Skip to navigation – Site map

HomeIssues16PapersAn Analysis of French Borrowings ...

Papers

An Analysis of French Borrowings at the Hypernymic and Hyponymic Levels of Middle English

Louise Sylvester, Megan Tiddeman and Richard Ingham

Abstracts

This paper analyses a large dataset of Middle English vocabulary from nine domains which has been arranged into a semantic hierarchy. It focuses on the distribution of French-origin borrowings at various levels of technicality and at various levels of co-hyponymic density (i.e. the number of words per sense). Overall, results show that French loanwords are concentrated in higher proportions at the hypernymic (or more general) level rather than at the hyponymic (or more technical) level. These findings run counter to the orthodox view that borrowings are used to fill lexical gaps for new technical terms in a semantic field.

Top of page

Full text

Introduction

  • 1 We are grateful to the Leverhulme Trust for funding this project across 2017-2020.

1This paper reports on findings from our project Technical language and semantic shift in Middle English.1 The project aims to address unanswered questions about word borrowing, obsolescence and semantic shift by looking at a large corpus of terms arranged in a semantic hierarchy. One of our aims is to examine the effects on Middle English vocabulary of the contact between French and English following the Norman Conquest. The main focus is on the levels of the semantic hierarchy at which French borrowings are found. In particular, we are interested in the technical register; that is, whether the contact with French impacted on English at the lowest levels, where the terms with most precise senses are found.

2Early discussions of the motives for lexical borrowing tended to highlight speakers’ perceptions of gaps, “the need for names for certain objects, concepts, etc. with which a language community is newly faced in a changed cultural situation” [Weinreich 2011: 53] (see also Weinreich [1953: 60]; Emeneau [1962: 431]; Grosjean [2010]). The second commonly discussed motivation is that of prestige attached to the source language (Weinreich [2011: 53]; Matras [2009: 168]; Campbell [2013: 58]). A third reason, not relevant to our data, is stylistic, the need to create special effects or to avoid homonymic clash (Weinreich [1953: 60]; [2011: 54]; Matras [2009: 168]).

  • 2 I argue elsewhere that multilingualism seems to be key to standardisation of the lexicon, since it (...)

3A further reason for focusing on technical lexis is that the later Middle Ages saw numerous innovations in many occupational domains, including metalworking, construction and domestic work, leading to the potential need for new conceptualisations. At this stage in its history, English had not yet reached the level of linguistic development characterising an Ausbau language; that is, “an autonomous standard variety together with all the nonstandard varieties from the dialect continuum which are heteronomous with respect to it” [Trudgill 1992: 169]. One of the markers of the final level of an Ausbau language is a technical register (Kloss [1967: 29]), and the Middle English period sees the emergence of a precise vocabulary, pointing to the emergence of a technical register fuelled by the multilingual situation of medieval Britain (Sylvester forthcoming)2. Arguing that words of low frequency are less stable than frequently used terms, and are thus more subject to obsolescence and replacement, Weinreich offers an example of a technical term: the word denoting ‘the place on a scythe where the blade is attached to the handle’ in dialectal Russian. He reported that “the infrequent and unstable designations for parts of tools, which vary greatly from one locality to the next, are, in the Finnish contact area, represented by Finnish loanwords” [Weinreich 1953: 53]. This example is in line with early suggestions that a significant proportion of the French borrowings in Middle English were technical terminology (Serjeantson [1935]; Prins [1941]). A question worth addressing, then, is the role of a technical register in speakers’ everyday lives. There is evidence that “concepts pertaining to the immediate surroundings: orientation in space, time and quantity, the private domain of mental and physical activity, and the nearest human environments (body and close kin)” have a greater stability, while those involving negotiation with other people are more open to borrowing. Matras’s investigation of Romani and studies of other languages including Selice, Maltese and Old English, suggest the likelihood of a proximity constraint on borrowing that overrides other motivations such as fashion and prestige (Matras [2009: 169-172]). This paper takes a new approach to these issues through the examination of the semantic hierarchical levels at which loanwords are found.

4Use is made here of Laura Wright’s [1995] suggestion that technicality equates to greater specificity of meaning and we investigate the levels at which French borrowings occur. A pilot study (Sylvester [2018]) found that the impact of French on the native lexicon in a small set of data relating to the semantic domain field of Building was most evident at the higher (superordinate and basic) levels of the lexicon, which had almost equal numbers of native and borrowed terms, while at the hyponymic level native terms were in the vast majority. This finding indicated resistance to borrowed vocabulary not at the lowest section of the social stratum, but rather by the class of skilled workers who would have made use of technical vocabulary in their work. This suggests that the most technical vocabulary (within the occupational domain of Building) was understood by speakers as forming part of their intimate, everyday lexis and was thus resistant to French borrowing, despite the prestige of French in this period. It was clear, however, that a larger set of data was needed for further investigation of the levels of the semantic hierarchy in which French impacted the vocabulary of Middle English.

1. Methodology

1.1. Measuring technicality

  • 3 We would like to express our grateful thanks to Dr Harry Parkin (Research Fellow for the first half (...)

5Our dataset totals 5276 words and 2307 senses from the augmented corpus of the Bilingual Thesaurus of Everyday Life in Medieval England (BTh). These have been arranged into an extensive semantic hierarchy, modelled on the taxonomic categories devised for the Historical Thesaurus of English (HT).3 This organisational system is made up of seven broad divisions or Hierarchical Levels (HLs) which arrange words on the basis of hyponymic relations – the most general, or superordinate terms, are found at the top and the most technical, or most specific in meaning, are found at the bottom.

6In the example below from the domain of Manufacture, we can see thirteen words with the sense Worker with skins/hides which are found at HL7 within the hierarchy. Each HL is then further sub-divided into a maximum of five Category Levels (CLs), with hypernyms at CL0, at the very top, and hyponyms found underneath at CL1 through to CL4. Below we can see the same thirteen words arranged across the CLs within HL7, allowing us to identify increasing levels of technicality within the overall sense of Worker with skins/hides. Skinner(e), pelliper and pelter are the most general at CL0 with coveiser one level below at CL1, meaning Worker with leather. This sense is then split into different kinds of leather workers (Tanner, One who tans leather and One who dresses leather), giving us additional lexis at CL2: tanner(e), barker, tauier, teuer(e), curr(e)iour. Finally, the most technical vocabulary in this sub-category of the hierarchy is found at CL3 under senses designating specific kinds of leather tawers who work with white leather or grey pelts: scoudere, whit-tawier(e), whit-lether teuere, and grei-tauier.

Hierarchy example 1. Extract from Manufacture showing HLs and CLs

2. Society

[HL1]

2.6. Occupation and work

[HL2]

2.6.1. Worker

[HL3]

2.6.1.1. Workers according to type of work

[HL4]

2.6.1.1.1. Manual/industrial worker

[HL5]

2.6.1.1.1.3 Workers with specific materials

[HL6]

2.6.1.1.1.3.1. Workers with skins/hides

[HL7] [CL0]

skinner(e) 1255-1450+ Old English

pelliper 1391-1450+ Latin

pelter 1318-1450+ Old French

.Worker with leather 

[CL1]

corveiser c1130-1450+ Anglo-French

..Tanner 

[CL2]

barker 1250-1450+ Old Scandinavian

tanner(e) a1325-1450+ Old English;Old French;Anglo-French

..One who taws leather

[CL2]

tauier 1320-1450+ Old English

teuer(e) 1440-1450+ Old English

...One who prepares or dresses white leather

[CL3]

scoudere 1287 ?Latin;?Old French

whit-tawier(e) 1333-1450+ Old English-Old English

whit-lether teuere 1384 Old English-Old English Old English

...One who prepares or dresses grey pelts

[CL3]

grei-tauier 1381 Old English;Old Scandinavian-Old English

..One who dresses leather

[CL2]

curr(e)iour 1286-1450+ Old French

  • 4 To take just two examples from the wider hierarchy: ‘duck’ (HL4) and ‘egg’ (HL6) which we would con (...)

7For the purposes of this study, we will be focusing on CLs as indicators of technicality, rather than on HLs. Clearly, moving from Worker (HL3) to Worker with skins/hides (HL7) involves increased specificity of reference, but HLs are not a consistent or quantitative indicator of technicality in individual cases and at any specific level.4 This is because they were designed to cover a huge range of lexis (i.e. every word in OED2) across the whole diachrony of English, as well as providing a taxonomy for the semantic fields within the HT. Overall, the broader the domains become, the greater the number of possible ways to organize them and, as the HT’s editor highlighted, there is “no perfect or inevitable way of classifying the world or the lexis which refers to it” [Kay 2004: 67]. Therefore, we should avoid attributing much inherent meaning to the fact that words under Ploughing Equipment and Equipment for Food Preparation are classed as HL5 in the hierarchy whereas those under Metal-working Equipment and Hunting Equipment are apparently less technical at HL4. The differences in HL value in these cases are simply due to the hierarchy’s wider framework. When carrying out cross-domain analysis of technicality of meaning, it is much more useful to compare CLs independently of HLs – e.g. plough-gere (‘ploughing equipment’) and harneis (‘hunting equipment’) are both general CL0 terms at the very top of their sub-category under Ploughing Equipment and Hunting Equipment respectively, and peni-bred (‘bread board for penny loaves’) and led panne (‘pan for melting lead’) are both found at CL3 and are the most technical in their sub-categories under Equipment for Food Preparation and Metal-working Equipment, respectively. We recognise that CLs do not offer a flawless solution as it is impossible to create an entirely objective and quantitative system to measure technicality – the HT framework we are using was created to fit the data it houses and not the other way around. However, we believe that CLs offer the best available method of comparing technicality across different domains, irrespective of their HL.

1.2. Delimiting French-origin lexis in the dataset

8In order to focus on the proportions of French loanwords in the dataset, it was first necessary to categorise all words in the corpus according to their language(s) of origin. Our hierarchy contains 231 different language tags based on the etymologies given in the Middle English Dictionary (MED). These tags have been retained in our project database, but each word has now also been allocated to one of twenty language origin sub-groups to enable analysis (see Table 1, below). For example, 537 words have been included in the ‘Latin and / or French’ sub-group (‘L + OF / AF’) and they include words with any of one of nineteen language tags, containing various combinations of Latin, Old French and Anglo-French, e.g. ‘Latin;Anglo-French’, ‘Latin;Old French;Anglo-French’, Latin;Old French-Old French’, etc.

  • 5 Notes on the language labels used are as follows: OE = Old English; OF = Old French, including ‘Con (...)

Table 1. Number and percentage of words per language origin sub-group across the corpus5

Language origin

No.

%

OE

2187

41.5%

OF / AF

1245

23.6%

L + OF / AF

537

10.2%

L

340

6.4%

OE + OS

217

4.1%

OE + OF / AF

168

3.2%

OS

133

2.5%

OE + L + OF / AF

117

2.2%

Germanic

107

2.0%

OE + Germanic

42

0.8%

OE + L

25

0.5%

OF / AF + Germanic

11

0.2%

Other Romance

5

0.1%

Celtic

4

0.1%

OF / AF + OS

3

0.1%

OS + Germanic

3

0.1%

OS + L

2

0.04%

Arabic

1

0.02%

Uncertain

69

1.3%

Other

60

1.1%

Total

5276

100%

9Once all words had been divided amongst the language origin sub-groups, results were analysed to see if there were French-specific trends at the general, hypernymic levels of the hierarchy (defined for the purposes of this study as CL0 and CL1 combined) and the technical or hyponymic levels (CL2-CL4 combined), as well as at individual CLs. Comparisons were made across the corpus as a whole and between the nine semantic domains. In each case, calculations are based on the percentage of words belonging to a particular language origin group at a particular CL in one domain or the corpus, or across all CLs in one domain or the corpus. This allows us to compare the relative proportions of French borrowings at the various levels of the hierarchy.

2. Results

2.1. Overview of Category Level distribution

10Within the nine semantic domains, words are distributed proportionately across the five CLs in the hierarchy as follows in Table 2:

Table 2. Number and percentage of words at each CL per domain

Domain

CL0

CL1

CL2

CL3

CL4

Total

Building

141

291

193

41

7

673

21.0%

43.2%

28.7%

6.1%

1.0%

100%

Domestic Activities

132

158

80

25

0

395

33.4%

40.0%

20.3%

6.3%

0.0%

100%

Farming

222

384

267

53

3

929

23.9%

41.7%

28.6%

5.5%

0.3%

100%

Food Preparation

56

218

62

6

0

342

16.4%

63.7%

18.1%

1.8%

0.0%

100%

Hunting

98

287

89

7

0

481

20.4%

59.7%

18.5%

1.5%

0.0%

100%

Manufacture

109

323

162

38

2

634

17.4%

50.8%

25.6%

6.0%

0.3%

100%

Medicine

222

487

94

20

3

826

26.9%

59.0%

11.4%

2.4%

0.4%

100%

Trade

146

254

103

31

1

535

27.1%

47.9%

19.1%

5.8%

0.2%

100%

Travel by Water

134

231

78

18

0

461

29.1%

50.1%

16.9%

3.9%

0.0%

100%

Corpus

1260

2633

1128

239

16

5276

23.9%

50.0%

21.4%

4.4%

0.3%

100%

11In all domains, the pattern is the same. CL1 (the second most general) is always the highest populated category, and half of all words in the corpus overall are found at this level. Words at CL0 and CL2 account for 23.9% and 21.4% of the corpus, respectively, whereas the percentage of words at CL3 and CL4, the most specific, is much lower and accounts for just under 5% of the corpus. CL4 lexis is especially rare with only sixteen words, or 0.3% of the corpus, found at this, the most technical level of the hierarchy. Building (35.8%), Farming (34.4%) and Manufacture (31.9%) clearly stand out as the domains with the highest relative proportions of vocabulary at the technical levels (CL2-4).

12It is also very useful to analyse the distribution of senses up and down the hierarchy to see how fine-grained the distinctions are across the corpus and within each domain (see Table 3, below). Once again, CL1 is the most popular category overall and half of all senses are categorised at this level, but we have more senses at CL2 than we do at CL0. This means that whilst only a quarter (26.1%) of words overall are technical in our dataset (i.e. found at CL2-4), technical senses make up over a third (36.2%) of the hierarchy.

Table 3. Number and percentage of senses at each CL per domain

Domain

CL0

CL1

CL2

CL3

CL4

Total

Building

26

108

139

31

5

309

8.4%

35.0%

45.0%

10.0%

1.6%

100%

Domestic Activities

40

69

46

17

0

172

23.3%

40.1%

26.7%

9.9%

0.0%

100%

Farming

69

166

122

36

2

395

17.5%

42.0%

30.9%

9.1%

0.5%

100%

Food Preparation

12

88

38

6

0

144

8.3%

61.1%

26.4%

4.2%

0.0%

100%

Hunting

22

151

63

7

0

243

9.1%

62.1%

25.9%

2.9%

0.0%

100%

Manufacture

34

173

101

21

2

331

10.3%

52.3%

30.5%

6.3%

0.6%

100%

Medicine

47

184

56

11

2

300

15.7%

61.3%

18.7%

3.7%

0.7%

100%

Trade

33

128

55

17

1

234

14.1%

54.7%

23.5%

7.3%

0.4%

100%

Travel by Water

35

87

47

10

0

179

19.6%

48.6%

26.3%

5.6%

0.0%

100%

Corpus

318

1154

667

156

12

2307

13.8%

50.0%

28.9%

6.8%

0.5%

100%

  • 6 Medicine has both the lowest proportion of hyponymic words (14.2%) and hyponymic senses (23.1%) in (...)

13However, the same three domains with the highest proportions of hyponymic words also have the most hyponymic senses: Building (56.6%), Farming (40.5%) and Manufacture (37.4%). Graph 1 (below) shows the split between CL0-1 and CL2-4 senses across all nine domains and clearly highlights the fact that Building is the only domain where technical senses outnumber general ones. Conversely, there are fewer technical senses compared to the corpus average (36.2%) in the following domains: Medicine (23.1%),6 Hunting (28.8%), Food Preparation (30.6%), Trade (31.2%) and Travel by Water (31.9%).

Graph 1. Percentages of senses at general (C0-1) and technical (C2-4) levels per domain

Graph 1. Percentages of senses at general (C0-1) and technical (C2-4) levels per domain

14The hierarchy extract below shows numerous examples of technical vocabulary and senses from Building (seemingly our most fine-grained domain), listed under the superordinate sense, Stone/rock. For example, we find ornel (‘inferior building stone’) at CL2, robel (‘small stones used in making mortar’) at CL3 and sconchoun creste (‘splayed building stone for the top of a gate’) at CL4.

Hierarchy example 2. Extract from Building with examples of technical lexis at CL2, CL3 and CL4

Stone/rock

ston a1150-1450+ Old English

.Stone as material for paving

..A paving stone

paving(e)-ston 1400-1450+ Old English-Old English

...Cobble

ston c1300-1450+ Old English

cobel-ston a1425-1450+ uncertain-Old English

cogel c1450-1450+ uncertain

cogel-ston 1450+ uncertain-Old English

.Building stone

ston a1150-1450+ Old English

stene kin c1275 Old English Old English

werk ston 1364 Old English Old English

fir-ston 1399-1415 Old English Old English

docelette 1428 ?Old French

min(e) ?1440 Latin;Old French

crop ston 1441 Old English Old English

moder ston 1442 Old English Old English

ende ston 1444-1446 Old English Old English

heth-ston 1445-1448 Old English;Old Scandinavian-Old English

pier c1450 Old French

..Stone for constructing a wall

wal ston 1424-1450+ Old English Old English

..Stone for building a gutter

goter ston 1427-1450+ Latin;Old French Old English

..Small building stone

...Small stones used in making mortar

robel ?1440 ?Anglo-French

..Inferior building stone

ornel 1338-1450+ Old French

ornel ston 1349-1432 Old French Old English

..Foundation stone

ground table ston 1434 Old English Old English;Latin;Old French Old English

..Support stone

somer 1288-1428 Old French;Anglo-French

..Stone used to lay a horizontal course along the base of a building (nouns)

bench-tableston 1434 Old English-Old English;Latin;Old French

..Flat building stone

platener ston 1399 Old French Old English

...Flat stone for the top or ridge of a wall

cres-table 1427 Latin;Old French-Old English;Latin;Old French

..Hewn or shaped stone

ston c1300-1450+ Old English

assheler 1339-1450+ Old French

assheler ston 1423-1450+ Old French Old English

...Hewn stone for corners or buttresses

assheler coin 1369 Old French Old French

...Wedge-shaped stone for building a vault or arch

vousour 1288-1450+ Old French;Anglo-French

...Splayed building stone

sconchoun 1293-1450+ Old French

sconchoun assheler 1430 Old French Old French

....Splayed stone forming a quoin

sconchoun angler 1422-1443 Old French Old French

....Splayed building stone for the top of a gate

sconchoun creste 1372-1429 Old French Latin;Old French

..Hewn stone or rough stone to be cut or hewn

heu(e)-ston 1432-1448 Old English Old English

..Stone of the nature of slate

...For roofing

sclat 1307-c1425 Old French

sclat ston 1361-1450+ Old French Old English

....Individual piece of slate for roofing

rof til 1361-1450+ Old English Old English

sclat ston 1361-1450+ Old French Old English

sclat a1382-1450+ Old French

15This extract from Building also illustrates the way in which the vocabulary was classified. Componential analysis, considered a cognitively salient tool in the lexicographer’s toolkit (Kay [2000: 56]), is used so that each time a new component is added to a definition, we move down a level in the hierarchy. The length of some of the defining headings in the extract shows how precise and technical the senses of the terms collected below them are. We move, for example, from .Building stone to ..Small building stone down to the very particular ...Flat stone for the top or ridge of a wall and ...Wedge-shaped stone for building a vault or arch, in which both the shape of the stone and its purpose are specified.

2.2. Focus on French loanwords at general vs technical levels

  • 7 For example, a previous study on the penetration of French-origin lexis in six Middle English occup (...)
  • 8 The occupational domain of domestic activities was added later than the others to the BTh project c (...)
  • 9 Fifty-one contiguous senses (with 212 words) were added to the occupational domains in the dataset, (...)

16When we analyse the etymology of all words in the corpus (see Table 4, below), the ‘French’ category is the second largest after ‘Old English’ and accounts for 23.6%. A further 10.2% of lexis is classed as ‘Latin and/or French’, giving us a total of 33.8% for ‘French +’ (this equates to ‘French’ and ‘Latin and/or French’ combined). Our French loanword results are noticeably higher than those found in previous analyses of BTh data7 and there are two main reasons for the increase. The first is the inclusion of three new semantic domains into this project dataset: Domestic Activities,8 Hunting and Medicine. Whilst the first of these domains exhibits near average levels of borrowing from French (compared to our project corpus), Hunting and Medicine have significantly higher than average percentages of either ‘French only’ or ‘French+’ lexis, as we shall see in more detail below. Secondly, in order to examine semantic shift, this project has added the contiguous higher levels in the hierarchy to the data from the BTh for which only strictly domain-specific lexis was collected (cf. Sylvester, Marcus & Ingham [2017]; Sylvester & Marcus [2017]). As we discuss in our analysis below, this increased number of general senses may also have contributed to an increase in the percentage of French borrowings in the dataset.9

  • 10 Eleven sub-groups (including Arabic, Celtic languages, Romance languages other than French, and wor (...)

Table 4. Number and percentage of language origin sub-group per domain (condensed version)10

Domain:

B

DA

F

FP

H

Man.

Med.

T

TbW

Corpus

OE

289

211

536

117

161

268

210

193

202

2187

42.9%

53.4%

57.7%

34.2%

33.5%

42.3%

25.4%

36.1%

43.8%

41.5%

OF / AF

118

83

155

129

218

129

154

165

94

1245

17.5%

21.0%

16.7%

37.7%

45.3%

20.3%

18.6%

30.8%

20.4%

23.6%

L + OF / AF

72

32

52

28

34

42

204

42

31

537

10.7%

8.1%

5.6%

8.2%

7.1%

6.6%

24.7%

7.9%

6.7%

10.2%

L

15

12

23

18

2

27

203

26

14

340

2.2%

3.0%

2.5%

5.3%

0.4%

4.3%

24.6%

4.9%

3.0%

6.4%

OE + OS

27

3

67

7

22

45

11

18

17

217

4.0%

0.8%

7.2%

2.0%

4.6%

7.1%

1.3%

3.4%

3.7%

4.1%

OE + OF / AF

24

14

22

14

11

31

10

28

14

168

3.6%

3.5%

2.4%

4.1%

2.3%

4.9%

1.2%

5.2%

3.0%

3.2%

OS

27

6

34

7

12

19

9

2

17

133

4.0%

1.5%

3.7%

2.0%

2.5%

3.0%

1.1%

0.4%

3.7%

2.5%

OE + L + OF / AF

33

5

13

6

5

24

10

10

11

117

4.9%

1.3%

1.4%

1.8%

1.0%

3.8%

1.2%

1.9%

2.4%

2.2%

Germanic

13

13

6

1

2

19

4

16

33

107

1.9%

3.3%

0.6%

0.3%

0.4%

3.0%

0.5%

3.0%

7.2%

2.0%

All other

55

16

21

15

14

30

11

35

28

225

8.2%

4.1%

2.3%

4.4%

2.9%

4.7%

1.3%

6.5%

6.1%

4.3%

Total

673

395

929

342

481

634

826

535

461

5276

100%

100%

100%

100%

100%

100%

100%

100%

100%

100%

17The main stage of our investigation focused on the distribution of French loanwords across the hypernymic and hyponymic levels of the hierarchy to see if any specific trends emerged. A wide variety of examples of French-origin borrowings from all domains, and from the most general level of technicality (CL0) to the most specific (CL4), are set out in Table 5, below. Each box of the table shows the contiguous levels of meaning that make up a loanword’s position in the hierarchy, with the sense highlighted in bold corresponding to its CL. As we can see from the final column where ‘n/a’ is indicated, there are no French-origin words found at CL4 in five of our domains: Domestic Activities, Farming, Hunting, Food Preparation and Travel by Water.

Table 5. Examples of French-origin loanwords at each CL from each domain

CL0

CL1

CL2

CL3

CL4

Building

A building

edifice

c1390-1450+

Builder

.Plasterer

plasterer

1286-1450+

Wood in specific form

.Wooden beam or timber

..Horizontal beam

traversein

1277-1348

Stone/rock

.Building stone

..Stone of the nature of slate

...For roofing

sclat

1307-c1425

Stone/rock

.Building stone

..Hewn or shaped stone

...Splayed building stone

....Splayed stone forming a quoin

sonchoun angler

1422-1433

Domestic Activities

Clothing

coverture

a1325-1450+

Garden

.Kitchen/herb garden

herber

c1400-c1450

Brewing

.Vat/vessel used in brewing/fermenting

..Device for boiling wort

furnais(e)

1369-1450+

Manufacture of thread

.Winding

..Winding on spool/bobbin

...Spool / bobbin

roket

1440

n/a

Farming

Farmer

gaineier

c1450

General Equipment

.Harness of draught animal

harneis

1333-1450+

Herding/pasturing/ confining

.Herding

..Herdsman

wardein

c1300-1450+

Ploughing equipment

.Plough

..Plough beam

...Fastening device on front of plough beam

clivie

a1325-1450+

n/a

Food Preparation

Providing/receiving food

sustenaunce

c1300-1450+

General preparation processes

.To sift

bulten

?c1175-1450+

Equipment for food preparation

.Baker’s equipment

..Rolling pin

roller(e)

a1399-1450+

n/a

n/a

Hunting

Hunting

chace

?c1250-1450+

Hunter

.Hunter using ferret/weasel

fereter

c1275-1450+

Trap/snare

.Net for hunting

..Net for hunting rabbits/hares

hai(e)

1389-1450+

Hunting with hounds

.Pack of hounds

..First group of hounds

...Member of first group of hounds

vaun-chasour

c1410

n/a

Manufacture

Metal

metal

c1230-1450+

Metalworker

.Worker in ironware

lorimer

c1230-1450+

Alloy

.Brass

..Types of brass

latoun

c1325-1450+

Glass and glass-like material

.Glass

..Pane

...Diamond shaped pane

quarrel

1450+

Containers for drink

.Large container for drink

..Barrel/cask for liquor

...Barrel/cask for wine

....Barrel/cask for rumney wine

romenei(e)

1450+

Medicine

To heal/cure

amenden

c1300-1450+

Process of healing of an injury etc.

.To heal a wound

souden

a1400-c1425

Medicines

.A medicine/

medicament

..Prescribed

medicine

receit(e)

a1398-1450+

Surgeon

.Surgeon performing specific operations

..Blood-letter

...Blood-letter who uses a cupping glass

ventoser

c1350

Treatments uniting/replacing parts

.Uniting fractures/wounds etc.

..Uniting a wound

...Uniting a wound by stitching

....A stitch

pointe

a1400-1450+

Trade

Merchant

marchaunt

c1225-1450+

Seller

.Itinerant/pedlar

pe-poudre

1376-1450+

Coins collectively

.A coin

..A debased coin

bas(e)lard

a1253-c1400

Sellers of specific things

.Seller of provisions

..Seller of fruit ...Female seller of fruit

fruitestere

c1390

n/a

Travel by Water

Sailor

mariner

c1300-1450+

Vessel/ship/boat

.Small vessel

spinace

1342-1450+

Transportation by water

.Loading /unloading cargo

..To load cargo

chargen

c1300-1450+

Part of vessel above water

.Deck superstructure

..Deck-house

...Types of deck-house

somer-huche

1346-1434

n/a

  • 11 The following are the seven ‘French’ or ‘Latin and/or French’ words found at CL4 in our hierarchy: (...)

18When average results for the corpus as a whole are analysed (see Table 6, below), French vocabulary is found clearly in higher proportions at the higher, more general levels: 25% (‘French’) / 36% (‘French+’) at CL0-1 compared to 20% (‘French’) / 28% (‘French+’) at CL2-3. This pattern mirrors that found in Sylvester [2018], which analysed a small subset of tool names in Building and found a lower concentration of French loanwords at the hyponymic levels than at the hypernymic levels. However, when results are examined individually per domain, we find that there are considerable differences in the distribution of borrowings. Note that CL4 words were excluded from this section of the analysis as they count for only sixteen out of 5276 words (or 0.3% of the corpus) overall and they only feature in any language in five out of the nine domains.11

Table 6. number and percentage of ‘French’ and ‘French+’ loanwords at hypernymic (CL0-1) and hyponymic (CL2-3) levels per domain

 

 

All words

French

French+

French

French+

Building

 

CL0-1

432

76

125

18%

29%

CL2-3

234

40

62

17%

26%

Domestic Activities

 

CL0-1

291

72

98

25%

34%

CL2-3

104

11

17

11%

16%

Farming

 

CL0-1

609

106

142

17%

23%

CL2-3

317

49

65

15%

21%

Food Preparation

 

CL0-1

274

112

134

41%

49%

CL2-3

68

17

23

25%

34%

Hunting

 

CL0-1

385

167

196

43%

51%

CL2-3

96

51

56

53%

58%

Manufacture

 

CL0-1

432

89

119

21%

28%

CL2-3

200

39

50

20%

25%

Medicine

 

CL0-1

709

135

307

19%

43%

CL2-3

114

18

49

16%

43%

Trade

 

CL0-1

401

129

162

32%

40%

CL2-3

133

36

45

27%

34%

Travel by water

 

CL0-1

365

81

107

22%

29%

CL2-3

96

13

18

14%

19%

Corpus

 

CL0-1

3897

967

1390

25%

36%

CL2-3

1363

274

385

20%

28%

19Firstly, Hunting (our sole ‘elite’ domain) stands out. Not only does it have the highest overall saturation of French loanwords of any domain (cf. Table 4, above), but it also exhibits the reverse pattern from the corpus average, with proportions of ‘French’ being 10% higher at the CL2-3 (43%) than at CL0-1 (53%). The same is true of ‘French+’ which is 7% higher at the technical level. In contrast, the following domains clearly have more French hypernyms than hyponyms: Food Preparation (16% more), Domestic Activities (14% more), Travel by Water (8% more) and Trade (5% more). The same trend is observed when we focus on ‘French+’ results for all these domains.

  • 12 Some examples of the 204 words with a ‘Latin and/or French’ language tag in Medicine include: cirur (...)

20Results for Medicine are inconclusive. There is a small preference for ‘French’ borrowing at the general (19%) over the technical level (16%) but the proportions of ‘French+’ at CL0-1 and CL2-3 are identical at 43%. The marked differences (the greatest in any domain) in the amount of lexis classed as ‘French’ as opposed to ‘French+’ is due to the particularly high level of ‘Latin and/or French’ vocabulary in Medicine. 12

21In Building, Farming and Manufacture, the differences between proportions of general and technical loanwords (in both the ‘French’ / ‘French+’ categories) range from only 1-3%. The figure for CL0-1 is always slightly higher in each case but it is also true to say that French lexis is evenly spread out in these domains with almost the same proportions of hypernyms as hyponyms. Note that these three domains all have lower than average levels of French overall (cf. Table 4, above), as well as a higher than average proportions of technical senses overall (cf. Table 3, above).

2.3. Focus on French loanwords at individual category levels

22Next, we broke down the French loanword data further into individual category levels of the semantic hierarchy. Giving the overall finding that borrowings seem to dominate at the general levels of meaning, we were particularly interested to see if there was a higher concentration at CL0 or CL1. Similarly, would borrowing decrease as we moved down the technical levels, from CL2 to CL3? For reasons discussed above, CL4 results have been excluded from the analysis.

Table 7. Number and percentage of ‘French’ and ‘French+’ loanwords at individual CLs per domain

 

 

All words

French

French+

French

French+

Building

 

CL0

141

28

46

20%

33%

CL1

291

48

79

16%

27%

CL2

193

33

53

17%

28%

CL3

41

7

9

17%

22%

CL0-3

666

116

187

17%

28%

Domestic Activities

 

CL0

132

42

52

32%

39%

CL1

158

30

46

19%

29%

CL2

80

7

10

9%

13%

CL3

25

4

7

16%

28%

CL0-3

395

83

115

21%

29%

Farming

 

CL0

222

45

58

20%

26%

CL1

387

61

84

16%

22%

CL2

266

45

59

17%

22%

CL3

51

4

6

8%

12%

CL0-3

926

155

207

17%

22%

Food Preparation

 

CL0

56

30

34

54%

61%

CL1

218

82

100

38%

46%

CL2

62

17

23

27%

37%

CL3

6

0

0

0%

0%

CL0-3

342

129

157

38%

46%

Hunting

 

CL0

98

34

38

35%

39%

CL1

287

133

158

46%

55%

CL2

89

47

52

53%

58%

CL3

7

4

4

57%

57%

CL0-3

481

218

252

45%

52%

Manufacture

 

CL0

110

25

32

23%

29%

CL1

322

64

87

20%

27%

CL2

162

35

44

22%

27%

CL3

38

4

6

11%

16%

CL0-3

632

128

169

20%

27%

Medicine

 

CL0

222

43

95

19%

43%

CL1

487

92

212

19%

44%

CL2

94

12

38

13%

40%

CL3

20

6

11

30%

55%

CL0-3

823

153

356

19%

43%

Trade

 

CL0

145

54

62

37%

43%

CL1

256

75

100

29%

39%

CL2

102

30

37

29%

36%

CL3

31

6

8

19%

26%

CL0-3

534

165

207

31%

39%

Travel by water

 

CL0

134

26

33

19%

25%

CL1

231

55

74

24%

32%

CL2

78

11

15

14%

19%

CL3

18

2

3

11%

17%

CL0-3

461

94

125

20%

27%

Corpus

 

CL0

1260

327

450

26%

36%

CL1

2637

640

940

24%

36%

CL2

1126

237

331

21%

29%

CL3

237

37

54

16%

23%

CL0-3

5260

1241

1775

24%

34%

23As we can see in Table 7 (above), the overall pattern for the corpus as a whole is clear and we do indeed see a consistent decrease in ‘French’ borrowing as we move down the semantic hierarchy – 26% at CL0, 24% at CL1, 21% at CL2 and 16% at CL3. Once again, Hunting bucks this trend and we see the opposite – a consistent increase in the levels of ‘French’ loans as we move down through the category levels.

24When we look at the individual eight remaining domains, however, matters are less clear cut. Only Food Preparation conforms to a clear pattern with a consistent decrease in ‘French’ at each level from CL0-CL3. However, most domains do have a higher proportion of ‘French’ lexis at CL0 than at CL1. The difference is most striking in Food Preparation (16% more), Domestic Activities (13% more) and Trade (8% more). Building (4% more), Farming (4% more) and Manufacture (3% more) all have smaller but still noticeably higher proportions at CL0 than CL1. The difference in French-origin lexis at CL0 and CL1 in Medicine is negligible at under 1%. Travel by Water is the only domain, apart from Hunting, which has relatively more ‘French’ borrowing at CL1 (24%) than at CL0 (19%) for reasons which are not entirely clear.

  • 13 Medicine has only twenty words at CL3, six of which are classed as ‘OF / AF’: overture (‘opening ma (...)

25It is also important to note that while decreases may not be consistent down through the category levels, nearly all the domains have more ‘French’ loanwords at CL0 than at CL3. The only exceptions are Hunting (as expected) and Medicine which has an especially high percentage of 30% for French-origin lexis at CL3.13 In addition, the same three domains which have higher than average levels of ‘French’ overall compared to the corpus average (cf. Table 4, above) also maintain higher than average levels of ‘French’ across all four levels from CL0-CL3: Hunting, Food Preparation and Trade.

26When analysis was carried on a CL by CL basis for the ‘French+’ language category, results were virtually identical (see final column in Table 7, above). Proportions of ‘French+’ lexis were higher at CL0 than CL1 in all domains except Hunting and Travel by Water. Overall, the corpus showed a decrease in ‘French+’ borrowing as we move down through the levels, but Hunting results move in the opposite direction. Patterns in the remaining eight individual domains vary but all have a higher level of ‘French+’ at CL0 than at CL3 except Medicine (which, again, has very high proportion of ‘Latin and / or French’ words at CL3).

2.4. Proportion of French loanwords and number of words per sense

27The final stage of our analysis focuses on hyponymic density (i.e. the number of co-hyponyms listed under any one sense in the hierarchy) and whether this seems to be linked to borrowing from French.

28As Table 8 (below) shows, the project dataset comprises 2307 senses of which 52.8% are single-item (i.e. have one word only). A further 36.8% have 2 to 4 co-hyponyms, 10.2% of senses have 5 or more co-hyponyms and only 1.9% have ten or more.

Table 8. Number and percentage of lexical items per sense in corpus

No. of items

per sense

CL0

CL1

CL2

CL3

CL4

Total

1

93

581

425

110

9

1218

4.1%

25.2%

18.4%

4.8%

0.4%

52.8%

2

64

247

143

32

2

488

2.8%

10.7%

6.2%

1.4%

0.1%

21.0%

3

45

135

47

6

1

234

2.0%

5.9%

2.0%

0.3%

0.0%

10.1%

4

25

79

24

4

0

132

1.1%

3.4%

1.0%

0.2%

0.0%

5.7%

5

24

41

14

0

0

79

1.0%

1.8%

0.6%

0.0%

0.0%

3.4%

6

13

25

7

1

0

46

0.6%

1.1%

0.3%

0.0%

0.0%

2.0%

7

14

9

2

2

0

27

0.6%

0.4%

0.1%

0.1%

0.0%

1.2%

8

10

14

2

0

0

26

0.4%

0.6%

0.1%

0.0%

0.0%

1.1%

9

9

4

0

1

0

14

0.4%

0.2%

0.0%

0.0%

0.0%

0.6%

10

3

5

1

0

0

9

0.1%

0.2%

0.0%

0.0%

0.0%

0.4%

11 to 20

15

12

2

0

0

29

0.7%

0.5%

0.1%

0.0%

0.0%

1.3%

20+

3

2

0

0

0

5

0.1%

0.1%

0.0%

0.0%

0.0%

0.2%

Total

318

1154

667

156

12

2307

13.8%

50.0%

28.9%

6.8%

0.5%

100%

29Another way of looking at these data is to examine the number of lexical items when senses are grouped under CL across the corpus. Results are shown in Graph 2 (below) which demonstrates that 1-item / 2-item / 3-item senses are found throughout the hierarchy, in all five CLs from CL0-CL4. We can see that the first three columns of the graph are made up of five different colours, representing all the levels of technicality. As the level of co-hyponymic density increases however, we find that senses becoming increasingly restricted to the hypernymic, general levels at CL0 and CL1 (shown in dark blue and orange).

Graph 2. Proportionate distribution of CL vs. no. of lexical item per sense (whole corpus)

Graph 2. Proportionate distribution of CL vs. no. of lexical item per sense (whole corpus)

30We wanted to find out if there were variations in individual domains when compared to the pattern of the corpus of a whole and if this seemed to be linked to the presence and/or position of French loanwords in the hierarchy. From our discussions above, we know that French lexis dominates at the hypernymic levels and that these levels have, on average, more words per sense. So, will Hunting, for example, show a markedly different pattern? We have seen that this domain has a higher proportion of French loanwords at technical senses – will these senses show comparatively higher levels of lexical richness (i.e. have a higher proportion of senses with multiple co- hyponyms)? Or can we find a link between the extent of penetration of French into an individual domain and its relative lexical richness – will Building and Farming, which exhibit the two lowest percentages of French borrowings overall in our corpus, exhibit greater or smaller co-hyponymic density than other domains?

31To this end, we analysed proportions of co-hyponymic density (defined as 1 / 2+ / 5+ and 10+ lexical item subsets) for each domain as a whole, and for each domain divided by category level. However, results were inconclusive.

Table 9. Proportions of single (1) and multiple (2+ /5+ /10+) item senses at each CL per domain

CL

No. of items per

sense

B

DA

F

FP

H

Man.

Med.

T

TbW

Corpus

CL0

No.

 

 

1

5

13

25

1

10

11

15

4

9

93

2+

21

27

44

11

12

23

32

29

26

225

5+

10

7

14

6

8

8

13

13

12

91

10+

3

1

2

0

3

1

7

3

1

21

Total

26

40

69

12

22

34

47

33

35

318

CL0

%

1

19%

33%

36%

8%

46%

32%

32%

12%

26%

29%

2+

81%

68%

64%

92%

55%

68%

68%

88%

74%

71%

5+

38%

18%

20%

50%

36%

24%

28%

39%

34%

29%

10+

12%

3%

3%

0%

14%

3%

15%

9%

3%

7%

CL1

 No.

 

 

1

53

28

78

38

97

98

85

66

38

581

2+

55

41

88

50

54

75

99

62

49

573

5+

9

4

20

9

10

10

26

9

15

112

10+

5

1

2

1

2

0

6

0

2

19

Total

108

69

166

88

151

173

184

128

87

1154

CL1

 %

 

 

1

49%

41%

47%

43%

64%

57%

46%

52%

44%

50%

2+

51%

59%

53%

57%

36%

43%

54%

48%

56%

50%

5+

8%

6%

12%

10%

7%

6%

14%

7%

17%

10%

10+

5%

1%

1%

1%

1%

0%

3%

0%

2%

2%

CL2

 No.

 

 

1

105

28

60

24

43

61

41

32

31

425

2+

34

18

62

14

20

40

15

23

16

242

5+

3

3

12

3

0

1

2

2

2

28

10+

0

0

1

0

0

0

2

0

0

3

Total

139

46

122

38

63

101

56

55

47

667

CL2

 %

 

 

1

76%

61%

49%

63%

68%

60%

73%

58%

66%

64%

2+

24%

39%

51%

37%

32%

40%

27%

42%

34%

36%

5+

2%

7%

10%

8%

0%

1%

4%

4%

4%

4%

10+

0%

0%

1%

0%

0%

0%

4%

0%

0%

0.5%

CL3

 No.

 

 

1

23

12

24

6

7

17

9

8

4

110

2+

8

5

12

0

0

4

2

9

6

46

5+

0

0

0

0

0

2

1

1

0

4

10+

0

0

0

0

0

0

0

0

0

0

Total

31

17

36

6

7

21

11

17

10

156

CL3

 %

 

 

1

74%

71%

67%

100%

100%

81%

82%

47%

40%

71%

2+

26%

29%

33%

0%

0%

19%

18%

53%

60%

29%

5+

0%

0%

0%

0%

0%

10%

9%

6%

0%

3%

10+

0%

0%

0%

0%

0%

0%

0%

0%

0%

0%

CL4

 No.

 

 

1

4

0

1

0

0

2

1

1

0

9

2+

1

0

1

0

0

0

1

0

0

3

5+

0

0

0

0

0

0

0

0

0

0

10+

0

0

0

0

0

0

0

0

0

0

Total

5

0

2

0

0

2

2

1

0

12

CL4

 %

 

 

1

80%

0%

50%

0%

0%

100%

50%

100%

0%

75%

2+

20%

0%

50%

0%

0%

0%

50%

0%

0%

25%

5+

0%

0%

0%

0%

0%

0%

0%

0%

0%

0%

10+

0%

0%

0%

0%

0%

0%

0%

0%

0%

0%

All CLs

 No.

 

 

1

190

81

188

69

157

189

151

111

82

1218

2+

119

91

207

75

86

142

149

123

97

1089

5+

22

15

46

18

18

21

42

25

29

236

10+

8

2

5

1

5

0

15

3

3

42

Total

309

172

395

144

243

331

300

234

179

2307

All CLs

 %

 

 

1

61%

47%

48%

48%

65%

57%

50%

47%

46%

53%

2+

39%

53%

52%

52%

35%

43%

50%

53%

54%

47%

5+

7%

9%

12%

13%

7%

6%

14%

11%

16%

10%

10+

3%

1%

1%

1%

2%

0%

5%

1%

2%

2%

32When all the CLs in a domain are grouped together, (see bottom row in Table 9, above), Hunting, a French-heavy domain, stands out as displaying a lower than average co-hyponymic density overall (i.e. it has a higher proportion of 1-item senses than the corpus average). Its proportion of 5+ item senses is also lower than the corpus average. However, these statistics are almost identical to those for Building which has low levels of French loanwords overall (cf. Table 4, above).

33Indeed, there is no obvious link between amounts of French in a domain and its rates of single or multiple item senses. Trade, Food Preparation, Domestic Activities and Farming have virtually identical proportions of single-item senses but the first two domains have higher than average rates of borrowing from French and the second two have lower than average rates (cf. Table 4, above). This is further demonstrated by Graph 3 (below), which shows that relative proportions of single (in blue) and multiple (in orange, grey and yellow) item senses per domain – we can see that overall there is little deviation from the corpus average (final column) by any of the nine domains. In addition, results for 10+ items (in yellow) at all CLs are very close with senses with 10 or more co-hyponyms being rare at between <1-5% or less in any given domain.

Graph 3. Relative proportions of single (1) and multiple item (2+ / 5+ / 10+) senses per domain

34When we separate the data even further into CLs for each domain (see again Table 9, above), results are still inconsistent. Once again, Hunting stands out as having a higher than average rate of single item senses (and hence fewer multiple item senses) at the most general levels of meaning, CL0 and CL1. In the corpus overall, 71% of CL0 senses and 50% of CL1 senses have 2+ lexical items compared to 55% of CL0 senses and 36% of CL1 senses in Hunting. However, the domain does not display a higher than average number of co-hyponyms at the more technical levels of meaning; in fact, it displays lower than average numbers (even though French loanwords are more common at this level of technicality in this domain). In the corpus overall, 36% of CL2 and 29% of CL3 senses have 2+ lexical items but only 32% of CL2 and 0% of CL3 senses in Hunting have more than one lexical item. We cannot say that the normal pattern for other domains is reversed for Hunting (as we might expect).

35When we analyse the eight remaining domains, there seems to be no discernible pattern that allows us to link co-hyponymic density at the hypernymic and hyponymic level with the rate of borrowing from French in any domain. e.g. Building (low rates of French) has a low rate of single word senses at the most general level of meaning, CL0, but so do Trade and Food Preparation (high rates of French). In addition, Farming (low rates of French) has higher than average rates of multiple item senses at the technical level, CL2, whereas Building (low rates of French) has the opposite – high levels of single-item senses and lower co-hyponymic density at CL2. Based on these results, we would suggest, therefore, that any differences in the domains in terms of lexical richness in the corpus is independent of the presence of French loanwords.

3. Discussion of main results

36When our corpus is taken as a whole, proportions of French loanwords are clearly higher at the hypernymic level (CL0-1) rather than at the hyponymic level (CL2-3), replicating the findings of the pilot study on a micro-domain in Building for the Technical Language and Semantic Shift in Middle English project (Sylvester [2018]). When corpus results are further broken down into individual CLs, we also see a consistent decrease in French borrowings as we move down the hierarchy from CL0 to CL3. The inclusion of Hunting and Medicine plus contiguous, general senses to the occupational domains appears to have increased the overall levels of French-origin lexis in the dataset compared to that of the BTh.

37However, the extensive analysis carried out so far on the thousands of words in this much larger dataset has also brought to light significant variations in the distribution of French-origin hypernyms and hyponyms in individual semantic domains. Hunting displays strikingly different results to the occupational domains. Given that it was deliberately selected to add culturally elite lexis to the project, it is unsurprising that it has by far the greatest proportion of French-origin vocabulary. The proliferation of a complex noble jargon associated with hawking and venery (especially hunting with hounds) following the Norman Conquest has been examined at length by historical linguists and literary scholars (e.g. Rooney [1993]; Marvin [2006]; Hunt [2009]). It is also unremarkable that French dominates in this domain at the technical level with proportions of French borrowing increasing consistently as we move down the hierarchy from CL0 to CL3.

38Nevertheless, in this dataset, it seems that there is no simple link between the penetration of French lexis in a domain overall and the division of that lexis between technical and general senses. All of the occupational domains, regardless of their rate of borrowing from French, exhibit the opposite pattern to Hunting with hypernyms outweighing hyponyms. Food Preparation and Trade also have significantly higher than average amounts of French loanwords, but these tend to cluster at the general level. It seems likely that there were bilingual speakers at the top social strata of these professional speech communities (in the guilds or in the manorial kitchens, for example) and a class of skilled workers beneath them, with a high level of knowledge and expertise in their field. It could be (as is postulated in Sylvester [2018: 260-261]) that there was a collective urge to resist foreign borrowings at a certain level of technicality to protect craftsmanship or ideas.

39It should be noted that the findings for Medicine (added as an emerging profession to the database) are inconclusive and they do not fit in to the patterns exhibited by either the aristocratic Hunting or the occupational domains. Results are no doubt skewed by the very high numbers of ‘Latin and/or French’ words. The domain contains lower than average levels of ‘French’ loanwords but higher than average of ‘French+’ These ‘French+’ loanwords are spread evenly across general and technical levels but ‘French’ loanwords are slightly more dominant at the general level. The domain also has, surprisingly, a lower than average number of hyponymic words and senses overall.

  • 14 It is important to re-iterate that these ‘French-light’ occupational domains still have higher than (...)

40Overall, however, there may be a correlation in our data between the margin of difference between the relative proportions of French hypernyms and hyponyms and either the levels of French overall or the number of technical senses (or perhaps both). The most French-heavy domains in the corpus seem to have a greater imbalance in the spread of loanwords across the hierarchy (as we have seen) whereas those with the lowest proportions of French (i.e. Building, Farming and Manufacture)14 have a much more even spread. These three domains also have the highest proportions of technical senses (and words) in the corpus. Our findings run counter to the orthodox view that innovation led to French-origin loans being used to fill lexical gaps for terms for new technology in a semantic field.

  • 15 Note that thirty-two out of the fifty-six French technical terms in our dataset under Hunting are f (...)

41One key variable in our data is the sources from which the vocabulary is drawn and, by implication, the audiences towards which the different texts were aimed. The difference in permeability to borrowing seems to reflect the social class difference. The hunting vocabulary, for example, comes almost exclusively from manuals written for the English aristocracy. Much of the lexis of the occupational domains collected for the BTh are early examples of code-switching found in administrative texts. We suggest that in this text-type, the code-switching is downwards, from Latin (the chief language of record) into the vernacular, recording terms in use in what had been an L variety, restricted almost exclusively to the spoken medium. The hunting manuals, by contrast, introduce French-origin terms in their translations into English from French originals as seen in the Livre de chasse, written between 1387 and 1391 by Gaston de Foix and translated (and adapted) into English by Edward of Norwich, 2nd Duke of York, between 1406 and 1413 (Baillie-Grohman & Baillie-Grohman [2005: xi-xii]), presumably for an English aristocratic audience that was by then less literate in French.15 This is code-switching (or borrowing foreign-language lexical items) upwards. The title of the English version, The Master of Game, itself contains a suggestion of aspiration. It is tempting to see a suggestion of distance between the aristocratic class and the accoutrements of its pastimes, and a sense of intimacy between the users of the tools and materials of the occupational domains such as building and manufacture. In that case, our data would support Matras’s observation about a proximity constraint on borrowing [2009: 169-172]. Matras’s suggestion that this constraint appears to override other motivations, such as prestige, would seem to apply more to the social classes that were further from the French ruling élite, and this distance is reflected in their attitudes in relation to the language expressing the tools of their trades. This interpretation would refine the early suggestions that much of the French borrowing was technical terminology, since the proportions of French borrowing at the technical levels in our data do not differ markedly from the numbers of French loanwords we find at the more general levels of the semantic hierarchy. It is noticeable, too, that Weinreich’s suggestion that the terms such as those denoting parts of tools are unstable and therefore subject to obsolescence and replacement [1953: 53] does not seem to be borne out by the lexis in the occupational domains where we find more survival of native terms than might be expected from many accounts of lexical borrowing in the later medieval period.

42Our final main finding is that there seems to be no link between levels of French borrowing in any given domain and the number of words per sense, either across that domain as a whole or at individual CLs. This again suggests that it is not the case that borrowing fills perceived gaps in the lexicon. It also appears to confirm that lexicalisation is a function of speakers’ needs: where a term is richly lexicalised it is further enriched by borrowing and where there is a small amount of vocabulary, speakers do not seek to amplify it by drawing on available foreign-language material.

4. Conclusion and future prospects

43The Technical Language and Semantic Shift in Middle English project aims to look at the contact effects on the technical terminology of ME using hypernymic frameworks – these semantic hierarchies were established by the Historical Thesaurus but have not yet been used to analyse the impact of loanwords at various levels of specificity. This has allowed us to examine the language ecology of later medieval English society from an entirely new angle.

44We have found that, overall, French-origin terms dominate at the more general levels of meaning, with the clear exception of Hunting. The cultural prestige attached to French in the late medieval period does not seem to have been a strong motivating factor in promoting its use as part of the most technical terminology in the occupational domains. This is possibly as a result of the different text types from which the vocabulary items were drawn, reflecting different target audiences for that lexis. French loanwords also tended to occupy semantic spaces where there was more, not less, lexical choice (i.e. a greater number of co-hyponyms per sense). Both these findings suggest we may need to develop a new conceptualisation of how French borrowings inhabit semantic spaces in Middle English.

  • 16 A preliminary study for this project on the domains of Farming and Trade has suggested that autohyp (...)

45Future investigations will expand on these findings by mapping the outcomes of borrowings and any corresponding native terms from the Middle English period to PDE, including narrowing and broadening,16 figurative and metonymic use, rates of polysemy and obsolescence. Consequently, we hope to be able to develop of model of lexical and semantic development in Middle English that is not based on competition as a leading idea and show that the achievement of standardisation in vocabulary depended on increased technical specification linked to borrowing history.

Top of page

Bibliography

Baillie-Grohman William & Baillie-Grohman F. N. (Eds.), 2005, The Master of Game. Philadelphia: University of Pennsylvania Press.

Campbell Lyle, 2013, Historical Linguistics, Edinburgh: Edinburgh University Press.

Durkin Philip, 2014, Borrowed Words, A History of Loanwords in English, Oxford: Oxford University Press.

Emeneau M. B., 1962, “Bilingualism and Structural Borrowing”, Proceedings of the American Philosophical Society 106.5, 430-442.

Grosjean François, 2010, Bilingual: Life and Reality, Cambridge MA: Harvard University Press.

Hunt Tony, 2009, Three Anglo-Norman Treatises on Falconry, Oxford: Medium Aevum.

Ingham Richard & Marcus Imogen, 2016, “Vernacular Bilingualism in Professional Spaces, 1200 to 1400”, in Classen Albrecht (Ed.), Multilingualism in the Middle Ages and Early Modern age: communication and miscommunication in the premodern world, Berlin: De Gruyter, 145-165.

Ingham Richard, Sylvester Louise & Marcus Imogen, 2019, “Penetration of French-origin lexis in Middle English occupational domains”, in Cennamo Michela & Barbirzio Claudia (Eds.), Historical Linguistics 2015: Selected papers from the 22nd International Conference on Historical Linguistics, Naples, July 2015, Amsterdam: John Benjamins, 461-479.

Kay Christian J., 2000, “Historical Semantics and Historical Lexicography: will the twain ever meet?”, in Coleman Julie & Kay Christian J., (Eds.), Lexicology, Semantics and Lexicography, Amsterdam: John Benjamins, 53-68.

Kay Christian J., 2004, “When Ignorance is Wisdom: Some day-to-say problems of classification”, in Kay Christian J & Smith Jeremy J. (Eds.), Categorization in the History of English, Amsterdam: John Benjamins, 59-67.

Kloss Heinz, 1967, “‘Abstand languages’ and ‘ausbau languages’”, Anthropological Linguistics 9(7), 29-41.

Marvin William Perry, 2006, Hunting Law and Ritual in Medieval English Literature, Cambridge: D.S. Brewer.

Matras Yaron, 2009, Language Contact, Cambridge: Cambridge University Press.

Prins A. A., 1941, “On the Loss and Substitution of Words in Middle English”, Neophilologus 26, 280-298; 27, 49-59.

Rooney Anne, 1993, Hunting in Middle English Literature, Cambridge: D.S. Brewer.

Schendl Herbert, 2000, “Linguistic Aspects of Code-Switching in Medieval English Texts”, in Trotter David (Ed.), Multilingualism in Later Medieval Britain, Cambridge: D S Brewer, 77-92.

Serjeantson Mary S., 1935, A History of Foreign Words in English, London: Kegan Paul.

Sylvester Louise & Marcus Imogen, 2017, “Studying French-Origin Middle English Lexis Using the Bilingual Thesaurus of Medieval England: A comparison of the vocabulary of two occupational domains”, in Louviot Elise & Delesse Catherine (Eds.), Studies in Language Variation and Change 2: Shifting, Switching and Alternating Patterns in the History of English, Newcastle upon Tyne: Cambridge Scholars, 217-228.

Sylvester Louise, Marcus Imogen & Ingham Richard, 2017, “A Bilingual Thesaurus of Everyday Life in Medieval England: Some issues at the interface of semantics and lexicography”, International Journal of Lexicography 30.3, 309-321.

Sylvester Louise, 2018, “Contact Effects in the Technical Lexis of Middle English: a semantic hierarchic approach”, English Language and Linguistics 22.2, 249-264.

Sylvester Louise, forthcoming, “The Role of Multilingualism in the Emergence of a Technical Register in the Middle English Period”, in Wright Laura (Ed.), The Multilingual Origins of Standard English, Berlin: De Gruyter Mouton, 365-379.

Trotter David, 2012, “Middle English in Contact: Middle English Creolization”, in Bergs Alexander & Brinton Laurel J. (Eds.), English Historical Linguistics Volume 2. Handbooks of Linguistics and Communication Science 34/1, Berlin: De Gruyter Mouton, 1781-1793.

Trudgill Peter, 1992, “Ausbau Sociolinguistics and the Perception of Language Status in Contemporary Europe”, International Journal of Applied Linguistics 2.2, 167-177.

Weinreich Uriel, 1953, Languages in Contact: Findings and Problems, Berlin: De Gruyter.

Weinreich Uriel, 2011, Languages in Contact: French, German and Romansch in Twentieth-century Switzerland, Amsterdam: John Benjamins.

Wright Laura, 1995, “A Hypothesis on the Structure of Macaronic Business Writing”, in Fisiak Jacek (Ed.), Medieval Dialectology, Berlin: Mouton de Gruyter, 309-321.

Corpus

BTh = Bilingual Thesaurus of Everyday Life in Medieval England: https://thesaurus.ac.uk/bth/.

HT = Historical Thesaurus of English: https://ht.ac.uk/.

MED = Middle English Dictionary: https://quod.lib.umich.edu/m/middle-english-dictionary/dictionary.

OED = Oxford English Dictionary, 2nd / 3rd ed., www.oed.com.

TOE = Thesaurus of Old English: https://oldenglishthesaurus.arts.gla.ac.uk/.

Top of page

Notes

1 We are grateful to the Leverhulme Trust for funding this project across 2017-2020.

2 I argue elsewhere that multilingualism seems to be key to standardisation of the lexicon, since it enabled the possibility of synonyms that have different sociolinguistic connotations (such as prestige) or functions (such as technicality). Old English is, rather, marked by polysemy: a glance at the Thesaurus of Old English (TOE) shows the large number of semantic domains in which individual lexical items appear, showing how hard terms in Old English have to work to express a wide range of meanings (Sylvester forthcoming).

3 We would like to express our grateful thanks to Dr Harry Parkin (Research Fellow for the first half of the project and now at the University of Chester) for all his work on the creation of the hierarchy.

4 To take just two examples from the wider hierarchy: ‘duck’ (HL4) and ‘egg’ (HL6) which we would consider to be both Basic Level Terms are found at different HLs; ‘a seller of feathers’ is found at HL4 whereas a ‘seller of bread’ is one HL lower down, at HL5.

5 Notes on the language labels used are as follows: OE = Old English; OF = Old French, including ‘Continental French’ tags in the MED; AF= Anglo-French, including ‘Anglo-Norman’ tags in the MED; L = Latin; OS = Old Scandinavian, including ‘Old Danish’, ‘Norse’, ‘Old Norse’, ‘Old Swedish’, ‘Norwegian’ and ‘Old Icelandic’ tags in the MED; Germanic = ‘Middle Dutch’, ‘Flemish’, ‘Middle Low German’, ‘Middle High German’, ‘Old High German’ and ‘Low German’ tags in the MED; Celtic = ‘Welsh’, ‘Irish’ and ‘Scottish Gaelic’ tags in the MED; Other Romance = ‘Old Italian’, ‘Italian’, ‘Old Provencal’ and ‘Spanish’ tags in the MED; Other = the small minority of words (sixty) which have MED tags which combine 3+ major language groups, e.g. ‘Middle Dutch;Middle Low German;Latin;Anglo-French’; Uncertain = Etymology given as unknown in the MED of which there are sixty-nine examples in the hierarchy. Note that a further 152 etymologies are flagged up as ‘unsure’ by the MED and the source language is preceded by a question mark. These have been included here within the relevant language group e.g. ‘? Middle Dutch’ is categorised under Germanic.

6 Medicine has both the lowest proportion of hyponymic words (14.2%) and hyponymic senses (23.1%) in our dataset. This raises interesting questions about modern views on what constitutes ‘technical language’; anachronistically, we may have assumed this to be a domain with the largest number of highly specific terms.

7 For example, a previous study on the penetration of French-origin lexis in six Middle English occupational domains from the BTh has shown that 26.1% of words can be classed as ‘French+’ loans: 17.8% ‘French’ plus 8.3% ‘Latin and/or French’ (Ingham, Sylvester & Marcus [2019: 469]). These figures mirror those found by Durkin [2014: 255] in his analysis of borrowings from French in the MED. His results were 16% (‘French’) plus 11% (‘Latin and/or French’), giving a total of 27% (‘French+’).

8 The occupational domain of domestic activities was added later than the others to the BTh project corpus and was not included in the analysis in Ingham, Sylvester & Marcus [2019].

9 Fifty-one contiguous senses (with 212 words) were added to the occupational domains in the dataset, for example: ‘Animals’ above ‘Domestic Animals’ in Farming, ‘Providing/receiving food’ above ‘Supplying food/catering’ in Food Preparation, ‘Brushing/sweeping’ above ‘One who sweeps’ in Domestic Activities and ‘Stone/rock’ above ‘Stone as material for paving’ in Building.

10 Eleven sub-groups (including Arabic, Celtic languages, Romance languages other than French, and words of uncertain origin) contain very small numbers of words and account for only 4.3% of the corpus when added together. These have been collated under ‘All other’ in the bottom row of the table to give a simpler, condensed overview. Note also the abbreviations for domain names in the table as follows: B = Building; DA = Domestic Activities; F = Farming; FP = Food Preparation; H = Hunting; Man. = Manufacture; Med. = Medicine; T = Trade; TbW = Travel by Water.

11 The following are the seven ‘French’ or ‘Latin and/or French’ words found at CL4 in our hierarchy: sconchoun angler (‘Splayed stone forming a quoin’), sclat (‘Individual piece of slate for roofing’), sconchoun creste (‘Splayed building stone for the top of a gate’) in Building; romenei(e) (‘Barrel/cask for rumney wine), cedre scobe (‘Sawdust from cedar wood’) in Manufacture; pointe (‘A stich to unite a wound’), cicatrizen (‘to unite fractures/wounds etc. by forming a cicatrice’) in Medicine. The tiny amount of CL4 lexis overall means the presence of a single loanword from the ‘French’ or ‘French+’ category can significantly skew the results; e.g. out of seven CL4s in Building, two are ‘French’, giving a relatively high proportion of 29%. Similarly, when we look at ‘French’ borrowings at CL4 across the corpus, we get a very large percentage: 44%. But this calculation is based on just seven lexemes out of sixteen. For this reason – while it is, of course, important to note the presence of French loanwords at the most technical level of the hierarchy – we have chosen to exclude CL4 from this part of the analysis.

12 Some examples of the 204 words with a ‘Latin and/or French’ language tag in Medicine include: cirurgie (‘surgery’), cure (‘a cure/remedy’), defensif (‘preventative treatment’), experiment (‘medical treatment’), farmacie (‘a medicine/medicament’), fissure (‘an incision’), phisik(e) (‘art/science of Medicine’), restoren (‘to heal/cure’), pronosticacioun (‘prognosis’), valerian(e) (‘valerian’).

13 Medicine has only twenty words at CL3, six of which are classed as ‘OF / AF’: overture (‘opening made in an abscess’), regendren (‘causing a wound to heal over’), ventoser (‘blood-letter who uses a cupping glass’), fermeresse (‘female officer in charge of a religious infirmary’) and maldrie / meselri(e) (‘specific leper-house’).

14 It is important to re-iterate that these ‘French-light’ occupational domains still have higher than average proportions of French when compared to MED corpus (cf. Durkin [2014: 255]).

15 Note that thirty-two out of the fifty-six French technical terms in our dataset under Hunting are found in first attestations in the MED from this aristocratic manual.

16 A preliminary study for this project on the domains of Farming and Trade has suggested that autohyponyms (i.e. words which narrow or broaden in sense) are more likely to be of French than Old English origin.

Top of page

List of illustrations

Title Graph 1. Percentages of senses at general (C0-1) and technical (C2-4) levels per domain
URL http://journals.openedition.org/lexis/docannexe/image/4841/img-1.png
File image/png, 24k
Title Graph 2. Proportionate distribution of CL vs. no. of lexical item per sense (whole corpus)
URL http://journals.openedition.org/lexis/docannexe/image/4841/img-2.png
File image/png, 21k
URL http://journals.openedition.org/lexis/docannexe/image/4841/img-3.png
File image/png, 19k
Top of page

References

Electronic reference

Louise Sylvester, Megan Tiddeman and Richard Ingham, An Analysis of French Borrowings at the Hypernymic and Hyponymic Levels of Middle EnglishLexis [Online], 16 | 2020, Online since 17 December 2020, connection on 30 March 2023. URL: http://journals.openedition.org/lexis/4841; DOI: https://doi.org/10.4000/lexis.4841

Top of page

About the authors

Louise Sylvester

University of Westminster
L.Sylvester1@westminster.ac.uk

Megan Tiddeman

University of Westminster
meg_smales@yahoo.com

Richard Ingham

University of Westminster
R.Ingham@westminster.ac.uk

Top of page

Copyright

CC-BY-NC-ND-4.0

Creative Commons - Attribution-NonCommercial-NoDerivatives 4.0 International - CC BY-NC-ND 4.0

https://creativecommons.org/licenses/by-nc-nd/4.0/

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search