An analysis of word-final -e in Middle English verse, in Gower and Chaucer*
0 Introduction The following paper offers some contribution to one of the most crucial questions concerning Middle English (henceforth ME) metrics and linguistics: the issue of word-final -e. The question has been researched both in ME alliterative poetry (Cable 1991: chapter 3) and in iambic verse (see References in this article); from as early as Tyrwhitt (1775) to the present day. The research of ME iambic poetry is centred around Chaucer: the present paper will also examine the versification of Chaucer in his CANTERBURY TALES (henceforth CT) from the point of view of final -e, but its main focus is to attempt to draw conclusions about that of CONFESSIO AMANTIS (henceforth CONFESSIO), written by Chaucer’s contemporary, John Gower. This paper, like the overwhelming majority of work examining the question of -e, maintains that when the metre shows that -e is pronounced, its pronunciation is linguistically justifiable: that is, -e can be retained or dropped, but not added arbitrarily to satisfy the metre.1 The above statement – as the article itself (like most articles) – assumes a default iambic metre for Chaucer and Gower2, in which the number of syllables in each line is the same.3 We
- Let me thank my doctoral supervisor, Dr Ádám Nádasdy (Eötvös Loránd University,
Budapest), who has been giving me unceasing academic (and often typographical) help. I am also grateful to the following people (in alphabetical order), without whose assistance this paper may not have been written: Rosamund Allen (Queen Mary, University of London) for many valuable comments (also pointing out directions towards further research); Zsófia Gyarmathy and Mitra Najafi-Gheidari (Eötvös Loránd University, Budapest) for helping to give the paper its required format, and pointing out various problems to me; Dr Attila Starčević (Eötvös Loránd University, Budapest) for additional typographical help and secondary literature; Prof László Varga (Eötvös Loránd University, Budapest) for his helpful review; Prof Robert F. Yeager (University of West Florida, Pensacola) for sending me several articles referred to in this study and discussing Chaucer and Gower with me; and all those people who helped me to access secondary literature. 1 The greatest advocate of the opposite school is Southworth (1947, 1954, 1962, 1964), who questions even the equal number of syllables in Chaucer’s lines. Robinson (1971) does not go this far, but he believes that the use of final -e is more irregular than according to the majority of scholars. 2 Throughout these pages, we will be assuming that the default was an iambic pattern both for Chaucer and for Gower. This means that in a verse line, the sequence X / (unstressed The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 2
will support the view (expressed by statistical means in Barber & Barber 1991) that -e is either retained, or its dropping can be justified: and if retained -e’s are included in the syllable count, the number of syllables in each line is (approximately) equal. The -e examined throughout this paper is by default a weak vowel (for possible exceptions see 3), realised as a schwa, and it occupies the weak position of an iambic foot. It can be of two types: A) lexical ending: OE heorte ME herte MoE heart; B) grammatical inflection: OE specan ME speke(n) MoE (to) speak. In ME, there are words borrowed from French like cite – which can be also spelt as citee (meaning ‘city’): originally, these words had stress on their final syllable. For Chaucer, these words already behaved regularly, like Germanic content words, carrying primary stress on the first syllable ([‘site] – see Minkova 2000); for Gower, the restructuring was not yet fully completed. There are reasons to believe that for Chaucer, very infrequently, these vowels could be reduced to a schwa – but in this paper, these instances are disregarded. Likewise disregarded (and tokens concerned are not included in the calculations) in the present article is the retention or dropping of line-final -e. Line-final -e does not contribute to the iambic metre: it can merely form an extra (“feminine”) syllable, so the metre does not allow any conclusion as to its retention or dropping. We have to note, however, that both Chaucer and Gower knew such poetry (French – and as regards Chaucer, Italian, too), in which the presence of an extra syllable at the end of an iambic line is undeniable. Notice also that – although certainly -e is the most frequent extra syllable – it is not only -e that can fill the line-final extra position: see, for /
X
/
X
/
X
instance, the rhyme werkis: clerkis: derk is (‘works: clerics: is dark’ – CT, II. 478, 480, 481). Throughout this article, pronounced -e’s are referred to as retained, and non-pronounced ones as dropped (whether by a metrical rule or otherwise – as discussed below). Retained -e’s will be indicated as ë, whereas dropped -e’s as e0. Since the paper examines the circumstances under which word-final -e could be dropped, it considers retention the default case. Hence, whereas the dropping of -e is always indicated (even in parts of lines where the analysed
3
syllable followed by a stressed syllable) must be repeated a given number of times. (In Chaucer’s case, we will be mostly concerned with pentameter lines; Gower wrote his English poetry – around 33 000 lines – in tetrameter (only 469 lines in pentameter)). A line may optionally end in an extra-metrical unstressed (“feminine”) syllable (usually, but not necessarily, in a word-final -e). This syllable does not contribute to the metre, hence it will be disregarded (see below).
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 3
words do not occur), retention is indicated only in the words under analysis. The dropping or retention of -e is inferred from the metre: therefore the metrical scansion shows unambigously whether an -e is (considered to be) dropped or retained.4 The paper is organised as follows. In 1, I will show that apocope (defined below) is significantly more frequent in Chaucer’s poetry than in Gower’s – in Gower, it is virtually non-existent. In 2, I shall introduce and review Smithers’s (1983) article from the aspect of the suffix -en, and we shall see that -en’s replacement by -e is – at least partly – metrically regulated. Further to this, I shall examine a statement concerning the final -e of strong participial forms with short stems, made by ten Brink in 1884 (2nd English edition 1901), which is, I believe, of greater significance than ten Brink supposed. Finally, in 3, analysing -e in metrically prominent (strong) positions, I will suggest that Gower may have used two variants of the article the.
1 Apocope in Chaucer and Gower 1.1 Preliminary concepts Let us start the discussion with the revision of two concepts, namely, two types of dropping the final unstressed vowel, which are indispensable in a paper on final -e: apocope (see (1)) and elision (see (2)). Definition of apocope: (optional) The final unstressed vowel (= -e) of a word is dropped. (1) Apocope: (a) -e in myghte is apocopated (dropped) /
X
X
that
X
/
saugh
X
/
myghte0
X
/
come0
/
therby |
‘When he saw he could not find it out’
4
Albeit line-final -e cannot add anything to our analysis (in fact, it would be a mistake to include it, unless our purpose were to attempt to draw conclusions about line-final -e itself), we will consider it retained, and indicate this in the scansion, by assigning it an extra foot.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 4
(b) -e in myghte is not apocopated (is retained) X
/
X
/
X
/
X
/
X
/
X
But hoom | he gooth; | he mygh | të nat | sojour | ne |
(CT, III. 987) ‘But he goes home; he cannot stay’ Definition of elision: (obligatory) If the final unstressed vowel (= -e) is followed by (an optional h and) another vowel in the next word, the final vowel (= -e) is dropped. e 0 / # (h)V Elision is a rule to eliminate hiatus, well known in many languages and historical periods. (2) Elision: (a) -e in come is elided (dropped) /
X
X
/
X
/
X
/
X
/
How that | a man | shal come0 | unto | this stoon |
(CT, VIII. 1474) ‘How a man shall come by this stone’ (b) -e in come is not elided (is retained) X
/
X
/
X
/
X
/
X
Schal co | më Thai | se for | to fet | te |
1362)5 ‘Shall come to fetch Thaise’ In (1a) we can see that the -e of myghte is dropped, even though it is part of the preterite ending -te. The same -e is retained in (1b). There is no phonological reason for its dropping in (1a): the only way we can know it is dropped is that its retention would violate the metre. This, then, is apocope. In (2a), the -e of come is not retained, although it is an infinitival ending. Its retention would violate the metre; its being silent is justified by its being followed by a vowel. This, then, is elision. In (2b), the same infinitival ending 5
Out of context, this line might raise the suspicion that Thaise was trisyllabic, stressed on X
/
X/
X
the penult (Thaísë): and the -e of come apocopated (|schal come |Thai|së). But that scansion would be impossible in all of its other occurrences (12 in number). 0
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 5
(attached to the same verb) is retained – this time required by the metre, and not followed by a vowel. The result of apocope and elision is the same: e 0; but whereas elision is phonetically conditioned, apocope is not conditioned – at least not immediately. Elision applies obligatorily: the number of exceptions is insignificant (for a possible exception see (3d1) below). Apocope, on the other hand, operates optionally. The problem that the research of ME word-final -e is highly concerned with is: to what extent is apocope a process (i.e. an optional rule), and to what extent are the apocopated forms “ready-made”, lexicalised. And conversely: which -e-retaining forms are still present in the contemporary language, and which ones are brought back by the poet, as conscious archaisms, for the sake of the metre. What is for certain is that apocope is not (or not entirely) an artificial device: it was concurrent with an ongoing diachronic process, whose result it is that by today there is no trace of word-final -e in English. (Of course there is word-final schwa – as, for instance, in visa, drama, or in Southern British English, error – but etymologically, these have nothing to do with ME final -e.)6 Smithers (1983: 213–14) claims that “there is no such thing as ‘apocope’, in the sense of an independent process”: he assumes that elision was an operative rule in non-metrical speech, too, well before the composition of HAVELOK (1310). He suggests that apocope is the analogical extension of elision. He is aware though (1983: 213) that “paradigmatic processes (i.e. in sub-systems of the ‘language’) may have worked along with [elision] in some varieties of ME”. But if the result of these paradigmatic processes (he mentions the syncretism of grammatical cases) was the loss of final -e, Smithers is also admitting the existence of apocope, although perhaps of a limited scope. He is also aware that the only tool which can show elision is metre – though I believe that if Smithers’s theory (that elision predated apocope in non-metrical prose) were right, there ought to be extant prose texts (most probably not of literary value – diaries, private notes, for instance), in which spelling would show that -e is retained preconsonantally but dropped prevocalically. Another problem with his hypothesis is that, as we have noted, inflectional reduction was an ongoing process – so if specan could be reduced to speke, it could possibly lose its -e, too, without this loss necessarily being triggered by an immediate phonological context. In addition to the above, we have to notice that because apocope is not sensitive to its immediate context, theoretically, it can take place whether the word-final vowel is followed by a vowel (#V) or a consonant (#C). However, 6
We have to note that the ending which became 0 by today did not have to start out from being a schwa – see OE specan above.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 6
as we have discussed, the -e#V context is the elision site: and elision, as a rule, operates obligatorily. Therefore, at elision sites, elision may override the operation of apocope: and it is only inferences drawn from other relevant metrical material that can help to distinguish which of the two is responsible for the dropping of -e before a vowel-initial word. We shall treat this problem in more detail below. 1.2 The data (Chaucer and Gower) These being stated, compare the occurrences of four monosyllabic7 nouns in CT and CONFESSIO; these nouns originally ended in -e (or in a vowel which had been reduced to -e by the time of ME). 1.2.1 Two Germanic nouns: name and herte Firstly, observe name (OE nama) in (3)–(4), and herte (OE heorte) in (5)–(6). 1.2.1.1 The noun name (3) name in CT (a) name0#V (elision): 11 times X
/
/
X
thus X
/
X
0
/
X
while
0
/
X
name
0
spron|ge| (CT, I. 1437) ‘And thus within some time his name has grown’ (b) name0#C (apocope): 3 times /
X
X
/
X
0
0
/
God have | hir soule ! |
(CT, IV. 530) ‘God have her soul! Her name was Alison’
X 0
/
X
/
was A | lisoun |
© namë (retention!): 7 times X
/
X
/
X
/
X
/
X
/
And looke0 | that thou | reporte0 | his na | më weel |
(CT, VI. 669) ‘And mind you say his name right’ 7
The present analysis is concerned with the presence or absence of final -e. Hence the nouns examined will be considered monosyllabic, with -e not counted as a syllable.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 7
(d1) Ambiguous: 1 example (1st reading) /
X
X
/
X
/
X
/
X
/
Telle me | the na | më of | the pri | vee stoon |
VIII. 1452) ‘Tell me the name of the secret stone’ 0
(d2) Ambiguous: (2nd reading) /
X
X
/
X
/
Tellë | me the |
0
X
/
X
/
of |the pri|vee stoon|
(4) name in CONFESSIO (a) name0#V (elision): 63 times X
/
X
/
X
/
X
/
X
His rih | te name0 | it is | Ravi | ne |
‘His true name is Rape’ (b) name0#C (apocope): 0 times © namë (retention!): 48 times X
/
X
/
X
/
X
/
X
Whos rih | të na | me was | Domil | de |
‘Whose true name was Domilde’ 1.2.1.2 The noun herte (5) herte in CT (a) herte0#V (elision): 86 times X
/
X
/
X
/
X 0
/
And in | his gen | til herte |
(CT, I. 1772) ‘And in his gentle heart he thought immediately’
X 0
/
anon |
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 8
(b) herte0#C (apocope): 5 times X
/
X
/
X
/
X
/
X
/
0
Into | myn herte , | that wol | my ba | ne be |
1097) ‘Into my heart, that will be my destruction’ © hertë (retention!): 87 times /
X
X
/
X
/
X
/
X
/
X
She was | agast, | and sey | de, her | të dee | re |
VII. 2889) ‘She was aghast, and said, “sweetheart”’ (d1) Ambiguous: 2 examples (1st reading) X
/
X
/
X
/
wel
X
woot
/
X
doost
/
X
her|të
to0
er|me| (CT, VI. 312) ‘But I know well you make my heart grieve’ (d2) Ambiguous: (2nd reading) X
/
/
X
wel
X
/
X
woot
/
X
doost
/
X
herte
0
er|me| (6) herte in CONFESSIO (a) herte0#V (elision): 178 times X
/
X
/
X
/
0
With al | his herte |
(CONFESSIO, I. 155) ‘With all his heart and make them glad’
X
make
0
/
X
chie|re|
(b) herte0#C (apocope): 0 times © hertë (retention!): 163 times X
/
X
/
X
/
X
/
X
Bot on | ly that | here0 her | të sou | cheth |
I. 315) ‘But only that their heart suspects’ The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 9
Before moving on to the other pair of nouns, let me note that in (3), the first reading of (3d) seems to be more probable than the second, since in (3d2) name does not get stress, although stress is expected because name is a content word. In (3d1), the elision rule is violated (i.e. ignored), but perhaps that is a minor offence, especially because name is followed by a phrase boundary (caesura), which may allow for a pause, as a result of which the final -e and the next vowel (that of of) are not in close enough contact for the rule to trigger, therefore they do not form a hiatus. In (5), more research would be necessary to decide which reading of the ambiguous example is more likely – that is, the one with the contraction of to plus a verb beginning in a vowel (i.e. t’erme); or the one in which the -e of herte is apocopated. We must also notice the syntactic construction in (4a): here, a resumptive it is inserted to license elision. Instead, name could be disyllabic (with -e retained): but that would result in a hiatus, that is, a violation of the elision rule. 1.2.2 Two Romance nouns Secondly, we shall observe the Romance nouns place (7)–(8) and cause (9)– (10). 1.2.2.1 The noun place (7) place in CT (a) place0#V (elision): 15 times X
/
X
/
X
/
X
/
0
X
And in | the grove , | at tyme |
(CT, I. 1635) ‘And in the wood, at the appointed time and place’
0
/
yset |
(b) place0#C (apocope): 1 times X
/
X
/ 0
X
/
And ther | fore in |
(CT, V. 186) ‘And therefore they left it at that place’
X 0
/
X
/
they han | it laft |
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 10
© placë (retention!): 11 times X
/
X
/
X
/
X
/
X
/
X
Lat se | now if | this pla | cë may | suffi | se |
4125) ‘Let us see now if this place can suffice’ (d1) Ambiguous: 1 example (1st reading) X
/
X
/
X
/
X
/
0
X
/
0
Heere in | this pla | cë, si | ttynge by | this post |
(CT, I. 800) ‘Here in this place, sitting by this post’ (d2) Ambiguous: (2nd reading) X
/
X
/
X
0
/
X
/
X
/
0
Heere in | this place , | sittyn | gë by | this post |
(8) place in CONFESSIO (a) place0#V (elision): 25 times X
/
X
/
X
/
0
And thanne | out of |
(CONFESSIO, I. 908) ‘And then he crept out of his place’
X
place
0
/
X
crep|te|
(b) place0# C (apocope): 0 times © placë (retention!): 44 times X
/
X
/
X
/
X
/
X
In pri | ve pla | cë wher | thei sto | de |
141) ‘In a private place where they stood’
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 11
1.2.2.2 The noun cause (9) cause in CT (a) cause0#V (elision): 28 times X
/
X
/
X
/
X
/
X
/
X
He knew | the cause0 | of eve | rich ma | lady | e |
419) ‘He knew the reason of every malady’ (b) cause0#C (apocope): 4 times X
/
X
/
X
/
X
/
X
/
By cause0 | that ther | was swich | diver | sitee |
II. 220) ‘Because there was such diversity’ © causë (retention!): 17 times X
/
X
/
X
/
X
Ye been | the cau | së wher |
I. 1568) ‘You are the reason why I die’
/ 0
X
/
X
that |I dy|e| (CT,
(10)cause in CONFESSIO (a) cause0#V (elision): 127 times X
/
X
/
X
/
X
/
X
And yit | the cause | is noght | deci | ded |
Prol. 334) ‘And still the cause is not decided’ 0
(b) cause0#C (apocope): 0 times © causë (retention!): 108 times X
/
X
/
X
/
X
/
And of | the cau | së ge | neral |
‘And of the general cause’ In order to decide whether (7d1) or (7d2) is more probable, we have to consider whether síttynge0 (with -e “apocopated”), or sittýngë (retention); and whether The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 12
place0 (with -e apocopated) or placë (retention) is more probable. In Werthmüller (2008) I pointed out that in Chaucer, monosyllabic verbs suffixed by -inge have the stress significantly more often on the stem than on the suffix; and in agreement with – among others – Halle & Keyser (1966) I claimed that (in Chaucer) unstressed -ing does not have an underlying final -e, which entails that in stem-stressed versions of sittynge the final -e is merely scribal and never corresponds to a pronounced sound (*[‘sittingə]).8 Furthermore, as we can see in (7b), place0 (with -e apocopated) virtually does not exist. Because there is no apocope in place, and because stress on the stem is significantly more frequent than on the suffix, I suggest the first and not the second reading. It is of great importance that all four apocopated instances of cause (as in (10b)) occur in by cause. This shows that by cause0 was on the verge of reaching lexicalised compound conjunction status (or perhaps that it had already reached it). The reason its status is not obvious is the fact that cause can also occur with retained -e in the structure by cause. 1.2.3 Tabulated summary Before drawing any conclusions about examples (3)–(10), in (11) we give a tabulated summary of our findings. (11) Tabulated summary of (3)–(10) (a) name SCHWAS: name DROPPED
elided Chaucer 11 (52%) Gower 63 (57%)
8
RETAINED
apocopated Σ 3 (15%) 67% 0 57%
TOTAL
7 (33%) 48 (43%)
21 (100%) 111 (100%)
In historical phonology, * can also introduce a reconstructed phonological entity. In this study, however, it refers to illicit forms only.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 13
(b) herte SCHWAS: herte DROPPED
elided Chaucer 86 (48%) Gower 178 (52%)
RETAINED
apocopated Σ 5 (3%) 51% 0 52%
TOTAL
87 (49%) 163 (48%)
178 (100%) 341 (100%)
© place SCHWAS: place DROPPED
elided Chaucer 15 (54%) Gower 25 (36%)
RETAINED
apocopated Σ 1 (4%) 58% 0 36%
TOTAL
12 (42%) 44 (64%)
28 (100%) 69 (100%)
(d) cause SCHWAS: cause DROPPED
elided Chaucer 28 (57%) Gower 127 (54%)
RETAINED
apocopated Σ 4 (8%) 65% 0 54%
TOTAL
17 (35%) 108 (46%)
49 (100%) 234 (100%)
Alternatively, we could represent the percentages by the following bar diagrams.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 14
(12) Proportion of retention, elision and apocope in CT and CONFESSIO (a) name 60% 50% 40%
elided
30%
apocopated
20%
retained
10% 0% Chaucer
Gow er
(b) herte
60% 50% 40%
elided
30%
apocopated
20%
retained
10% 0% Chaucer
Gow er
© place 70% 60% 50% 40%
elided apocopated
30% 20% 10% 0%
retained
Chaucer
Gow er
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 15
(d) cause 60% 50% 40%
elided
30%
apocopated
20%
retained
10% 0% Chaucer
Gow er
1.2.4 Discussion of the findings The most striking fact these data reveal is that apocope is absent in Gower’s verse. The corpus examined is not large enough for us to be certain that Gower did not apocopate at all,9 but it is doubtless that the loss of -e for him was not as well-advanced as for Chaucer, and was still a strictly phonological rule (elision). If one looks at Gower’s proportion of the retained vs. dropped (that is, elided) -e’s, one must come to the conclusion that elision was rather an obligatory phonological rule for Gower than a device to ensure the dropping of the -e’s (since a great number of the -e’s are retained). Compare the Gowerian diagrams of (12) with this imaginary one: 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% elided
9
retained
In fact, he did apocopate: mainly in auxiliary verbs – which is not surprising, bearing in mind the fact that one of the categories most often used in the language is that of auxiliaries. A language change can be first observed on the most widely used elements.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 16
If such a diagram could be drawn, one would be bound to claim that Gower used elision, as it were, to license the dropping of -e by means of the immediate phonological context. This would mean that certain words (most probably those in frequent use) were further along in the process of losing their -e (lexical diffusion). For Chaucer it is also true that the sum of the dropped (the apocopated plus the elided) -e’s does not outnumber the retained ones so significantly that one should believe that the retention was just a last resort. Nevertheless, with a greater corpus, one may find that for certain words, the proportion of retention is significantly lower than for other words. The enlargement of the corpus must be the task of future research. Farrington Babcock (1914: 81) suggests that there is “at least a tendency for the Romance noun to retain e more frequently [in other words, apocopate less – my comment] than the Germanic one”. She claims this about Chaucer and illustrates it with a table, in which no data from CT are included. Her calculations do not break down to individual nouns (we do not even know which nouns she analysed), they are concerned only with the total of Germanic and Romance nouns. This does not exclude the possibility that some nouns contribute to her result by a significantly greater share than others. When counting the proportion of apocopated instances, Farrington Babcock (1914: 62) disregarded examples where “elision naturally occurs”:10 that is, for her the hundred percent was made up from the apocopated and the -e-retaining instances. This means that, if I performed her count on my corpus, the percentages received that way would be greater than those counted including the elided instances, although their relation to each other would certainly be the same. (13) compares the percentages of apocope counted (a) according to my method (data almost the same as in (3–10), but now rounded off to the first decimal digit), and (b) according to Farrington Babcock’s (that is, elided cases excluded). (13) Percentages of Chaucerian apocope in name, herte, place, cause (a) elided cases included (my method) name (14.2%) > cause (8.1%) > place (3.5%) > herte (2.8%) (b) elided cases excluded (Farrington Babcock’s method) name (30%) > cause (19%) > place (7.6%) > herte (5.4%) 10
From this it does not emerge whether she included those examples where elision should occur but does not (as in (3d1) or (18a)): however, these instances are so few in number that they would not alter the result significantly.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 17
The above calculation can be summarised by the following table: Apocopated word name0 cause0 place0 herte0
My count (all occurrences
100%)
Farrington B.’s count (retention+apocope
100%)
14.2 % 8.1 % 3.5 % 2.8 %
30.0 % 19.0 % 7.6 % 5.4 %
It can be seen that the two counts yield the same pattern. Farrington Babcock (1914: 62) says that if -e is followed by a vowel or h-, elision “naturally occurs”; this is why she disregards the elided cases. But as we have seen in 1.1, an elision site, by definition, obscures any possible apocope at the same place. It is highly probable that Chaucer would not have apocopated every -e at the elision sites: but the possibility is there that he would have dropped some of those -e’s, where they were followed by a consonant. This possibility is not to be assumed for Gower, however, because he did not apocopate an -e anywhere. The findings of (13) do not exactly confirm Farrington Babcock’s (1914: 81) assumption: of these four nouns, although it is a Germanic noun (name) that apocopates most, it is also a Germanic noun (herte) that apocopates least. The corpus examined, however, is not large enough to refute her theory – and it must be observed that a Germanic noun which had been part of the language for a longer time than a Romance one, is likely to apocopate more regularly than a Romance one. If we look at the Gowerian forms, we can see that the -e is underlyingly present in them: it is either retained, or its dropping has a contextual explanation (elision). The deletion is always post-lexical. In Chaucer, however, there are forms (the apocopated ones), whose -e-dropping may be required by the metre, but in order to exist at all, they had to be allowed by his lexicon. About the very frequently used word have, Guthrie (1988: 391) says that “vestigial final -e may still be retrieved at the poet’s discretion, perhaps for metrical reasons, but probably not for metrical reasons alone: the poet must be willing to accept the stylistic effects which come with the ending, as a contemporary American poet who uses oft instead of often must be able to make use of its stylistic register.” This may suggest that Chaucer had two forms of have stored in his lexicon: an -e-dropping one and an -e-retaining one. To extend this assumption to every word may be too far-fetched: but what The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 18
is certain is that whereas Chaucer’s (poetic) phonology allows him to delete (i.e. apocopate) certain -e’s without any more reason than the metre and the grammatical process of deleting inflections and word endings, Gower was keen to make sure that the dropping of -e always had a phonological reason. Without treating them at the same length as monosyllabic nouns, let me show that the same is true for Gower’s disyllabic nouns borrowed from French (like nature) – except for one single noun: manere. These words could be stressed in the Romance way, on their penultimate syllable (natúre?), with -e retained or dropped; or in the Germanic way, on the first syllable (náture0), with the -e probably dropped. We must note though that it would be possible for náturë to occur in a verse line, if we assumed an inverted foot: (14) Dactylic náturë? /
X
X
/
X
/
X
/
X
Natu | rë, tok | hem in | to lo | re |
‘Nature [who] took them into lore’ But foot inversion cannot always be assumed (only after a line or phrase boundary), and Romance stress was possible anyway – moreover, more probable than Germanic stress, as (15) (a typical scenario) shows. As it stands, then, the following variations are possible:
Dropped Retained
Romance stress
Germanic stress
natúre0 natúrë
náture0
- náturë
Let us examine the occurrences of nature (see (15)) and manere (see (16)) in Gower. (15) the pronunciations of nature in CONFESSIO V
C
nátur#V: 10 nátur#C: 0 natúr#V: 21 natúr#C: 0 natúrë#V: 0 (violating elision, hiatus) natúrë#C: 11 náturë#: 0 (impossible for metrical reasons)
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 19
(16) the pronunciations of manere in CONFESSIO V
C
máner#V: 11 máner#C: 39 manér#V: 34 manér#C: 0 manérë#V: 0 (violating elision, hiatus) manérë#C: 3 mánerë#: 0 (impossible for metrical reasons) (15) shows that no matter whether nature is front-stressed or end-stressed, its -e is dropped only if it is followed by a vowel. That vowel is necessary to ensure the non-pronunciation of the -e – which in the front-stressed tokens is never retained. That the -e (schwa) is always present underlyingly appears from the fact that its dropping can always be justified. That no vowel can follow the retained -e of end-stressed natúre can be accounted for by the fact that hiatus is normally avoided in ME verse. In (16), the end-stressed examples behave the same way as in (15): if the -e is dropped, it is always followed by a vowel to justify the elision. If the -e is retained (hence no elision site is required), it is always followed by a consonant, to avoid hiatus. However, in the majority of the front-stressed tokens, mánere0 is followed by a consonant. No final -e is retained, though no vowel is present to create the elision site. This may allow for the assumption that the -e of mánere normally apocopates in Gower: but if something “normally apocopates” and is hardly ever pronounced, we might as well claim that it is not there underlyingly any more, i.e. that this word already had a form máner in Gower’s lexicon. Furthermore, if we observe other disyllabic Romance nouns which can have the stress on either syllable, we find that they behave like nature, not like manere. The data in (3)–(10) also show that apocope is very atypical of Gower. Hence we had probably better conclude that manere was one of the very first words in Gower to go through restructuring: unlike other words borrowed from French, it usually had the stress on the first syllable; and if it did, the -e (schwa) was underlyingly not present. This suggests that manere was further advanced in the process of losing final -e than most (if not all) other nouns in Gower. Windeatt (1977: 57-8) remarks that “[t]he marked tolerance of variable stress is suggested by use of the more ordinary loanword manere, which in [Chaucer’s] TROILUS appears to be mostly accentuated in the modern way when occurring in midline”. Indeed, maner was a commonly used word (more commonly than today), appearing in such expressions as maner thing (‘sort of thing’). The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 20
On the foregoing pages my intention has been to express my agreement with the school which holds that word-final -e was working (more or less) regularly, and its function was grammatical and (in the data analysed in (3)– (10)) lexical. I have to reject Robinson’s (1971: 95) opinion, according to which “[…] in a few cases Chaucer felt that the ordinary pronunciation of a word was with sounded -e (deere for instance frequently appears to break the elision rule and have its -e sounded before a vowel or h-, especially in the phrase deere herte), though in most cases the -e is only to be sounded on special occasions, and on others left unambiguously silent through elision. […] in all these cases he tends to avoid unnecessary problems whether by leaving the reader discretion at a place where some readers might otherwise damage the verse, or by preventing the problem arising at all when he elides the -e.”11 Robinson seems to suggest that elision helps ensuring the usual dropping of -e. This would be a tenable assumption, if there were no final -e’s which solely the metre shows are to be dropped (that is, apocopated). It is true that Robinson is not completely against the sounding of word-final -e’s, and he is aware that final -e’s cannot be inserted at the end of any word. He presumes that the -e’s once justified by grammar can still be pronounced, but most of the time this has a rhetorical effect. I believe that the data and the analysis above do not confirm this theory.
2 Smithers’s theory and the issue of -en As we have mentioned in 1, the infinitival -en could be reduced to -e in ME. This is also true for the other verbal -en suffixes: the past participial -en, the plural indicative and the plural subjunctive -en. Smithers (1983) is very much concerned with the problem of -en and -e. In this chapter, I shall briefly examine his theory, and investigate a question invoked by a statement by ten Brink [1884 (1901)]. 2.1 Smithers’s theory on -en Smithers enumerates the contexts where graphic -en can appear; in (17) I reproduce his list in an abbreviated version (omitting some comments and the line numbers of further examples, but preserving his grouping, even though I disagree with its methodology – see below).12 11 12
Robinson does not notice that the -e of deere is the vocative inflection. He ignores the grammatical role of inflectional -e’s throughout his book. Smithers distinguished his types by Arabic numerals. In order to avoid ambiguity, these are replaced by Roman numerals throughout this paper. Where Smithers writes “vowels or diphthongs”, I simply write “vowels”, as it is a cover term for diphthongs as well in modern linguistic terminology. Smithers did not provide translations for his examples: the
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 21
(17) Phonetic and rhythmical contexts (Smithers 1983: 203–205) [lines quoted by Smithers from his own edition of HAVELOK] (I)
when -en is followed by an unaccented syllable with initial vowel: X
/
X
/
X
X
/
X
/
And he gart komen into the tun (1002) ‘And he got to come to town’ (II)
when followed by an unaccented syllable with initial h- + vowel: /
X
/
X
/
X
X
/
X
With the wende shulen he yern(e) (1347) ‘They will yearn to go with you’ (III) when followed by an unaccented syllable with initial consonant (other than h-) + vowel: /
X
/
X
X
/
X
/
0
Ther he greten for hunge r and cold (449) ‘There they cry for hunger and cold’ (IV) when -en is the second of two unaccented syllables, and is followed by an accented syllable (IVa) with initial vowel: X
/
X
X
/
X
/
X
/
X
0
And men haue den of him mikel dred(e) (181) ‘And people were very much afraid of him’ (IVb) with initial h-: X
X
/
X
/
X
/
X
/
X
Wolden he nomore0 to putting gang(e) (1058) ‘They would not go to throw again’ (IVc) with initial consonant (other than h-) + vowel: X
/
X
X
/
X
/
X
/
That ye aren comen to me now (161) ‘That you have come to me now’ translations are mine. Note that in this text, he is both the 3rd sg. masc. and the 3rd pl. personal pronoun (see especially IVb and VIb). The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 22
(V) when followed by an accented syllable with any initial consonant: X
/
X
/
X
/
X
/
And putten with a mikel ston (1024) ‘And they threw with a great stone’ (VI) when followed by an accented syllable (VIa) with initial vowel: X
/
X
/
X
/
X
/
X
0
And mikel sorwe haue den all(e) (238) ‘And they all had great sorrow’ (VIb) with initial h- + vowel: X
/
X
/
X
/
X
/
X
He wrungen honde0s and wepen sor(e) (152) ‘They wrung their hands and wept bitterly’ For ease of understanding, see the following table, summarising Smithers’s types. Stress pattern
Preceding syll is
Next word begins with
Next word’s 1st syll is V
/xx
accented
-en
unaccented
xx/
unaccented
-en
accented
/x/
accented
-en
accented
I kómen intó IVa hav’den óf VIa háv’den álle
C (not h) II III shúlen he gréten for IVb IVc wolden aren hé cómen VIb V wrúngen pútten hóndes wíth hV
Smithers’s (1983: 205–208) basic idea is that where -en would make one of two unstressed syllables, the letter -n was unnecessarily inserted by a scribe: the poet intended only an -e. This -e, if it is prevocalic, gets elided; if it is preconsonantal, it is apocopated. He thinks that unnecessary scribal insertion The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 23
of -n is also possible in type V, “since the metrically obligatory unaccented syllable is equally well supplied by -e” (1983: 206). Before turning our attention to this theory and the problems it faces, let us notice that in the article mistakes appear, caused by Smithers’s ignoring some of his own data. He notes the following: “What is extremely significant is that in types I, II, III, and IVc -en makes one of two successive unstressed syllables” (Smithers 1983:206). This comment is correct, as far as it goes: but Smithers fails to notice that in IVa and IVb it does so, too. The same lack of attention on his part is visible when he writes: “In type I, if -e was the author’s form, it could (and would) have been fused with the following initial vowel by elision. In type II, when initial h- was ‘dropped’ (as it can be, and intermittently is, today), the result would have been as in type I. […] in types III and IVc elision of the -e in the -en is phonetically not possible13 (because of the following consonant) […]” (Smithers 1983: 205). Here, Smithers wishes to list the contexts where elision is possible (if the -n is considered to be merely graphical), but again he omits IVa and IVb. In (17), a methodological inconsistency also appears. Three groups could be made: one in which -en is the first of the unaccented syllables (I)–(III); one in which -en is the second of the unaccented syllables (IV); and one in which -en is the only unaccented syllable (V)–(VI). Since in all these contexts -en is followed by the same types of segmental material, each group should be divided into an equal number of types. This is shown by the table above. Smithers’s (1983: 205–208) idea that unstressed -en (as it stands in the manuscripts), if it is next to another unstressed syllable, is really an -e, while truly pronounced -en is only necessary if it has to fill a hiatus, may seem to be attractive, but it poses some problems. Two unstressed syllables next to each other, if they have to occupy the same metrical position (unlike in the rejected scansion in (14)) make a line hypermetrical: and Smithers seems to be oblivious of the fact that lines exist in which neither of the two successive unstressed syllables is (or would be) an -en. For instance, he scans hunger (III) and hondes (VIb) as monosyllabic, but he does not justify this. In these cases, he would be unable to invoke elision. About hunger, it could be supposed that the -e is syncopated – but seeing the tokens in CT, one cannot consider this to be a tendency, so an explanation would be necessary. In CT, altogether 21 non-line-final instances of hondes, handes, londes and landes appear: of these, only two are monosyllabic. Hunger has 6 occurrences, none of which are monosyllabic. Nevertheless, I could accept Smithers’s argument about -n’s being only scribal insertions in syllables which make a line hypermetrical, and whose -e could be dropped (elided or apocopated) without the -n coda, if he 13
Phonetically, -e could be dropped – phonologically, elision would indeed not be possible.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 24
had pointed out that there are a significantly greater number of (would-be) hypermetric lines with -en than without -en. Smithers (1983: 207), to prove his theory, shows cases where the -n is surely spurious: he is both the 3rd sg. masculine and the 3rd pl. nominative pronoun, but the -en inflection (in HAVELOK) can only express plural. Therefore, he + verb + -en, if he is singular, is corrupt. Smithers (1983: 208) believes that the only role of -en (both in speech and verse) was to fill a hiatus: hence he does not think that the -en was authorially intended in type V either, because – no elision being possible here – an -e would have sufficed. Everybody who has dealt with the structure of at least one language could list countless examples where less material would suffice, yet the language resorts to more. In standard Modern German, one says and writes Sie singen mit dem Lehrer, yet *Sie singe mit dem Lehrer (‘they sing with the teacher’) would suffice just as much as and putte with a mikel ston (‘and they threw with a great stone’) in type V would: yet singen, and not singe, is the standard use. Another question is, why the hiatus ought to be avoided at all. Smithers himself admits that “incontestable cases of hiatus are very common in HAV[ELOK]” (1983: 206).14 Minkova (1987: 454) claims that “there is no evidence in English, at any point in its history, for any phonotactic constraint on hiatus, word internally or across word boundaries. Nor is there any universal, or even a tendency, for languages to disallow hiatus; in fact languages with and without hiatus are about evenly distributed”. Up to this point we maintained that elision operates in order for the hiatus to be avoided. This is the traditional working definition – but it is inaccurate. Hiatus refers to two vowels next to each other, which can occur within one word ([leun] – leoun, meaning ‘lion’), or across word boundary (citee of, to eten – ‘city of’, ‘to eat’). This paper is not concerned with how these may have been pronounced – whether, for instance, the hiatus was filled by a glottal stop (as in Modern German) or not. But English has had no general tendency to avoid two vowels occurring next to each other (hiatus sites have always been present). What is to be observed, however, is that in some accents of British English (RP, for instance), VV sequences do not occur at all if the first vowel is a schwa. The indefinite article is not /ə/ but /ən/ before a vowel-initial word (an evening as opposed to *a evening); the article the has a long [i] in the same environment ([ði] as opposed to *[ðə] evening); otherwise, an intrusive [r] is pronounced between the schwa and the next vowel (China[r] and Japan). The situation is not that straight-forward though: historically speaking, an and not a was the default for the indefinite article; and intrusive [r] is present after 14
See also (3d1).
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 25
[ɑ] and [], too: mama[r] and papa, draw[r] it. Nevertheless, this can certainly not be termed a general rule for avoiding hiatus. In ME, due to word-final -e not yet having been lost, most occurrences of hiatus across word boundary would have entailed that the first vowel was a schwa. Smithers does not go into detail about the definition of hiatus, but it is highly probable that he means it only in this narrowed sense (as does our definition of elision). It is beyond doubt that hiatus across word boundary is infrequent in ME verse (see (3d1)): but if we remember that it did exist; that in the manuscripts -en does not only appear at places where it can fill a hiatus; and that languages do not always work economically, then we have good reasons to doubt that the -en was exclusively intended by the author as a hiatus-filler: that is, in type V, where -en does not make the line hypermetrical, but -e would have sufficed, the -n may well have been also intended by the author, not inserted accidentally by a scribe. 2.2 h- and the question of hiatus In order to investigate how frequent hiatus actually was in ME verse, the question of word-initial h- also ought to be settled. As discussed in 1.1, and shown in (17), elision is supposed to take place before a vowel and an initial h-. This suggests that initial h- is not pronounced, so words beginning in h- in fact begin in a vowel. But the analysis needs to be more fine-grained: Guthrie (1988: 92) mentions “hard h-”, and his context lets us believe that before hard h- elision did not operate. But as (18) shows, this is still not satisfactory. (18) -e before h(a) -e is retained X
/
X
/
X
/
X
/
X
/
X
Among | the prin | ces in | this er | thë hie | re |
PRAISE OF PEACE, 51) ‘Among the princes on this earth here’ (b) -e is dropped X
/
/
lat
X
/
X
li|tel
/
child
X
/ 0
heer
X
thee| (CT, II. 859) ‘Let my little child live here with you’
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 26
In (18a) and (18b), the h-initial word is the same: heer (spelling is irrelevant). In (18a) the schwa of erthe is lexical; in (18b) the (elided) schwa is a grammatical ending, the infinitive marker. If we assume that this initial h- was not pronounced, then (18a) contains a hiatus; the -e preceding it does not get elided, whereas that in (18b) does. We have to note, however, that in (18b), dwelle0, a lexical verb, fills a weak position, and two weak syllables next to each other are not allowed in an iamb, so the -e gets dropped. Dwelle can fill a weak position because it has a low information content: whereas here can be focussed. The secondary literature tends to explain the elision before h- along the description of ten Brink [1884 (1901: §269)]: “[…] weak e is also elided before following h. This affects in English words chiefly the initial h in he, him, his, hire, here, hem, […] how, heer, and various forms of the verb have, in Romance words the mute h as in honour, honest, humble, humilitee etc. […] It is noteworthy that the aspirated French h also occasionally permits of elision [harneys]”. The English words in ten Brink’s list are most often unstressed; so are the initial syllables of the listed French words. Schwas which are elided would be unstressed if they were pronounced, but two unstressed (metrically weak) syllables are never located next to each other. Elision is a process which helps to avoid this situation. It follows from the metre (and often from the phonology) that the syllable preceding the final (elidable) -e is stressed.15 In other words, then, elision makes room for the unstressed syllable following the elided -e. Since the listed h-initial words (can) also begin with an unstressed syllable, their first syllable can fill that space. Hence, the tendency seems to be that elision operates before those h-initial words whose first syllable is unstressed. But this leads on to the question of what happens to a final -e before unstressed syllables not beginning in h- (or in a vowel, in which case elision must operate, so that issue is ignored here), as in (19). (19) Elision impossible X
/
X
/
X
/
X
/
X
/
O goode | Custan | ce , | allas! | so wo | is me |
II. 817) ‘O, good Custance, alas! I am so sad’ 0
0
Goode Custance is a vocative phrase, in which the adjective ought to receive an inflectional word-final -e. In (19), this -e would be one of two successive unstressed syllables, and it is the only one in its surroundings (goode Cu-) 15
See (23) for a possible exception.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 27
which can be left unrealised: since it is not followed by a vowel, this must be an instance of apocope. This may suggest that final -e before h-, which, most frequently, gets dropped under the same metrical circumstances (i.e. when the next word is, or begins with, an unstressed syllable), does not really elide, but apocopates: and the fact that the next consonant is initial h- is just coincidental. We must note that – as listed by ten Brink – the elision rule operates before frequently used function words such as hire (‘her’) or have. It is probable that – as it is often the case in Modern English as well – the initial h- of these words did not have to be pronounced, they could begin in a vowel. Moreover, before the word herte, for instance, no obvious elision occurs either in CONFESSIO or CT. There are ambiguous examples, where either the -e or the preceding consonant is syllabic (as in tendre? herte); and the -e is dropped in the genitive pronouns (hire0 herte): but that -e is normally dropped in any other context. Nevertheless, this analysis faces at least one crucial problem: in ME, words beginning in h- were not preceded by the determiners a, my, thy: they took an, mine, thine, like vowel-initial words. A search for the string “my h” in the entire CORPUS text of CT (see References) yields one single occurrence: my hand, which may be a mistake for the normal myn hand, made by a medieval scribe or a modern editor. This result may imply that word-initial hwas not pronounced at all, even in words which begin in a stressed syllable. But perhaps we are experiencing a similar phenomenon as in Modern English, when people, reading out the phrases an hotel or an historical fact, although they pronounce [ən] and not [ə], sound the [h], too. The problem definitely needs a more in-depth scrutiny. 2.3 Ten Brink’s statement on the participial -en Ten Brink observes the following concerning participial -en: “Final weak e often becomes mute immediately after the tonic syllable, and in the following cases it is never syllabic [that is, never pronounced – my comment]: […] In the strong P[ast] P[articiple] of verbs with an originally short root, when the -n is apocopated: come, drive, write, stole etc. (but on the other hand, comen, driven, writen are naturally dissyllabic and stolen occurs beside stoln)” [1884 (1901: §260)]. The present paper, like ten Brink’s book, has drawn conclusions about pronunciation by assuming the existence of a regular metre. The above observation, however, cannot be confirmed or refuted by the same method: the question here is not whether a syllable was present or absent, because both comë and comen are disyllabic; but whether it did or did not have a specific coda (in a specific grammatical role).
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 28
It must be mentioned that in the case of the verb come the stem vowel (short [u]) of the participle and the infinitival forms is identical: both of them can end in -en ([kumən]), -e ([kumə]); they can be monosyllabic ([kum]). The infinitive and plural indicative and subjunctive can be disyllabic ([kumə]), and only the homophonous participial disyllabic -ë form ([kumə]) is ruled out by ten Brink. His statement is summarised in (20). (20) Ten Brink’s findings
1 syll 2 syll’s 2 syll’s
Infinitive come0 comë comen
Past Pple come0 comen
To illustrate this with an example: according to ten Brink, (21b) is impossible for Chaucer (no example can be found in the CORPUS edition of CT either); but the form (disyllabic comë) has ample appearances in Peck’s (2000) critical edition of CONFESSIO. (21) Short-stemmed participial form in ME (a) Original line /
X
X
/
X
/
X
/
X
/
Whan that | Arcite | to The | bes co | men was |
1355) ‘When Arcite arrived in Thebe’ 0
(b) Impossible for ten Brink, found in one manuscript (Ha4)16 /
X
X
/
Whan that |
16
X 0
/
X
/
X
/
to The | bes co | më was |
Let me express my thanks to Rosamund Allen, who kindly made her collated manuscripts of the KNIGHT’S TALE available for me. The collation was prepared as part of a Leverhulme Funded project titled AUTHORIAL AND SCRIBAL METRICAL PRACTICE IN MIDDLE ENGLISH TEXTS; only parts of which were published.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 29
© Cf. its appearance in Gower X
/
X
/
X
/
X
/
X
Which now | is co | më to | my myn | de |
1406) ‘Which has now come to my mind’ Ten Brink’s discovery may be more significant than he might have supposed. I believe that nobody is able to find phonetic material which would prove that Chaucer did not use disyllabic forms such as comë: the metre does not help to decide whether a non-syllabic consonant ([n]) was actually pronounced. What the data may reveal is that in the manuscripts examined by ten Brink, no participial form of these short-stemmed strong verbs appears which ends in -e and should be read disyllabically. Ten Brink compared the most reliable manuscripts – and the present study cannot undertake an extensive manuscript examination. But my comparison of 10 manuscripts of Chaucer’s THE KNIGHT’S TALE revealed that (21b) appears in a manuscript ten Brink did not examine. Nevertheless, ten Brink may have found a systematic gap in manuscripts. But the manuscripts were produced after the lifetime of Chaucer, which implies that Chaucer may have used these forms, which were found ungrammatical by later scribes. Unlike Chaucer, Gower could revise his extant manuscripts, and the fact that Peck’s (2000) edition contains a number of examples of the disyllabic participial comë suggests that Gower used such forms, too, not only such disyllabic forms as comen; therefore one might reasonably assume that his contemporary, Chaucer did so, too, and ten Brink found only that the scribes regularised Chaucer’s use, having inserted -n everywhere. However, perhaps (22), which examines the (superficially) monosyllabic tokens of the participle come, also suggests that forms such as disyllabic comë were indeed not part of Chaucer’s grammar. These forms could be monosyllabic, either because final -e got dropped, or because it was underlyingly not present. (22) come (past part.) in CT (total: 31 times) (a) -e0 #V (elision): 21 times X
/
X
/
X
/
X
/
X
/
X
At nyght | was come0 | into | that hos | telry | e |
I. 23) ‘At night had come to that hostel’
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 30
(b) -e0 #C (apocope): 10 times X
/
/
day X
X
/
come
X 0
/
X
hom|ward
/
moste
X 0
tour|ne| (CT, III. 988) ‘The day had come that he had to turn homewards’ (22b) shows that approximately in one third of the instances (10 out of 31), the -e of come was followed by a consonant, so the underlying schwa – if it was there at all – had to apocopate due to metrical pressure. But perhaps one third is significant enough for us to stipulate that the -e was not present underlyingly in any of the examples: that is, no elision took place, and the form [kum] was followed by a vowel 21 times without any metrical significance. In future research, more verbs must be examined from this perspective, and the question is worthy of a thorough manuscript research. But if ten Brink’s statement is true, then there is at least one minor detail where most manuscripts agree – which is important because in minor details manuscripts usually diverge. Furthermore, we have seen that -n instead of -e may not only have a phonological role (hiatus-filler), but also a grammatical one. Hence ten Brink’s observation undermines Smithers’s (1983) theory introduced above: it refutes his view that the only reason for writing -en instead of -e was to fill the hiatus.
3 Schwa stressed? A problem with modern theories of poetic metre is that they do not pay enough attention to what cannot fill a strong position. Halle & Keyser (1966, 1972), though they started out from Chaucerian lines, did not consider it a great metrical offence if an unstressed syllable occupied a strong position (if the neighbouring syllables were not unstressed). They did not really distinguish the degrees of being unstressed. Their main concern was for a stressed syllable not to occupy a weak position between two unstressed syllables: in their theory, only that offence made a line unmetrical. Kiparsky’s first metrical rule (1975: 583) says that “a primary stress can be freely replaced by any other stress” – which includes zero stress, too. It can be seen that these authors are not really interested in ME – in which schwa, the most unstressed vowel, occurs much more frequently than in later stages of English: that is, researchers of later English metre do not have to be that much concerned with unstressed syllables as the researcher of ME metre.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 31
Infrequently though, schwa could probably fill a strong position in Chaucer’s and Gower’s English poetry. The question is so unsettled that modern scholars cannot even unambiguously decipher ten Brink’s [(1884) 1901] opinion, let alone that of Chaucer. According to Ian Robinson (1971: 91) ten Brink believed that a weak -e could fill a strong position, whose effect is “to turn -e into -er, that is, to give the sound more prominence than we can believe it to have had in speech” (1971: 99). (In modern phonetic terms Robinson’s statement probably means “to turn [ə] into [з]”.) Robinson does not cite where he thinks ten Brink suggests that an -e can fill a strong position. On the other hand, Guthrie (1988: 389) maintains that in ten Brink’s (implicitly expressed) view a weak -e cannot fill a strong position. The section of ten Brink’s book from which Guthrie drew this conclusion (§194) describes what can occupy a strong position: and weak -e is not listed. I agree with Guthrie, but I believe that there is a more direct hint in ten Brink: “cf., for instance, weddede [A. 868], where [the manuscripts] read wedded, and the verse seems to require the complement of a monosyllable” [1884 (1901: §256)]. The line referred to by ten Brink is shown in (23). (23) Chaucerian schwa (?) in strong position X
/
X
/
X
/
And wed | dede |
‘And wedded the queen Hyppolita’
X 0
/
X
/
Ypo | lita |
Ten Brink seems to suggest that weddede is disyllabic: the -e is not pronounced. It is admitted by many metrists that there are lines which will have the required number of syllables only if a weak -e fills a strong position: (23) is one of these lines. Such lines do not occur frequently, and it would probably be inconsiderate to assume that such strong -e’s are indeed stressed. About (23) it was pointed out to me17 that weddede is followed by a major phrase boundary, a dramatic caesura, in anticipation of a new piece of information. Perhaps this phrase boundary (a pause in performance) licenses the weak syllable in the strong position. It is not my intention here to set up a hypothesis concerning the reason and the pronunciation of the Chaucerian final schwa in strong position: my aim is to claim that in CONFESSIO, a weak -e never fills a strong position, except in the case of the word the, as can be seen in (24).
17
By Dr Ádám Nádasdy (personal consultation).
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 32
(24) Strong schwa (?) in Gower (in CONFESSIO II. (3530 lines)) (a) Unambiguous (6 cases): X
/
X
/
X
/
X
/
X
Now af | ter Pri | de the | secoun | de |
‘Now after pride the second (one)’ (b) Ambiguous (4 cases): ?
?
X
/
X
/
X
/
X
0
That the | greteste | of Bar | bari | e |
‘That the greatest of Barbary’ I believe there are ambiguous cases, when the first (and possibly the third) foot is shared by the and another function word: these are potential inversion sites, so it cannot be decided which part of the foot is more prominent. Because in Gower, the -e of the was the only final -e able to fill a strong position, it is possible that Gower used two versions of the; one with schwa (in weak position, which was significantly more frequent) and one with a full vowel in strong position (similarly to Modern English, see 2.1). In order for us to make generalisations about their appearance in Gower, all instantiations of the strong the should be found and examined. However, it is striking that in the first 500 lines of the Prologue of CONFESSIO, five unambiguous cases can be found (compared to the six unambiguous cases of the 2530 lines of book II). The Prologue is a rhetorical text, examining the church and the people; (24a) also has an elevated tone, opening the description of the second deadly sin. So perhaps here the register has a decisive role in the choice of the poet – strong the is more frequent in Gower’s rhetorical verse.
4 Conclusions A definite short-coming of the present paper is that it does not gather data from manuscripts, but relies on critical editions instead (see References). This, undoubtedly, makes it seem less authoritative, since metrical divergences may remain unnoticed. Bearing in mind, however, that manuscripts diverge from one another (and there is no extant MS that Chaucer would have seen – Gower could supervise some of his MSS), we cannot think in terms of exact numbers, but only in terms of tendencies. This means that examining manuscripts would not have influenced these findings significantly. This paper, like any other paper on word-final -e, probably contains more questions than answers: for instance, instead of coming up with (and in order to be able to find) a solution for the question of hiatus, it had to tackle the The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 33
issue of h- being pronounced or not. Nonetheless, I have argued that, as traditional historical linguists maintain, word-final -e is more than a metrical device: it has grammatical importance. I have shown that -en (which, as a verbal suffix, is a “synonym” of -e) is more than a simple hiatus-filler. We have also seen that in ME there may have been two allomorphs of the article the. All the questions concerning -e need a much thorougher research than this paper can afford.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 34
References Barber, Charles and Nicholas Barber. 1990. The versification of the Canterbury Tales: a computerised statistical study. Leeds Studies in English 21: 81-103. Brink, Bernhard Aegidius Konrad ten. 1901. The language and metre of Chaucer. 2nd edn. (translated by M. Bentinck Smith) London: MacMillan and Co. Originally published: 1884. Chaucers Sprache und Verskunst. Leipzig: T. O. Weigel. Cable, Thomas. 1991. The English alliterative tradition. Philadelphia: University of Pennsylvania Press. Farrington Babcock, Charlotte. 1914. A study of the metrical use of the inflectional e in Middle English, with particular reference to Chaucer and Lydgate. MLA 29: 59–92. Gower, John. 2005. In praise of peace. In: Kathleen Forni (ed.). The Chaucerian apocrypha: a selection. Kalamazoo, Michigan: Medieval Institute Publications for TEAMS. Available online: http://www.lib.rochester.edu/camelot/teams/tmsmenu.htm. Last accessed: 23.05. 2010. Guthrie, Stephen R. 1988. Babcock’s curve and the problem of Chaucer’s final -e. English Studies 5: 386–395. Halle, Morris and Samuel Jay Keyser. 1966. Chaucer and the study of prosody. College English 28: 187–219. Halle, Morris and Samuel Jay Keyser. 1972. The iambic pentameter. In: William Kurtz Wimsatt (ed.). Versification: major language types. New York: Modern Language Association. 217–237 Herzman, Ronald B., Graham Drake and Eve Salisbury (eds.). Havelok the Dane. In: Four romances of England. Kalamazoo, Michigan: Medieval Institute Publications. Available online: http://www.lib.rochester.edu/camelot/teams/danefrm.htm. Last accessed: 23.05. 2010. Kiparsky, Paul. 1975. Stress, syntax and meter. Language 51: 576-615. Minkova, Donka. 1987. The prosodic character of early schwa deletion in English. In: Anna Giacalone-Ramat, Onforio Carruba and Giuliano Bernini (eds.). Papers from the 7th International Conference on Historical Linguistics. Amsterdam/Philadelphia: John Benjamins. 445–458. Minkova, Donka. 2000. Middle English prosodic innovations and their testability in verse. In: Irma Taavitsainen, Terttu Nevalainen, Päivi Pahta and Matti Rissanen (eds.). Placing Middle English in context. Berlin: Mouton de Gruyter. 431-461. Peck, Russell A. (ed.). 2000. John Gower: Confessio Amantis. Kalamazoo, Michigan: Medieval Institute Publications. Available online: http://www.lib.rochester.edu/camelot/teams/tmsmenu.htm. Last accessed: 23.05.2010. Robinson, Fred Norris (ed.). 1957. The works of Geoffrey Chaucer: Canterbury Tales. Boston: Houghton Mifflin. Available online: 1993. In: Corpus of Middle English Prose and Verse. University of Michigan. http://name.umdl.umich.edu/CT. Last accessed: 23.05.2010. Robinson, Ian. 1971. Chaucer’s prosody: a study of the Middle English verse tradition. Cambridge: CUP. Southworth, James Granville. 1947. Chaucer’s final e in rhyme. PMLA 62: 910–935. Southworth, James Granville. 1954. Verses of Cadence. Oxford: Basil Blackwell. Southworth, James Granville. 1962. The prosody of Chaucer and his followers. Oxford: Basil Blackwell.
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller
Werthmüller: An analysis of word-final -e in ME verse, in Gower and Chaucer 35 Southworth, James Granville. 1964. Chaucer: a plea for a reliable text. College English 26: 173–179 Smithers, Geoffrey Victor. 1983. The scansion of Havelok and the use of ME -en and -e in Havelok and by Chaucer. In: Douglas Gray and Eric Gerald Stanley (eds.). Middle English studies presented to Norman Davis in honour of his seventieth birthday. Oxford: Clarendon Press. 195–234. Tyrwhitt, Thomas. 1775. An Essay on the language and versification of Chaucer. In: The Canterbury Tales of Chaucer. In four volumes. London: T. Payne and Son. Vol. IV: 1– 111. Werthmüller, Gyöngyi. 2008. Middle English Iambic metre – intrametrical and linguistic problems. The Even Yearbook 8: 1–28. Available online: http://seas3.elte.hu/delg/publications/even/2008.htm. Last accessed: 23.05.2010. Windeatt, Barry A. 1977. “Most conservatyf the soun”: Chaucer’s Troilus metre. Poetica (Tokyo) 8: 44–60.
Gyöngyi Werthmüller Eötvös Loránd University
werthmuller.gyongyi@chello.hu
The Even Yearbook 9 (2010), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/delg/publications/even, © 2010, Gyöngyi Werthmüller