- 1 The Oxford English dictionary (OED) suggests 1602 for the first attestation of emotion in the sense (...)
1This paper reexamines results from several studies predominantly involving the language of everyday occupational domains in medieval England, and compares them with the lexis of anger, understood to be a domain expressing an emotion. As noted by Pons-Sanz [2022], the expression of anger in the Middle English period has been the subject of a great deal of scholarship (e.g., Gevaert [2007]; Diller [2014]). Much of this is devoted to understanding medieval conceptualizations of emotions, especially since the term emotion itself is not recorded in English until the sixteenth or early seventeenth centuries.1 The problem pervades the entire semantic field of terms relating to emotions, as Gevaert observes: “the danger of anachronism always lurks and methods must be found to verify if the conceptualizations assigned to expressions make sense historically speaking” [2007: 31]. In the absence of a generally accepted definition, particularly with reference to the historical context, Pons-Sanz elects to make use of the classification in the Historical thesaurus of English (HTE), though she makes the caveat that this methodology risks imposing a modern typology on the medieval data, since the HTE classification was designed to reflect the entire diachrony of English (cf. Sylvester, Marcus & Ingham [2017]).
2This paper examines questions associated with the lexicalization of the semantic domains in our dataset of Middle English lexis (our dataset is described below). In the literature on the vocabulary of emotion, there is some discussion of expressions for anger. There is some scholarship which discusses the reasons for new expressions in this field. Gevaerts [2007: 194] considers the possibility that stylistic requirements might have played a role, concluding that, while it is difficult to be sure about the demands of alliteration, anger was not introduced for reasons of rhyme. She is interested in tracing semantic differences between wrath and ire, and anger but finds that it is not possible to draw distinctions that hold across writers in her corpus (e.g., Chaucer and Langland) and cites Klaeber’s observation that the use of near-synonyms close together has the effect of neutralizing semantic differences between them. In this context we may note Weinreich’s suggestion that affective terms tend to lose their expressive force, creating a constant need for synonyms, what he terms an “onomastic low-pressure area” [1968: 58]. This idea is supported by our findings which show that anger is the most heavily lexicalized – and the only affective – domain in the corpus.
- 2 HTE category headings are in bold throughout.
3As detailed below, the lexical items in our dataset are tagged for language of origin. Our data shows that loanwords from French and / or Latin form a large proportion of the lexis below the headword anger, in particular in the densely lexicalized semantic categories from HTE such as Anger, To make angry, and Indignation or resentment.2 It is possible that the combination of the density of lexicalization expected in an affective category, and the predominance of loanwords, may indicate continued bilingualism in the Middle English period.
4Examining the non-literal expressions of anger, Gevaert suggests that “first attestations of a certain conceptualization can be considered transparent and motivated” [2007: 31], and that this also applies to new expressions showing the same conceptualization as those which have been in use for some time, since the conceptualization would not give rise to new, similar expressions if it were opaque [2007: 32]. Gevaert [2007: 188] concludes that metaphorical expressions are not as widespread as is assumed in cognitive linguistic scholarship. Our data, too, showed little evidence of the kinds of semantic shift we would expect at a time of high levels of lexical borrowing. In this context, we note the multilingual context of late medieval England, and Weinreich’s suggestions [1968: 58-59] about the need felt by bilingual speakers to borrow new terms. The notion of subjectivity has been proposed as an explanation for the mechanisms of semantic change (Traugott & Dasher [2001]), but though it seemed likely that an element of subjectivity would be found in a domain expressing strong emotion, despite our finding some differences between the lexis of anger and that of the other domains studied, in many respects the developments in the affective lexis mirrored those of the vocabulary in the domains of everyday life in medieval England.
5The paper first sets out the data and methodologies employed in the study (Section 2). In Section 3, we examine the density of lexicalization in the anger domain and compare it with the other domains. This section also examines levels of polysemy in anger compared with that of terms in the other domains. We thus compare relative numbers of lexical items and then the number of semantic categories in which the terms in each of the domains is found. Section 4 is concerned with long-term retention of lexis in the anger domain; specifically, we examine outcomes for the earliest attested term in a semantic category when it is joined by one, two or three (near-)synonyms. Outcomes for native terms are compared with those for loanwords. Finally, in Section 5 we examine a set of loanwords in order to investigate the range of senses in the source language(s) that is borrowed into Middle English; how far polysemy and semantic development of loanwords in Middle English are matched in the source language(s); and to what extent new senses develop in Middle English after a loanword has been borrowed.
6The following analyses are based on two datasets, one is used to examine lexicalization, polysemy and long-term retention of vocabulary; the other is used to investigate sense sharing across borrowing and source languages. In all cases, lexis relating to anger is compared with other, non-affective, domains.
- 3 Collected for the Bilingual thesaurus of everyday life in medieval England project. We are grateful (...)
- 4 Collected for the Technical language and semantic shift in late medieval English project. We would (...)
- 5 We are grateful to the Leverhulme Trust for funding for our current project, Semantics of word borr (...)
7The first dataset consists of 5818 terms attested in Middle English between ca. 1100 and 1500 collated from ten semantic domains: building, domestic activities, farming, food preparation, manufacture, trade, travel by water,3 hunting, medicine,4 and anger. This last domain was added for the current project because it seemed likely to provide more figurative language to study than the other nine domains.5
- 6 Note that only strictly domain-specific lexis was collected for the Bilingual thesaurus project (se (...)
8The vocabulary was classified into 2400 semantic categories and subcategories taken from HTE. These were arranged in a semantic hierarchy with the most general (hypernymic) terms at the top of each section, and the most technical (hyponymic) senses at the bottom, e.g., Emotion above Sudden access of emotion in anger; Stone / rock above Stone as material for paving; Animals above Domestic Animals in farming; and Providing / receiving food above Supplying food / catering in food preparation.6 Two examples of hierarchy sections are given below, showing semantic categories from anger and food preparation with co-hyponyms listed underneath:
9(1)
|
02. The mind
|
|
02.04. Emotion
|
|
blod c. 1275-1450+ Old English
|
|
talent c. 1330-1450 Latin;Old French
|
|
affeccioun c. 1390-1450 Latin;Old French
|
|
feling(e) c. 1395-1450+ Old English
|
|
sentement a. 1425 Latin;Old French
|
|
affect a1425-1450+ Latin
|
|
desir c. 1425-c1430 Old French
|
|
|
.An emotion
|
|
passioun a. 1250-1450+ Latin;Old French
|
|
issue c. 1350-1450+ Old French
|
|
meving(e) a. 1382-1450+ Old French;Anglo-French
|
|
spirit a. 1382-1450+ Latin;Old French;Anglo-French
|
|
mocioun c. 1390-1450+ Latin;Old French
|
|
affeccioun c. 1390-1450 Latin;Old French
|
|
feling(e)#1 a. 1425-1450 Old English
|
|
.Principle / power of an emotion
|
|
spirit a. 1387-1450+ Latin;Old French;Anglo-French
|
|
.Sudden access of emotion
|
|
acces(se) c. 1384 Latin;Old French
|
10(2)
|
01. The world
|
|
01.07. Food and drink
|
|
01.07.01. Food
|
|
01.07.01.21. Providing / receiving food
|
|
liveneth a. 1225-1340. ?Old Scandinavian
|
|
sustenaunce c. 1300-1450+ Old French;Anglo-French
|
|
feding a. 1387- c. 1450 Old English
|
|
norishing(e) c. 1387-1450+ Old French
|
|
ap(p)areilling c. 1390 Old French
|
|
refreshing(e) 1418-1450+ Anglo-French
|
|
incibacion ?a. 1425 Latin
|
|
provendringe a. 1450 Anglo-French
|
|
norishement 1450+ Old French
|
|
.Supplying food / catering
|
|
vitailing 1425-1450+ Old French;Anglo-French
|
|
advictailling 1435-1443 Old French
|
|
vitaillement 1450+ Old French
|
|
.Supplier of food / caterer
|
|
achatour 1240-1450+ Old French;Anglo-French
|
|
catourer 1281-1450+ Anglo-French
|
|
catour 1364-1450+ Anglo-French
|
|
vitailer c. 1380-1450+ Old French;Anglo-French
|
|
puveiour a. 1400-1450+ Anglo-French;Old French
|
- 7 This paper does not discuss the distribution of native and non-native words in the corpus across th (...)
11The etymology and dates of attestation during the Middle English period were recorded for each lexical item. This allows the gathering together of various types of data for each semantic category: how many words are used to express a particular sense in Middle English, the monosemy or polysemy of individual terms within the corpus and whether they undergo semantic shift, and the proportions of native terms and loanwords overall and at specific levels of the hierarchy.7 A selection of these semantic categories containing two, three or four co-hyponyms recorded in use in Middle English are examined to compare long-term survival rates (see Section 4).
- 8 Their classification as loanwords is based on the etymologies given in the Middle English dictionar (...)
- 9 For example, the French loanword croupe which was originally included in farming under Body parts o (...)
12The second dataset consists of a subset of 100 French and Latin loanwords selected from the 1759 borrowings from French and / or Latin in our corpus.8 Crucially, in these cases, all Middle English senses of the loanwords were recorded, regardless of the original semantic domain.9 The criteria used for selection mean that hapax legomena and highly polysemous borrowings are not included in this dataset. This ensures, first, that there is sufficient evidence of each borrowing’s use in Middle English; and, secondly, that semantic overlap with the source languages for polysemous loanwords could be reliably examined and quantified. With these two requirements in mind, it was decided that:
-
Each loanword must have between 1 and 8 senses recorded in Middle English-matrix citations before ca. 1450;
-
At least one sense must be recorded in at least three Middle English-matrix citations;
-
- 10 It is not unusual to find entries in MED which consist mainly or entirely of citations from sources (...)
Remaining senses must be recorded in at least one Middle English-matrix citation.10
- 11 Out of 1637 individual loanwords in the main corpus attested in Middle English by 1450, 57% are tag (...)
13These factors meant that it was not possible to ensure an even breakdown of ten loanwords from each of the ten original domains; however, all domains are represented in the loanword subset. Similarly, the distribution of source languages does not exactly mirror that of the main corpus, but the basic proportions are the same with the majority of etyma being tagged as French (73%), followed by French and / or Latin (21%), and then Latin (6%) as shown in Table 1, below.11
Table 1. Original domains and source language(s) of the 100-loanword subset
- 12 Domain combinations included eleven instances of two domains and one of three. All domains apart fr (...)
|
Original semantic domain
|
No. of senses of loanword in ME
|
Source language(s) of loanword
|
Total no. loanwords
|
Examples
|
|
Monosemes
|
Polysemes
|
French
|
French and / or Latin
|
Latin
|
|
Anger
|
1
|
12
|
7
|
6
|
0
|
13
|
irour
frounen
|
|
Building
|
2
|
6
|
6
|
2
|
0
|
8
|
carpentrie founden
|
|
Domestic Activities
|
2
|
5
|
5
|
1
|
1
|
7
|
fildor
beverage
|
|
Farming
|
4
|
7
|
9
|
2
|
0
|
11
|
cracche
jument
|
|
Food Preparation
|
1
|
3
|
4
|
0
|
0
|
4
|
endoren
viaund(e)
|
|
Hunting
|
1
|
6
|
6
|
1
|
0
|
7
|
bugle
les(se)
|
|
Manufacture
|
1
|
8
|
8
|
0
|
1
|
9
|
cade
foil
|
|
Medicine
|
2
|
11
|
5
|
5
|
3
|
13
|
opium
phisicien
|
|
Trade
|
1
|
9
|
7
|
2
|
1
|
10
|
aquitaunce
eschaunge
|
|
Travel by Water
|
2
|
4
|
6
|
0
|
0
|
6
|
barge
cable
|
|
2+ of the above domains12
|
2
|
10
|
10
|
2
|
0
|
12
|
pelur(e)
furnais(e)
|
|
Total
|
19
|
81
|
73
|
21
|
6
|
100
|
|
- 13 For Anglo-French: Anglo-Norman dictionary (AND); for Continental French: Französisches etymologisch (...)
14Once the 100 loanwords had been selected, senses in French / Latin recorded before ca. 1450 were also collated using the relevant historical dictionaries.13 Notes on attestation and obsolescence dates for each sense, text type and figurative use were also documented. To allow analysis of polysemic overlap, a “sense summary” was created for each loanword which maps the senses in all the languages onto a common framework using the semantic categories of the HTE. This allows us to identify which senses are borrowed from French / Latin, and which have developed independently in Middle English following the adoption of the loanword. Two examples of these summaries are given below for the loanwords peutre (from manufacture) and disdein(e) (from anger):
15(3) Loanword sense summary for peutre
|
MED [s.v. peutre]
< OF peautre, peutre, peaultre & ML peutrum, peautrum, peltrum.
|
OED3 [s.v. pewter]
< AF peutre and MF peautre
|
|
HTE senses:
|
ME
1348-1450+
|
BML
c. 1227-
1450+
|
AF
c. 1292-
1416
|
CF
c. 1195-1450+
|
|
Sense 1:
Pewter (< Alloy)
|
X
|
X
|
X
|
X
|
|
Sense 2:
Alloy of gold and silver [electrum] (< Alloy of precious metals)
|
X
|
|
|
|
|
Sense 3:
Zinc (< Base metal)
|
|
X
|
|
|
|
Number of shared senses between ME and source language(s):
|
1 of 3; One of Some
|
|
Native senses found only in ME:
|
1 = Alloy of gold and silver [electrum]
|
|
Figurative language:
|
None
|
- 14 Peutre is only used in this second sense as a gloss to Latin electrum in three glossaries: Medulla (...)
- 15 There is one citation of this sense from ca. 1227 (DMLBS [s.v. peltrum]). As with so many French-or (...)
16In the first example, the French loanword peutre is polysemous in Middle English and has two senses recorded in Middle English-matrix texts by 1450. The first sense (‘pewter, an alloy of tin and lead’) is borrowed from French and is recorded in Middle English, Anglo-French (AF), Continental French (CF) and British Medieval Latin (BML). However, the second (‘alloy of gold and silver, also known as electrum’) is only found in Middle English and so it is categorized as “native”.14 A third sense of the loanword (‘zinc’) is only recorded in British Medieval Latin.15 Of the three senses, only one is shared between Middle English and the donor language – hence this loanword is tagged “One of Some”. The full typology for categorizing semantic overlap is given in Section 4, below.
17(4) Loanword sense summary for disdein(e)
|
MED [s.v. disdein(e)]
< OF desdeigne
|
OED2 [s.v. disdain]
< OF desdeign, -daign, -daing, -dain, AF dedeigne
|
|
HTE senses:
|
ME
c. 1300-1450+
|
BML
s. xii-s. xv
|
AF
s. xiiex-s. xivin
|
CF
c. 1170-1450+
|
|
Sense 1:
Contempt (< Attention and judgement)
|
X
|
X
|
X
|
X
|
|
Sense 2:
Action of expressing contempt (< Contempt)
|
X
|
|
|
|
|
Sense 3:
Object of contempt (< Condition of being held in disesteem / contempt)
|
X
|
|
|
|
|
Sense 4:
Indignation or resentment (< Anger)
|
X
|
|
|
X
|
|
Sense 5:
Inflammation (< A disease)
|
X
|
|
|
|
|
Sense 6:
Loathing or detestation (< Hatred)
|
X
|
|
X
|
|
|
Number of shared senses between ME and source language(s):
|
3 of 6; Some
|
|
Native senses found only in ME:
|
3 = Action of expressing contempt; Object of contempt; Inflammation
|
|
Figurative language:
|
Metonymy: Action of expressing contempt; Object of contempt
Metaphor: Inflammation (‘angriness / irritation of wound’)
|
- 16 The metaphor is a sense hapax and is only recorded once ca. 1400. The metonymic senses are longer l (...)
18In the second example, the French loanword disdein(e) is also polysemous and is recorded in a total of six senses in Middle English by 1450. There are three senses which are shared between the borrowing and donor languages out of a total of six: the earliest sense, ‘contempt’ (which is also recorded in British Medieval Latin), followed by ‘indignation or resentment’, and ‘loathing or detestation’. The loanword is therefore categorized as “Some” for semantic overlap. Three further senses appear to have developed independently in Middle English following the borrowing of the loanword and they all involve figurative language: the metonymic extensions ‘action of expressing contempt’ and ‘object of contempt’, and the metaphorical ‘inflammation’, used to describe the angriness or irritation of a wound.16
19As noted above, the hierarchy corpus (i.e. Dataset 1) totals 5818 words, arranged under 2400 semantic categories. Of these, 542 words arranged under 93 sense categories belong to the domain of anger. When compared to the other domains, it is immediately noticeable that anger is the most heavily lexicalized domain in the corpus: the average number of co-hyponyms per semantic category is 5.8, compared to 2.4 for the corpus. This contrasts sharply with the other nine domains which all have very similar levels of lexicalization, as shown in Table 2.
Table 2. Average number of lexical items per semantic category across the ten domains
|
Domain
|
Lexical items
|
Semantic category
|
Average lexical items per sense
|
|
Anger
|
542
|
93
|
5.8
|
|
Building
|
673
|
309
|
2.2
|
|
Domestic Activities
|
395
|
172
|
2.3
|
|
Farming
|
929
|
395
|
2.4
|
|
Food Preparation
|
342
|
144
|
2.4
|
|
Hunting
|
481
|
243
|
2.0
|
|
Manufacture
|
634
|
331
|
1.9
|
|
Medicine
|
826
|
300
|
2.8
|
|
Trade
|
535
|
234
|
2.3
|
|
Travel by Water
|
461
|
179
|
2.6
|
|
Corpus
|
5818
|
2400
|
2.4
|
20When we look at the distribution of number of lexical items per sense in more detail, we see that anger is similar to the main corpus in that, in both cases, roughly a third of categories have between two and four words each (34% of anger compared to 37% overall). However, the domain has significantly fewer single-item senses (31% of anger compared to 52% overall) and more senses with five or more words per sense (34% of anger compared to 11% overall), as shown in Table 3.
Table 3. Distribution of number of words per sense, corpus vs. anger
|
Number of words per sense
|
Corpus
No.
|
%
|
anger
No.
|
%
|
|
1
|
1247
|
52%
|
29
|
31%
|
|
2 to 4
|
886
|
37%
|
32
|
34%
|
|
More than 5
|
267
|
11%
|
32
|
34%
|
|
5 to 10
|
217
|
9%
|
16
|
17%
|
|
11 to 20
|
40
|
2%
|
11
|
12%
|
|
More than 20
|
10
|
< 0.5%
|
5
|
5%
|
|
Total
|
2400
|
100%
|
93
|
100%
|
21Only a tiny majority of semantic categories (less than 0.5%) in the project hierarchy have more than twenty co-hyponyms. However, five out of these ten “super-lexicalized” senses belong to the domain of anger: Anger; Furious anger; To make angry; Indignation or resentment and To be / become furious / rage. The remaining five domains are building (which contributes three highly lexicalized semantic categories), medicine, and domestic activities. Table 4a sets out this information together with the languages of origin of the co-hyponyms.
Table 4a. Ten most lexicalized semantic categories in the dataset and languages of origin of co-hyponyms
- 17 The “Other” category here includes loans from Middle Dutch, mixed etymology compound nouns, and wor (...)
|
Domain and semantic category
|
No. of words
|
Old English
|
Old Norse
|
French
|
French and / or Latin
|
Latin
|
Other17
|
|
Anger
Anger (n.)
|
47
|
19
|
6
|
14
|
6
|
0
|
2
|
|
Building
To build and construct (v.)
|
41
|
23
|
4
|
6
|
3
|
1
|
4
|
|
Medicine
To heal / cure (v.)
|
40
|
19
|
1
|
8
|
9
|
3
|
0
|
|
Anger
Furious anger (n.)
|
38
|
21
|
5
|
6
|
5
|
0
|
1
|
|
Domestic Activities
Clothing (n.)
|
32
|
9
|
1
|
20
|
2
|
0
|
0
|
|
Anger
To make angry (v.)
|
31
|
16
|
2
|
8
|
4
|
0
|
1
|
|
Building
Wooden beam or timber (n.)
|
27
|
14
|
1
|
5
|
2
|
1
|
4
|
|
Anger
Indignation or resentment (n.)
|
25
|
5
|
2
|
12
|
4
|
0
|
2
|
|
Building
Wooden board or plank (n.)
|
23
|
11
|
1
|
2
|
2
|
0
|
7
|
|
Anger
To be / become furious / rage (v.)
|
21
|
12
|
1
|
6
|
0
|
0
|
2
|
22It is also important to note that all these super‑lexicalized senses are found at the top two hypernymic levels of the relevant sections of the hierarchy. This is to be expected as analysis of the wider corpus has shown a clear inverse correlation between the number of lexical items per semantic category and specificity of sense / technicality (Sylvester, Tiddeman & Ingham [2020: 20-22]).
- 18 The “Old Norse” category includes terms whose etyma have the following MED language tags: “Old Dani (...)
23When we examine the distribution of native and non-native terms, it can be seen that loanwords form a sizeable proportion of the lexis for these five highly lexicalized semantic categories in anger. Twenty out of forty-seven co‑hyponyms found in the category Anger are borrowings from French and / or Latin (e.g., felonie, folie, impacience, ire, irour, jelouste, spit(e), offense), as are twelve out of thirty-one in To make angry (e.g., achaufen, encensen, enticen, exciten, greven, meven, noien, provoken), and sixteen out of twenty-five in Indignation or resentment (e.g., despit, disdeine(e), displesir, grevaunce, grucched, indignacioun, maugre, rancour). Indeed, in this last example, French-origin terms outnumber native ones (twelve compared to five) which occurs for only one other semantic category: this is Clothing, where twenty out of a total of thirty‑two co‑hyponyms are French loanwords (e.g., amiture, ap(p)areil, atir, arrai, coverture, habit, raiment, vestment) compared to only nine native terms. This category (from domestic activities) is one where scholars would traditionally expect to find a high proportion of (Anglo)-French borrowings but, as Ponz-Sanz [2022: 366-367] points out, the field of emotion contains a surprisingly substantial amount of French-derived terms, and the anger subfield, in particular, “an interesting amalgamation of native, French- and Norse-derived words”. Terms classed as Old Norse in our corpus,18 which appear under these vocabulary-rich senses in Table 4a, include anger, angren, brathe eggen, greme, ournen thro, and sit(e).
- 19 Figures for the total numbers of French-origin loanwords across all semantic categories in these do (...)
24Table 4b shows the percentages of categories for the super‑lexicalized senses when they are condensed into their four domains of origin. Those located within the subfield of anger (28.4%) come second after domestic activities (62.5%) in proportions of French borrowings, but ahead of medicine (20.0%) and building (17.8%). Whilst this provides only a snapshot of the domains in question (and two of the domains are represented by a single semantic category), it is worth noting that these proportions are much closer to each other when we look at French borrowings across all sematic categories in these domains: anger (19.2%); domestic activities (21.0%); medicine (18.6%); building (17.5%).19 This suggests that, while the super‑lexicalized senses in medicine and building are representative of these wider individual domains in terms of French loanwords, the five semantic categories in anger (Anger; Furious anger; To make angry; Indignation or resentment and To be / become furious / rage) and one from domestic activities (i.e., Clothing) have especially high percentages of French terms amongst the numerous co-hyponyms attested in the late medieval period.
Table 4b. Ten most lexicalized semantic categories condensed by domain and languages of origin of co-hyponyms
|
Domain
|
Total no. Of words
|
Old English
|
Old Norse
|
French
|
French and / or Latin
|
Latin
|
Other
|
|
No.
|
No.
|
No.
|
No
|
No.
|
No.
|
No.
|
|
Anger
(5 semantic categories)
|
162
|
73
|
16
|
46
|
19
|
0
|
7
|
|
Building
(3 semantic categories)
|
91
|
48
|
6
|
13
|
7
|
2
|
15
|
|
Domestic Activities
(1 semantic category)
|
32
|
9
|
1
|
20
|
2
|
0
|
0
|
|
Medicine
(1 semantic category)
|
40
|
19
|
1
|
8
|
9
|
3
|
0
|
|
%
|
%
|
&
|
%
|
%
|
%
|
%
|
|
Anger
(5 semantic categories)
|
100%
|
45.1%
|
9.9%
|
28.4%
|
11.7%
|
0.0%
|
4.3%
|
|
Building
(3 semantic categories)
|
100%
|
52.7%
|
8.2%
|
17.8%
|
9.6%
|
2.7%
|
20.5%
|
|
Domestic Activities
(1 semantic category)
|
100%
|
28.1%
|
3.1%
|
62.5%
|
6.3%
|
0%
|
0%
|
|
Medicine
(1 semantic category)
|
100%
|
47.5%
|
2.5%
|
20.0%
|
22.5%
|
7.5%
|
0%
|
25We now turn to senses per word in Dataset 1, rather than words per semantic category. It was found that individual terms in anger are also more likely to be polysemous when compared to those in the corpus as a whole. Table 5 shows these results; note that this analysis was restricted to senses recorded within the ten semantic domains and does not include all senses recorded for each word in the Middle English period (as opposed to the loanword study which follows in Section 4.).
Table 5. Number of times each word appears under difference semantic categories within the corpus and within anger
|
Corpus
|
anger
|
|
Occurrence(s) of word in dataset
|
No. of words
|
Total no. of words
|
% of total
|
No. of words
|
Total no. of words
|
% of total
|
|
1 time
|
3649
|
3649
|
62.7%
|
269
|
269
|
49.6%
|
|
2 times
|
630
|
1260
|
21.7%
|
72
|
144
|
26.6%
|
|
3 times
|
158
|
474
|
8.1%
|
18
|
54
|
10.0%
|
|
4 times
|
66
|
264
|
4.5%
|
10
|
40
|
7.4%
|
|
5 times
|
25
|
125
|
2.1%
|
4
|
20
|
3.7%
|
|
6 times
|
4
|
24
|
0.4%
|
0
|
0
|
0%
|
|
7 times
|
2
|
14
|
0.2%
|
1
|
7
|
1.3%
|
|
8 times
|
1
|
8
|
0.1%
|
1
|
8
|
1.5%
|
|
Total
|
4535
|
5818
|
100%
|
375
|
542
|
100%
|
2663% of words in the corpus appear only once in a single semantic category, compared to 50% of words in anger. This means that 37% of words in corpus are polysemous (i.e., repeated in 2 to 8 additional semantic categories in the dataset), compared to 50% for anger. It can also be seen that most words occur between one and five times in the dataset; with only seven out of 4535 individual words (less than 1%) occurring between six to eight times.
27These words are given in Table 6. anger includes the most polysemous native word in the dataset, as well as the most polysemous non-native word. These are wratthe, which is found in eight semantic categories (Anger, Act done in anger, Fit(s) / outbursts of anger, Furious anger, Furious anger personified, Fury of the deity, Instance / fit of furious anger, Act done in indignation), and the French borrowing, greven, which is found in seven semantic categories (Be angry, Become angry, Be angry at / with, Make angry, Be / become furious / rage, Infuriate, Excite to indignation / resentment). Note that this is the only loanword amongst these seven highly polysemous words, the other six are all of Old English-origin: iren, ston, beten, washen and stagge.
Table 6. Most polysemous words in the dataset
|
Word
|
Occurrences of word in dataset
|
Domain(s)
|
|
wratthe (n.)
|
8
|
anger
|
|
greven (v.)
|
7
|
anger
|
|
iren (n.)
|
7
|
farming, food preparation, manufacture, medicine
|
|
ston (n.)
|
6
|
building, manufacture
|
|
beten (v.)
|
6
|
domestic activities, hunting, manufacture, medicine
|
|
washen (v.)
|
6
|
domestic activities, food Preparation, manufacture, medicine
|
|
stagge (n.)
|
6
|
farming
|
28Therefore, semantic categories in anger are more likely have numerous co-hyponyms when compared to the corpus average, but individual words in this domain are also more likely to be polysemous and appear within several semantic categories.
29The next set of results investigates the long-term outcomes for a selection of lexemes in anger recorded in Middle English to see if the widespread influx of (mainly) French borrowings prompted a large-scale replacement of existing terms. The investigation is an extension of an earlier study in which 1606 words from nine domains, divided into 453 2-item senses, 100 3-item senses, and 100 4-item senses (labelled as lexical pairs, trios and quads) are analysed (Sylvester, Tiddeman & Ingham [2022a]). That study found that, while replacement did occur when new terms entered into a semantic space, it was rarer than expected and certainly did not represent the accepted view of extensive relexification of English. The present paper examines the pairs, trios and quads relevant to anger which have since been added to Dataset 1, and compares outcomes with the main results of the nine-domain study.
- 20 For example, for the category Be / become irritated, Term 1 (angren, a. 1400) and Term 2 (chaufen, (...)
30There are thirty-two semantic categories in anger which have two, three or four co-hyponyms attested before 1500. This accounts for a third (34%) of the ninety-three semantic categories across the domain, almost the same total proportion of lexical pairs, trios and quads as found across the nine-domain corpus (a total 839 out of 2307 categories, i.e., 36%). Of these thirty-two semantic categories, one (Other types of anger) was excluded because the terms involved were co-hyponyms rather than synonyms, and a further nine were excluded because there was less than 25 years between the attestation of Term 1 and Term 2.20 This was to mitigate problems inherent in the use of dictionary citations as the basis for deciding whether one of two terms replaces the other (cf. Sylvester, Tiddeman & Ingham [2022a: 254]). The remaining data from anger consists of twenty-two semantic categories (eight pairs, ten trios and four quads) and a total of sixty-two words. This is a small sample compared to the nine-domain corpus but overall, it still allows us to obtain a general idea of Term 1 retention in the domain.
31The following typology (a condensed version of the one devised in Sylvester, Tiddeman & Ingham [2022a]) was used to categorize long-term semantic outcomes with the focus being on the earliest attested term (Term 1) in the group of two, three or four co-hyponyms. Examples of outcomes in each case are given from anger and from a selection of the other nine domains:
- 21 A word is defined as occurring in Present-Day English, if it is attested in the nineteenth century (...)
- 22 The Middle English period is defined here as 1100-1500.
32Term 1 drops out of use in a particular sense (relevant to our domains) before the Present-Day English period (PDE) 21following the arrival of incoming terms (Terms 2-4) during the Middle English period.22 At least one of Terms 2-4 remains in use until Present-Day English.
|
Pairs
|
anger: Action of raging
graming (a. 1225) is replaced by raging (ca. 1300–PDE)
Medicine: To make small incisions
garsen (a. 1398–16th c.) is replaced by scarifien (?a. 1425–PDE)
|
|
Trios
|
anger: One who / that which irritates
flie (c. 1230–17th cent.) is replaced by prik(e) (ca. 1386–PDE) and terrer(e) (a. 1425–1440) also falls out of use.
food preparation: Art of cooking
curie (a. 1387–16th cent.) is replaced by cokerie (a. 1393–PDE) and kichen(e) (c. 1400) also falls out of use
|
|
Quads
|
anger: To speak angrily
speken with onde (a. 1275) is replaced by ragen (a. 1499–PDE) and blusteren (c. 1422-PDE) and speken with passioun (ca. 1454) also falls out of use.
farming: Young pig
faren (OE–14th cent.) is replaced by pigge (a. 1250–PDE), grice (ca. 1230–PDE) and hogling (1377–PDE)
|
33Term 1 is joined by incoming terms (Terms 2-4) in a particular sense during the Middle English period, all or some of which then drop out of use before the Present-Day English period.
|
Pairs
|
anger: Angry speech
misword (ca. 1225-PDE) is not replaced by crabbednes(se) (ca. 1450)
Manufacture: Glass / crystal vessel
glas (ca. 1230–PDE) is not replaced by vitre (ca. 1450–16th cent.)
|
|
Trios
|
anger: Heat of anger
fir (ca. 1340–PDE) is not replaced by swelme (ca. 1400–1450) or by ferventnes(se) (ca. 1450)
domestic activities: Cider
sider (ca. 1350–PDE) is not replaced by pommade (ca. 1400) or by pomis (ca. 1450)
|
|
Quads
|
anger: To frown / scowl at
louren (ca. 1300–PDE) remains in use and is joined by gloumen on (a. 1425–PDE) but bilouren (ca. 1390) and scoulen on (ca. 1451-16th cent.) fall from use
hunting: Bird-lime
lim (OE–PDE) remains in use and is joined by visc (?a. 1425–PDE) and brid-lim (a. 1425–PDE) but gleu (ca. 1400–18th cent.) falls out of use
|
34Term 1 is joined by incoming terms (Terms 2-4) in a particular sense during the Middle English period. All terms go on to exist as (near) synonyms until the Present-Day English period.
|
Pairs
|
anger: To make sour / bitter
bitt(e)ren (ca. 1230–PDE) remains in use alongside souren (a. 1393–PDE)
manufacture: Wire-maker / worker
wir-drawer(e) (1368–PDE) remains in use alongside wir-smith (1438–PDE)
|
|
Trios
|
anger: No instances of this outcome in the sample
farming: Plough
sulou (OE–PDE) remains in use alongside plough (ca. 1175–PDE) and sul (a. 1225–PDE)
|
|
Quads
|
anger: No instances of this outcome in the sample
trade: Group / body of merchants
gild(e) (OE–PDE) remains in use alongside hanse (a. 1135–PDE), compaignie (1389–PDE) and livere (a. 1422–PDE)
|
35Term 1 is joined by incoming terms (Terms 2-4) in a particular sense during the Middle English period. One or all of the terms then undergoes semantic change through narrowing, broadening or metonymy prior to 1500.
|
Pairs
|
anger: Grudge
envie (ca. 1300–15th cent.) is joined by querele (1340–PDE); envie undergoes metonymic shift to mean ‘an instance of this feeling’ (a. 1393)
domestic activities: Brush / broom
besom (ca. 1000–?19th cent.) is joined by brom (1346–PDE); besom narrows to mean ‘broom made of specific material, e.g., the plants, broom or birch’ (ca. 1400–19th cent.)
|
|
Trios
|
anger: No instances of this outcome in the sample
building: Cement / mortar
lim (OE–PDE) is joined by morter (a. 1300–PDE) and ciment (ca. 1330–PDE);
morter broadens to mean ‘any substance that resembles or serves a similar purpose to mortar esp. plaster’ (1440–PDE)
|
|
Quads
|
anger: No instances of this outcome in the sample
farming: Domestic fowl collectively
foul (1131–PDE) is joined by pullain (1329–19th cent.), pultrie (1372–PDE) and polaille (ca. 1400–17th cent.); pultrie undergoes metonymic shift to mean ‘place where domestic fowl are sold’ (1423–PDE, now exists in place names only)
|
36All terms are hapaxes in a particular sense, attested once or in a single text during the Middle English period.
|
Pairs
|
anger: No instances of this outcome in the sample
medicine: Sharp / pointed instrument
pointel (?a. 1425) and radie (?a. 1425)
|
|
Trios
|
No instances of this outcome in the sample
|
|
Quads
|
No instances of this outcome in the sample
|
37Tables 7a and 7b show the distribution of outcome type across the 2-item, 3-item and 4-item semantic categories in the nine-domain corpus, and across anger.
Table 7a. Number and percentage of semantic outcomes across nine-domain corpus
|
Outcome:
|
Type 1
|
Type 2
|
Type 3
|
Type 4
|
Type 5
|
Total
|
|
Pairs
|
93
|
193
|
127
|
22
|
18
|
453
|
|
Trios
|
30
|
59
|
7
|
4
|
0
|
100
|
|
Quads
|
32
|
53
|
10
|
5
|
0
|
100
|
|
Total
|
155
|
305
|
144
|
31
|
18
|
653
|
|
24%
|
47%
|
22%
|
5%
|
3%
|
100%
|
Table 7b. Number and percentage of semantic outcomes across anger
|
Outcome:
|
Type 1
|
Type 2
|
Type 3
|
Type 4
|
Type 5
|
Total
|
|
Pairs
|
1
|
5
|
1
|
1
|
0
|
8
|
|
Trios
|
3
|
7
|
0
|
0
|
0
|
10
|
|
Quads
|
2
|
2
|
0
|
0
|
0
|
4
|
|
Total
|
6
|
14
|
1
|
1
|
0
|
22
|
|
27%
|
64%
|
4.5%
|
4.5%
|
0%
|
100%
|
- 23 This result mirrors findings from the Technical language and semantic shift in Middle English proje (...)
38A key result is that, as in our previous study, Type 2 (Term 1 remains in use and is not replaced upon the arrival of incoming terms into the semantic space) is the most common scenario. However, the proportion of outcomes categorized as Type 2 is considerably higher in anger than it is for the nine-domain corpus: 64% Type 2 outcomes compared to 47%. Type 1 outcomes (Term 1 is replaced by one or more incoming terms) is roughly the same: 24% for the 9-domain corpus and 27% for anger. Similarly, outcomes involving narrowing, broadening and metonymy (Type 4) are rare in both sets of data, accounting for only 5% in both cases.23 Finally, Type 3 outcomes (where all terms remain in use as synonyms) are much less prevalent in anger making up only 5% of outcomes, as opposed to 22% for the nine-domain corpus. This indicates higher levels of redundancy of incoming terms in anger as suggested by Weinreich [1968: 58].
- 24 As the sample size for anger is small (22 categories), it was not productive to break the data into (...)
39Next, only the three main outcomes are considered, Types 1, 2 and 3, which account for 93% of results in the nine-domain corpus and 96% of those in anger. These results are given in Tables 8a and 8b and are divided into two subgroups based on whether Term 1 in each pair, trio or quad is native or a loanword. For example, in the lexical trio quaken / resen / tremblen (‘to shake with anger’), Term 1 is native (and in this case, Terms 2 and 3 are loanwords) whereas in gref / displeasaunce / displesir (‘cause of indignation’), Term 1 is a French borrowing (as are Terms 2 and 3).24
Table 8a. Main outcomes across nine-domain corpus based on language of origin of Term 1
|
Outcome
|
Type 1
|
Type 2
|
Type 3
|
Total
|
|
Term 1
Native
|
82
23%
[Native term replaced by loanword = 56]
|
182
52%
|
87
25%
|
351
100%
|
|
Term 1
Loanword
|
73
29%
|
123
49%
|
57
23%
|
253
100%
|
|
Total
|
155
26%
|
305
50%
|
144
24%
|
604
100%
|
Table 8b. Main outcomes across anger based on language of origin of Term 1
|
Outcome
|
Type 1
|
Type 2
|
Type 3
|
Total
|
|
Term 1
Native
|
5
33%
[Native term replaced by loanword = 3]
|
9
60%
|
1
7%
|
15
100%
|
|
Term 1
Loanword
|
1
17%
|
5
83%
|
0
0%
|
6
100%
|
|
Total
|
6
28%
|
14
67%
|
1
5%
|
21
100%
|
40Once again, the main results from the nine-domain study are replicated in the domain of anger. The latter sample is small compared to the former, as we have noted, but similar overall patters can be discerned. In both datasets, Type 2 (failed replacement) is the most common outcome, regardless of whether the existing term (Term 1) is native or non-native. It can also be seen in both cases that borrowed Term 1s are more likely to be replaced by incomers (Type 1 outcome) than native Term 1s: the replacement rates for existing terms of Old English origin is 23% compared to 29% for existing terms which are loanwords in the nine-domain corpus, and 33% compared to 17% in anger. These figures suggest that the key findings of our previous study are replicated in an affective domain: rather than a widespread ousting of native terms by newly arrived loanwords in the Middle English period, such replacement is relatively rare. Out of a total 351 cases in the larger dataset where Term 1 is native, the latter is eventually replaced by (an) incoming loanword(s) in only fifty-six cases (16%): e.g., native breden is replaced by borrowed frien (‘to fry’); out of the fifteen instances in anger where Term 1 is native, replacement by loanword(s) occurred in three cases (20%): e.g., native speken with onde is replaced by ragen and blusteren (‘to speak angrily’).
41The final section of analysis examines 100 French / Latin loanwords in all their senses (i.e. Dataset 2) to investigate whether there are any differences in the rates of sense innovation in Middle English between loanwords selected from the domain of anger (such as disdein(e), fervour, frounen, irour, impacience, ire, maltalent, rancour) and those from the other nine non-affective domains (such as cisours, endoren, gleu, keie, retail(e), restoracioun, sicamour(e), phisik(e)).
- 25 Note that the average number of Middle English-matrix citations per sense is higher for anger (7.2) (...)
42As noted above, the 100 loanwords include nineteen monosemes and eighty-one polysemes recorded in Middle English-matrix sources by 1450. Polysemes range from 2-sense to 8-sense loanwords recorded in 1808 citations, with a total of 327 senses overall in the dataset. anger accounts for one monosemic loanword, twelve polysemic loanwords, totalling 384 citations and fifty-three senses, as shown in Table 9.25 It should also be noted that the loanwords selected are not ‘lexical rarities’ and seventy-four loanwords from of the non-affective corpus (74%) and ten from anger (77%) are still attested in Present-Day English in at least one sense.
Table 9. Number of words, senses and citations in the 100-loanword corpus and in Anger
|
Corpus
|
anger
|
|
No. of words
|
No. of senses
|
No. of citations
|
No. of words
|
No. of senses
|
No. of citations
|
|
1 sense
|
19
|
19
|
97
|
1
|
1
|
3
|
|
2 senses
|
22
|
44
|
291
|
4
|
8
|
55
|
|
3 senses
|
15
|
45
|
247
|
0
|
0
|
0
|
|
4 senses
|
17
|
68
|
363
|
2
|
8
|
72
|
|
5 senses
|
18
|
90
|
419
|
3
|
15
|
79
|
|
6 senses
|
4
|
24
|
123
|
1
|
6
|
55
|
|
7 senses
|
3
|
21
|
131
|
1
|
7
|
53
|
|
8 senses
|
2
|
16
|
137
|
1
|
8
|
67
|
|
Total
|
100
|
327
|
1808
|
13
|
53
|
384
|
43As detailed in the Section 2.2., Dataset 2 was created to trace the semantic development of loanwords in Middle English and to compare their meanings with those in French and Latin. Smaller studies by Durkin & Allan [2016], Durkin [2018] and Ingham [2021 and this volume] have found that cases of polysemy are nearly always mirrored between English and the source language. Our 100-loanword analysis provided similar results. The loanwords relating to anger pattern similarly, as can be seen in Table 10. In the eighty-one cases where the word is a polyseme in Middle English, the word is also a polyseme in French / Latin in seventy cases, so there is a match rate of 96%. Similarly, all twelve polysemic loanwords in anger are also polysemes in the source language(s).
Table 10. Number of Middle English loanwords which match source language(s) in terms of polysemy
|
Loanword is:
|
Corpus
|
anger
|
|
Monosemous in source language(s) and in ME
|
10
|
0
|
|
Polysemous in source language(s) and in ME
|
78
|
12
|
|
Monosemous in source language(s) and polysemous in ME
|
3
|
0
|
|
Polysemous in source language(s) and monosemous in ME
|
9
|
1
|
|
Total
|
100
|
13
|
44Next, we move beyond a binary monosemous / polysemous comparison and investigate the extent of semantic overlap between Middle English and the donor languages. This was achieved using the sense summaries compiled for each loanword (see examples in the Methodology section) which record whether individual senses were present in Middle English, French and Latin. The following categories were then devised to label sense-sharing scenarios:
-
One of one: the loanword has only one recorded sense and it is shared between Middle English and the source language(s): e.g., keie has one sense (‘wharf / quay’) recorded in French, Latin and Middle English.
-
One of some: the word has one sense shared between Middle English and source language(s), and the total number of recorded senses is more than one: e.g., irour has one sense (‘anger’) shared between French and Middle English but two additional senses (‘hatred’ and ‘distress’) recorded only in French.
-
Some: the word has two or more senses shared between Middle English and the source languages(s): e.g., rosari(e) has four senses shared between Latin and ME (‘rose-garden’; ‘place of supreme happiness’; ‘encyclopaedic work’; and ‘specific false coin’) but an additional sense (‘pattern of flowers’) recorded only in Latin.
-
All: Middle English shares all the senses found in the source language(s): e.g., sirup has two senses (‘syrup or linctus [medicinal]’ and ‘syrup [culinary]’), both recorded in French, Latin and Middle English.
-
- 26 This is probably due to a gap in the lexicographical record. According to the current OED entry [s. (...)
None: None of the senses found in the source languages(s) are found in Middle English (i.e., Middle English only has recorded senses of the word which are unique): e.g., disjoint(e) has two senses (‘predicament or straits’ and ‘great manifestation of feeling’) which are recorded in Middle English only. The original sense (‘disjunction, disunion or disconnection’) is found in French / Latin but is not recorded in Middle English.26
45Once again, loanwords in anger exhibit the same characteristics as those found across the other domains. As results in Table 11 show, complete semantic overlap is rare in the corpus (only ten out of eighty-one Middle English polysemes in the corpus share all their senses with the source languages, i.e., 12%), and non-existent in anger at 0%.
Table 11. No. of semantic overlap scenarios in the corpus and in anger
|
Semantic overlap
|
Corpus
|
anger
|
|
All
|
10
|
0
|
|
Some
|
62
|
10
|
|
One of one
|
7
|
0
|
|
One of some
|
19
|
2
|
|
None
|
2
|
1
|
|
Total
|
100
|
13
|
46When senses do not match up (which occurs in the majority of cases), it is either because the loanword has senses in the donor languages which are not borrowed into Middle English, or because new native senses develop independently in Middle English following the adoption of the loanword into the language.
47The final section of analysis considers homegrown loanword senses: e.g., disdein(e) (‘action of expressing contempt’); disjoint(e) (‘great manifestation of feeling’); gleu (‘bitumen / pitch’), gruel (‘poultice’); pipin (‘type of apple’); quirre (‘bag / collection of game’); sclice (‘surgical probe’); turnour (‘basket-maker’). It was found that the percentage of senses unique to Middle English (i.e., not attested in French and / or Latin) is nearly the same when anger was compared to the 100-loanword corpus: i.e., roughly one in five senses recorded in Middle English for a loanword is categorized as native. This accounts for seventy-one out of 327 senses (22%) of senses for the corpus and ten out of 53 (19%) for anger.
- 27 It is important to reiterate that the figures in the anger sample are low and that 70% in this inst (...)
48However, as Table 12 shows, 100% of native senses in anger are either metaphors or metonyms, compared to 60% for the corpus, with the metaphor rate being particularly high at 70%),27 compared to just 30% for the corpus. Examples of native metaphors from anger in the dataset include fervour (‘inflammation’ < ‘furious anger’), frounen (‘to present a gloomy aspect [of inanimate object]’ < ‘to frown’), noli me tangere (‘ill-tempered person’ < ‘eruptive disease’); and raumpen (‘to be / become furious’ < ‘to rise up on the hind legs’). Examples of native metaphorical senses from other non-affective domains include caudel (‘sloppy food’ < ‘caudle’); dauben (‘to soil’ < ‘to plaster’); meselri(e) (‘sin of gluttony’ < ‘leprosy’); and somer (‘bearer / porter’ < ‘packhorse’).
Table 12. Number and percentage of native / shared senses which are metaphorical and metonymic
|
|
|
Corpus
|
anger
|
|
Total no. of loanword senses in ME
|
327
|
53
|
|
No. of native ME senses
|
71
|
10
|
|
% of total senses in ME
|
22%
|
19%
|
|
No. of native metaphors
|
21
|
7
|
|
% of all native senses
|
30%
|
70%
|
|
No. of native metonyms
|
21
|
3
|
|
% of all native senses
|
30%
|
30%
|
|
No. of other native senses
|
29
|
0
|
|
% of all native senses
|
40%
|
0%
|
|
No. of ME senses shared with French / Latin
|
256
|
43
|
|
% of total senses in ME
|
78%
|
81%
|
|
No. of shared metaphors
|
38
|
8
|
|
% of all shared senses
|
15%
|
19%
|
|
No. of shared metonyms
|
33
|
4
|
|
% of all shared senses
|
13%
|
9%
|
|
No. of other shared senses
|
185
|
31
|
|
% of all shared senses
|
72%
|
72%
|
- 28 Examples of shared metaphor from anger are ire (‘heat’ < ‘anger’) and malencoli(e) (‘emotions of me (...)
49Crucially, when the levels of shared metaphor (i.e., those which are also recorded in French / Latin) are compared,28 it can be seen that figures for the corpus and for anger are much closer together at 15% and 19%, respectively. Therefore, whilst metaphors are more common amongst native senses than amongst shared ones in both the corpus and in anger, the domain of anger has a notably higher proportion of metaphors amongst those senses which have developed independently in Middle English. Anger was included in the current project in the hope that it would provide more figurative language to analyse than the non-affective domains. These figures from a sample of anger loanwords suggest that this is indeed the case but that metaphors are proportionally more prevalent amongst homegrown senses than among those borrowed from French / Latin.
50Various analyses have been carried out to compare the domain of anger to a larger corpus including nine non-affective domains. Results showed that anger is more heavily lexicalized: the average number of lexemes per sense is 5.8 (compared to 2.4 for the corpus) and only 31% of semantic categories are single-item (compared to 52% for the corpus). Five of the most heavily lexicalized categories (i.e. those with twenty or more co-hyponyms) in the dataset belong to this domain. Co-hyponyms in these super‑lexicalized categories from anger contained particularly high proportions of French borrowings (28.4%), whereas the overall percentage of French-origin loanwords across the domain (19.2%) was very similar to that found across domestic activities, medicine and building. Individual lexemes in anger are more polysemous, with 50% occurring under more than one semantic category (compared to 37% for the corpus), and two out of the seven most polysemous words in the dataset belonging to this domain.
51In terms of long-term retention of existing lexis, outcomes for anger appear similar to those for the non-affective domains. An examination of two‑, three‑ and four‑item senses found that the earliest recorded term (Term 1) tends to remain in use following the arrival of new vocabulary in the Middle English period, with only 27% being ultimately replaced by Present-Day English (compared to 24% for the corpus). In all cases, Type 2 (where incoming terms fail to replace Term 1) was the most common outcome (64% for anger and 47% for the corpus), regardless of whether Term 1 was of native or non-native origin. Cases where a native term were replaced by a loanword were relatively rare in both anger (20%) and the non-affective domains (16%). However, outcomes where all terms in the semantic category remain in use until Present-Day English (i.e. long-term synonymy) were comparatively very low in anger and accounted for only 5% of cases (compared to 22% for the corpus).
52Finally, French / Latin loanwords in anger were found to exhibit similar trends to the other domains in terms of sense sharing with their source language(s). Polysemy in the borrowing and donor languages was nearly always mirrored but complete semantic overlap (where all senses in all languages matched up) was rare in the corpus (12%), and non-existent in anger. All domains had roughly the same proportion (around one in five) of native loanword senses, i.e. senses which developed independently in Middle English and were not recorded in French / Latin. Notably, the metaphor rate amongst these homegrown senses in anger was particularly high at 70% (compared to 30% for the corpus).
53The investigations in this paper were undertaken in order to discover if the semantic and lexical development of an affective domain, anger, differed from the kinds of developments seen in domains of everyday life, and of an elite and a professional occupation. We were also interested in any differences between the affective and non-affective domains in connection with the impact of the influx of French word borrowing that occurred in the later medieval period.
54Weinreich [1968: 59] observes that comparison by a population of bilingual speakers between their language and one with which they are in contact may lead to the feeling that some of their semantic fields are insufficiently differentiated prompting them to borrow near-synonymous terms. This also fits in with our finding that individual loanwords in anger are more likely to be polysemous and to appear within several semantic categories within the domain. Note, however, that the polysemy found in loanwords in anger is mirrored by polysemy in the source languages (French and / or Latin), though complete semantic overlap is rare, pointing to the development of new senses in Middle English. Our results with regard to density of lexicalization support Weinreich’s [1968: 58] suggestion that the vocabulary of affective domains is perceived by speakers as being in constant need of renewal. The differing results with regard to the amount of loanwords in the anger domain also appear to give us a clue about the continued presence of speakers familiar with French in the Middle English period.
55In other respects, our results in relation to the vocabulary in anger mirror those found for the other semantic domains studied. While this seems to be a purely negative finding, it is important in the context of generally accepted ideas about the affective lexicon. We need to have a clear sense about what is particular about the emotional lexicon and what about it is true of the development of the vocabulary of English in a time of a trilingual textual culture. Our findings support those of Gevaert [2007], who found very little evidence of semantic shift through metaphorization in her corpus. Our results indicate that, despite the influx of French, vanishingly few of the native terms underwent semantic shift, in particular narrowing and broadening, in the Middle English period, even in domains such as anger, which have more loanwords than the average for the corpus. It is also the case that these borrowings did not oust the native terms within the Middle English period, again, even under the pressure of a greater amount of borrowing seen in the affective domain.
RSS feed
