feles, drihtenes, fugeles, micele, minstere, o__t`ere, y%i__seles etc
genitive forms without the -(e)s marker
This is a relic of two different OE noun classes - the feminine o_-stems of the strong declension, in which the genitive marker was a final e, e.g. sa_wol sa_wle, and the stems in -r of the minor declension, in which a genitive marker was absent, e.g. mo_dor mo_dor, and similarly for faeder, bro_d`or, dohtor, sweostor. The former is observed inconsistently in Orm, which has both sawle and sawless for the genitive.
For the reflexes of the stems in -r, we need to enlist principle 4. Against Orm's moderr, brot`err (gen) we can counterbalance moderes (Genesis and Exodus 1250) and faderes (Poema Morale 1175).
Ch has an occasional "zero" marker for the genitive in reflexes of a couple of OE weak declension nouns: hlae_fdig`e hlae_fdig`an, heorte heortan > lady lady, herte herte. Orm on the other hand, consistently has he(o)rrtess as the genitive.
Core eME: all genitive forms have the -(e)s marker
Levelling underway in all pre-1250 sources
This section includes examples from PC2 and Orm, the secondary eME sources and other pre-1250 sources found in the MED, as well as variant OE forms.
stem alternation (with or without <w>) in reflexes of wa- and wo_-stems
In OE feminine wo_-stems, a <w> appears at the end of the stem, before the case marker, throughout the paradigm, except in the nominative singular, which has a final <u>, or less often - no final vowel. The wa_-stems behave similarly, but with the expected differences characteristic of masculine and neuter strong noun declensions.
It's interesting to note that of the wa- and wo_-stem nouns that have survived to ModE, the reflexes of feminine mouns tend to retain the <w> - sinew, shadow, meadow, leasow (pasture), while the reflexes of neuter nouns do not - meal, smear, bale (harm). This is not surprising, since in the OE neuter nouns, four case forms - the nominative and accusative, both singular and plural, had no <w> in the stem, while for the feminine, only the nominative singular lacked the <w>.
Clark Hall 4 and the MED provides the following examples of levelling within paradigms, either with or without <w>, in texts dated before 1250:
- neuter wa- stems - genitive or dative without <w>: 1211 smeres gen, Orm tre dat, AW leohe dat, 1250 ter dat; plural in -es without <w>: 1250 bales, Orm cnes, tres, 1225 knes;
- masculine wa-stems - nominative or accusative sg with <w>: 1197 barewe (grove), 1242 berwe, Orm t`eww
- feminine wo-stems - nominative sg, with <w> - AW schadewe, TH 1200 shadewe, 1225 scadewe, 1250 schadwe, OE alt mae_dwa m sg; [and possiby: 1240 Hardemedwe (name), 1236 Smythesleswe (name)]
Core eME: no stem alternation in the reflexes of wa- and wo_-stems - bale bales, smere smeres, tere teres, barwe barwes, tre__ tre__s, cne__ cne__s, le__ le__s, t`ew t`ewes, scadwe scadwes, maedwe maedwes, laeswe laeswes
See simplification of OE noun paradigms: wa- and wo-stems for more detail.
stem alternation (with or without <h>) in reflexes of a-stems with final <h>
In OE masculine a-stems with final <h>, the <h> appears at the end of the stem in the nominative and accusative singular only. Throughout the rest of the paradigm, the <h> is missing and the preceding vowel, if short, is lengthened.
Clark Hall 4 and the MED provide the following examples of levelling within paradigms, either with or without <h/g/y%>, in texts dated before 1250:
- nom/acc sg without <h>: OE alt we_al, wal-, 1225 wale; OE alt mear, AW meare, 1200 mare; OE alt seol; OE alt fe_or, 1150 PA fere acc; OE alt fe_o, Lmn feo; OE alt sco_, Orm sho, 1200 TH sho;
- oblique forms (gen/dat/pl) with <h/g/y%>: OE alt slo_ges gen; OE alt wo_ges gen, wo_ge dat, Orm woy%h%e dat, PC1 woge dat, Lmn woy%e dat, AW wohe dat, Owl woy%e dat, 1200 wohy%e dat; OE alt eolc, eolces gen; OE alt salhas pl, saligum pl dat.
Core eME: no stem alternation in the reflexes of a-stems with final <h> - wale wales, mare mares, fere feres, fe__ fe__s, sco__ sco__s, slo__h slo__ges, wo__h wo__ges, elk elkes, salh salges, sele seles, halh halges, farh farges, alh alges, eh eges, horh horges
See simplification of OE noun paradigms: a-stem nouns ending in <h> for more detail.
Extending the pattern
The last of the five basic principles states that "anomalies within a given noun or verb paradigm can be levelled when evidence of levelling exists in that paradigm, or in a similar paradigm, in at least one eME source (PC2, Orm, Lmn, AW, Owl, SO, Ch)."
This principle, which we could call "extending the pattern", is used to decide whether the simplification of a particular noun or verb paradigm is appropriate in eME. To illustrate - OE has several verbs with dual stems, in which the stem of the first person, plural and infinitive differs (albeit slightly) from the stem of the other forms. One example is lecgan, with present singular forms lecge leg`st leg`t` and plural form lecga`t`. In the eME reflexes of such verbs, if any of the eME sources, has a first person singular, plural or infinitive form with a stem which matches the stem of the third person singular, we can apply that simplification across the paradigm for that verb, in eME. That is in fact the case for eME ley%en (or leggen), since Ch has the infinitive leye (as well as leggen). Hence the eME paradigm can be ley%e ley%est ley%et` ley%en. Another example is herien, in which the i extends across the paradigm, so that OE herian inf herode pt sg3 becomes eME herien inf heriede pt sg3. Ch provides the key support for the extension of the i to the preterite and present tense singular, although such forms are found as early as Lmn. See simplification of verb paradigms in eME for more detail and similar examples.
Note that even without that support in Ch, we could have extended the pattern to ley%en from say%en, which is a similar paradigm, and which displays levelling in PC2.
The role of Chaucer and Modern English in normalising eME
The Chaucer Test
All the eME sources listed above are from the 12th or 13th century. Ch alone comes after. He started writing around 1370 - three quarters of a century into the second half of the Middle English period. Why then, consider Ch forms? Because Ch is a crucial staging post in the development of OE to ModE. To many, Chaucer is synonymous with Middle English. His is a substantial body of work (of great literary merit) which gives us invaluable clues about the syntactical, semantic and phonetic development of English in the East Midlands prior to 1400 (and by a slight extrapolation - prior to 1300).
The Chaucer Test is used to decide whether to adopt or reject borderline grammatical features in eME. If traces of an OE grammatical feature are common (even if not predominant) in Ch, then that feature is retained in eME. Here are some examples:
- eME retains a distinction between singular and plural in the vowel of the 3rd person preterite forms of 4th and 5th class strong verbs - y%af s vs y%ae_fen pl, because this distinction remains to some extent in Ch - yaf s vs yeve pl; (note however that yaven pl is also found in Ch);
- eME lacks a genitive marker with family member nouns - mi_n fader hu_s, t`i_n mo_der brot`, hire suster bo_k etc, because the genitive marker es was often absent in these nouns in Ch;
- eME retains a distinction between short and long consonants, because this is reflected in Ch's verse - sone rhymes only with wone, which in OE had a single n, whereas sonne rhymes with yronne bigonne wonne, all of which had nn in OE);
- eME has a single prefix corresponding to OE g`e-; Orm has two reflexes of the OE prefix g`e- : y%e- for verbs and i- for adverbs and adjectives. PC2 has only one example of each, and again, they differ: ge- and o- /@`/?, respectively. Given that there was a single prefix in OE, it would seem needlessly complex not to continue that in eME. Two prefixes would also be inconsistent with the majority of eME texts, and Chaucer. The evidence points to a single pronunciation. Thus eME has a single prefix - i, which is found in Orm (and most eME sources).
- eME has no stem alternation in nouns; some OE nouns and verbs had a w or h which appeared in certain forms, e.g singular genitive and dative and all plurals, but not in others, e.g the singular nominative. Ch and ModE leave that complexity behind. w or h is either missing in all forms of a particular word, or included in all its forms1. Examples are smeoru smeorwes > smere smeres (smear) and sinu sinwes > sinwe sinwes (sinew). How can you tell if the eME form has a w or not? See the ModE test (below).
- grammatical features that persist in certain words up to Chaucer's time, despite being dropped in most words of the same verb or noun class, are retained in those words in eME. Examples are ferien and herien which retain the i of OE class 2 weak verbs, which is dropped elsewhere. Furthermore, that feature is propagated throughout the paradigm in eME, as it is in Chaucer.
the ModE test
This test covers groups of words with a shared characteristic in OE, which evolve inconsistently, so that the sound change in question can't be neatly summarised (in Table X).
The ModE test is only invoked in three groups of words. The reflexes of the affected words in ME and ModE can show either of two possible forms. The question is - which form do we prefer in eME? How do we choose? In short, ModE indicates the preferred form for each word in question, provided that the sound change in question has begun to show at all within the group in the Early Middle English period.
This test is important because it means that we don't have to sift through the attested forms for each member of the group. We can apply a single test to all members of the group. And that test is a simple question - what form does ModE have?
For those familiar with OE, who prefer to use forms based on standard OE, these cases don't present a problem. But for those without that knowledge, seeking an appropriate (additional) eME form can be somewhat difficult at times, without the ModE test.
Imagine for example a ModE speaker who is looking for the eME word for flour. She has an OE dictionary (the CASD) in which she finds melu n, melwes gs. The word has two different stems. She knows from the Chaucer test (see above) that there won't be stem alternation in the eME form. There will be a single stem for singular, plural and possessive. But will it have w or not? The appropriate forms are listed in Table X, but she is off-line temporarily (perhaps on a train) and doesn't have access to it. Not does she have access to the MED, and wouldn't have time to analyse the results, even if she did. If she bases the form on the OE nominative, it is mele. If she bases it on the OE genitive form, that's melwe. Which one is preferred?
This is where the ModE test comes in. There is no w in the ModE form - meal. So, instantly she knows that the preferred eME form is mele.
The ModE test applies in three contexts only:
OE ae_ > eME ae_ or e_?
Most instances of OE ae_ become ae_ in eME and ea in ModE. But a small number become e_ and ee respectively. How do we tell? Which is the preferred form in eME for OE strae_t, for example.
If we look at the eME source texts, we see that Orm has straete dat without any counter example in PC2. In the absence of the ModE test, the exclusive eME form would be strae_t. However, OE WS strae_t is one of a group of words in which the ancestral vowel was WGmc or Lat a_ which became ae_ in OE WS, but generally e_ in the Anglian dialects, and finally ee in ModE (e.g. street). If we apply the ModE test then, we get an additional eME form - stre_t. Similarly, OE WS dae_d, slae_pan, grae_dig`, nae_dl > de_d, sle_pen, gre_diy%, ne_del, on the basis of ModE deed, sleep, greedy, needle. See Normalisation 5: issues for more detail.
with or without w/h?
The ModE reflexes of OE wa- and wo-stems sometimes have final w, and sometimes not: e.g. melu melwes > meal meal's (flour) but mae_d mae_dwe > meadow meadow's. Similarly, the ModE reflexes of OE a-stems sometimes have final gh (or w), and sometimes not: e.g. sco_h - sco_s > shoe shoes but slo_h - slo_s > slough sloughs (quagmire).
So how does a a ModE speaker know whether to include or exclude the w or h in an eME form? The answer is quite simply - look to ModE. If there is a ModE reflex without w or gh2, then leave out w or h in the eME form. Otherwise, include it. Hence - mele meles, mae_dwe mae_dwes, sco_ sco_s, slo_h slo_ges.
Note that where ModE has no reflex of an OE word, the eME form is based on the gen sg OE form (the form with the fickle w or h).
For more detail, see the simplification of verb paradigms and the simplification of noun paradigms.
vowel lengthening before consonant clusters ld, nd, ng, mb, rd
The vowel was short in OE but has since lengthened. Examples are: OE c`ild, hund, strang, climban > child, hound, strong, climb.
This development was inconsistent however. In many cases the vowel either remained short or reverted. Examples of short vowels before nd are ModE land, hand. On the other hand, both Orm and PH3 appear to have a long vowel in land and loand respectively. See normalisation 5: issues for further examples.
ModE is the guide. Since the language of departure is Modern English, and one of the aims of englesaxe is to provide a pathway to Old English, temporary sound changes in the Middle English period, have been ignored. In other words, if a vowel is short in an OE word, is indicated as long in Orm or Ch, but is short in the modern continuations of that word, it is short in eME.
If a vowel is short in an OE word which has not survived to ModE, for example - swindan, the corresponding vowel is short in eME. This is the case even when similar words - bi_nden fi_nden wi_nden, have a long vowel in ModE.
- ferien, herien > ModE ferry, harry
More on normalisation
Looking for more background on the normalisation of the East Midland dialect c 1150, in englesaxe? For an outline of the basic principles applied to the normalisation of eME word forms, together with a list of the abbreviations and definitions used, see normalisation 1: principles. For examples of the principles in action - eME forms (eME changes applied to standard OE forms) and additional eME forms, see normalisation 2: additional eME forms. For a discussion of the principles governing the eME spelling scheme used in englesaxe, see normalisation 4: spelling. For a discussion of some of the finer points, see normalisation 5: issues.
- in one exceptional case, Ch and ModE aren't in accord: Ch has both stree and straw sg and stres pl while ModE has only forms with w - straw sg and straws pl; I've listed strawe as the preferred form in eME
- h never survives to ModE in the reflexes of words with stem alternation in OE
- Clark Hall J.R., A Concise Anglo-Saxon Dictionary, Cambridge University Press, 1960
- Wright, Joseph & Wright, Elizabeth Mary, Old English Grammar, London : H. Milford, Oxford University Press, 1914, p. 96
- Ibid. p. 169
- Clark Hall Op. Cit.