1Had the Norman Conquest never occurred, the English language would lack the Romance element that makes its lexicon “an etymological hybrid” (Stockwell & Minkova [2001: 38]). The Middle English (ME) period, conventionally said to range from about 1100 to 1500, saw an unprecedented influx of loanwords from French and Latin, the languages of power and literacy in use at the time, as well as the gradual disappearance of a portion of Old English (OE) lexical material. Durkin’s [2014: 256-257] investigation of the Oxford English Dictionary and Middle English Dictionary shows that 44%-48% of headword entries first recorded in the Middle English period derive from French and / or Latin. Furthermore, recent investigations of the Middle English lexis relating to everyday-life activities and occupations in the later Middle Ages have provided evidence opposing the conventional view of French-origin borrowing as something limited to the interests and activities of the ruling elite (Sylvester & Marcus [2017]; Ingham, Sylvester & Marcus [2019]). Specifically, a closer look at the lexis of occupational domains shows that, “in the contexts in which medieval people lived and worked, French-origin lexis played a very substantial role which should not be accounted for as a matter of elite culture” (Ingham, Sylvester & Marcus [2019: 475]).
2This paper investigates the lexis associated with two major institutions of late medieval England, namely the church and the manor. On the one hand, the Christian church as an institution was important throughout the Middle Ages and across Europe, including England. Its rituals were integral to major life events such as birth (baptism), marriage, and death (funeral and memorial services), and it provided a moral framework for everyday life that the laity was also required to adhere to – this adherence was enforced, e.g., through confession, which was made mandatory once a year by the decrees of the third Lateran Council of 1179 (Logan [2002: 195-197]). Churches (and chapels) as physical buildings were present in every major and minor location throughout England and usually served as the sites at which these rituals were performed (cf. Orme [2021: 13-41]). The church was also heavily involved in the education of its own members as well as the laity, funding the basic instruction of boys in singing and reading in song and grammar schools, as well as the instruction of (prospective) clerics in grammar, and sending (prospective) priests to universities (Orme [1973] and [2006]; Zieman [2008]; Logan [2014]).
3On the other hand, the manor represents a political and socio-economic system that “was central to the organisation of agrarian life in the Middle Ages and beyond” (Bailey [2002: 2]), allowing lords to effectively gain income from their holdings of land. As a result of “the overwhelming importance of agriculture at this time” (Ingham & Marcus [2016: 152]), “[m]ost medieval people made their living from [it]” (Dyer [2002: 13]), relying on a system of organization which became a fully-fledged institution under the control of officials responsible for its management by the late twelfth and early thirteenth centuries (cf. Dyer [2002: 119-137]. The term manor, originally referring to the lord’s dwelling on the estate, by extension came to refer to the whole administrative unit through which “land tenures, rents and many land transactions were administered and recorded, and a wide range of agricultural resources was controlled and exploited” (Bailey [2002: 2]).
- 1 On this point, see also Schendl [2012: 37], who notes this with regard to multilingual sermons wher (...)
4Both the manor and the church were multilingual environments, albeit in different ways, which makes them interesting domains in their own right as well as in comparison with each other. In the church, the upper echelons were usually francophone, as its members were either French and / or nobles themselves, especially just after the Conquest (Loyn [1988: 225]), or were at least conversant in French, as French was “a conversational language in the universities” in England (Orme [2006: 75]). The language necessary for conducting mass, however, was Latin, and so anyone who wanted to take higher orders was expected to be literate, i.e., know Latin to a certain extent, although the precise expectations and the actual levels of proficiency seem to have been rather varied throughout the Middle Ages (Logan [2014: 12-14]). On the other hand, the congregation, as well as many parish priests, would have been mostly comprised of Middle English native speakers, who would have had varying levels of French and Latin competence, from none to highly competent or even native-like, depending on their social background and the point in time during the Middle English period (see, e.g., Orme [1973]; Ingham [2014])1.
5Similarly, manors were characterized by the presence of different, numerically disproportionate linguistic groups. Estates were owned and managed by a relatively small number of bilingual individuals, who exercised authority over a much larger population of monolingual English speakers. Although Rothwell differentiates, perhaps somewhat simplistically, between “francophone landlords” and “anglophone labourers” [2009: 384], documentary evidence suggests that, from the late twelfth century onwards, it is more appropriate to refer to bilingual and monolingual English native speakers (see Short [1980]; Ingham [2017]). Besides landowners, the bilingual group included appointed officers responsible for estate management such as stewards and bailiffs, whose profession required competence in the languages of literacy (Ingham [2009]; Rothwell [2009]; Ingham & Marcus [2016]).
6Acknowledgement of the multilingual environments to which religious and agricultural practices were connected leads us to questions concerning the impact of French and Latin on the vocabulary used in these contexts. Beyond the general observation that borrowing occurred in both, recent studies have shown that (historical) thesauri can be fruitfully employed for further investigations into lexical change. For example, when it comes to the influence of borrowing in specific contexts, recent investigations of the technical vocabulary relating to everyday-life activities have shown that French borrowings tend to occur more frequently in more general senses, while more specific senses are more often expressed by native vocabulary (Sylvester [2018]; Sylvester, Tiddeman & Ingham [2020]). Sylvester, Tiddeman & Ingham’s [2020: 28] study also raises the potential usefulness of including obsolescence (among other factors) in future research in order to broaden our understanding of the lexical changes that occurred during the Middle English period.
- 2 The term lexical field (or domain) is used throughout this paper. Considering that, in this regard, (...)
7In light of the points outlined above, this paper aims to assess to what extent the influence of Romance borrowing (i.e., from French and Latin) varied depending on the lexical field2 in question and the level of specificity of meaning. Given the findings by Sylvester, Tiddeman & Ingham, we would expect higher proportions of English (i.e., native) lexical material at more specific levels of the semantic hierarchy. Additionally, the undeniably large numbers of loanwords that entered the Middle English lexicon, especially of French origin, have been made responsible for the obsolescence of native English items by replacing them (e.g., Dekeyser & Pauwels [1990]), a view which has also been challenged by Sylvester, Tiddeman & Ingham recently. We would therefore not necessarily expect to find higher amounts of loans and high rates of obsolescence to go hand-in-hand. Nouns denoting locations and objects connected to the church and the manor were collected in order to answer the following research questions:
-
What is the distribution of source languages in the two domains and in their sub-domains containing the lexis denoting locations and objects?
-
What is the distribution of lexical items and source languages across levels of specificity and how does it differ between the two domains?
-
What are the rates of obsolescence and survival up to 1500, and how do they differ between the two domains?
8The history of English has seen several periods of intense linguistic contact, broadly defined as “the use of more than one language in the same place at the same time” (Thomason [2001: 1]). Each of these inevitably led to varying degrees of language change, considering that “any linguistic change that would have been less likely to occur outside a particular contact situation is due at least in part to language contact” (Thomason [2001: 62]). The effects of language contact are especially visible in the lexicon, which is “the language layer most responsive to socio-political and cultural changes in the history of a nation” (Stockwell & Minkova [2001: 34]). Empirical investigations of the lexicon of a language are likely to show that “the most common specific type of influence is the borrowing of words” (Thomason [2001: 10]); in particular, there is statistical evidence of “the high borrowability of nouns” (Matras [2009: 167]; see also Durkin [2020: 177-178]).
9The relationship between languages in contact is often characterized as one between a dominant / superstrate and a non-dominant / substrate, or between equal / adstrate parties. While English has generally been a superstrate in more recent times, it was more often a substrate language during the Middle Ages, before England became a global colonial power. During the medieval period, English was notably influenced by two major donor languages: Old Norse after the Viking invasions, and French—initially Norman French, later Central French—after the Norman Conquest. While French is widely regarded as a superstrate, the relationship between English and Old Norse remains a topic of ongoing debate in the literature. Lutz [2013: 566-568] characterizes Old Norse as a superstrate, whereas Pons-Sanz [Forthc.b: Chapter 3] argues for an adstratal influence. Since Old Norse is not the focus of this study, we will not take a position in this debate. All the while, Latin has been an important adstrate throughout the centuries, although it has been often regarded as a superstrate in view of its significant role in religious and educational contexts (Lutz [2013: 577-584]). Depending on the areas of influence and expertise of the speakers of the superstrate languages, different areas of the lexicon were impacted. Old Norse borrowings are found especially in the domains of seafaring, “war, law, social ranks, and coins and measures” (Durkin [2014: 180-181]), although recent studies have identified Norse-derived terms in a broader range of semantic fields, such as emotions, faith and morality, or different types of movement and action (cf. Pons-Sanz [Forthc.a and Forthc.b: Chapter 3]). On the other hand, contact with French resulted in an extensive influx of words in a wide variety of lexical fields. Although the influence of French has traditionally been seen as “reflecting the leading position of the new aristocracy” (Stockwell & Minkova [2001: 38]), the presence of French-origin lexis in the vocabulary of non-elite contexts provides “evidence that French may well have been preserved in later mediaeval England not just or even primarily by the aristocracy, but by trade, professional and civic interests” (Collette & Fenster [2017: 6]). Latin played a larger role in written language and was “the language of the church, scholarship, and partly of law” (Kastovsky [2006: 250]) in medieval England. And although we can make general observations on particular influence from certain languages on certain lexical fields, these fields still warrant closer examination; the observation that we find loanwords outside the expected (or maybe stereotypical) fields has been made for both French and Old Norse (Ingham [2018a]; Durkin [2014: 181]).
10Another aspect to be taken into account when referring to borrowed words is their distribution at different levels of specificity. For example, Sylvester [2018] and Sylvester, Tiddeman & Ingham [2020] have recently shown that French borrowings tend to be concentrated at more hypernymic, or more general, levels of the semantic hierarchy of the lexicon, whereas hyponymic levels, or more specific senses, tend to show a higher proportion of native vocabulary. These results, obtained by examining lexical domains associated with medieval professional occupations, suggest “a resistance to the imported French vocabulary [...] by the class of skilled workers” (Sylvester [2018: 262]). In addition, one aspect of change in the Middle English lexicon that is often discussed in connection with borrowing is the loss and / or replacement of older, usually native, vocabulary as a consequence thereof. Sylvester, Tiddeman & Ingham [2022: 237], for example, note that “[i]t is generally accepted that large numbers of existing OE-origin terms were replaced by these new borrowings in the late medieval period”. However, by looking at patterns of semantic shift and obsolescence within specific lexical domains, Sylvester & Tiddeman [2023] and Sylvester, Tiddeman & Ingham [2022; 2023] found that, contrary to traditional interpretations of the interaction between native words and borrowings in terms of competition, “wholesale relexification did not happen during the later medieval period” (Sylvester & Tiddeman [2023: 182]) and that “loanwords did not generally oust native terms; indeed, there is not much difference in replacement rates for non-native terms replacing native ones and vice versa” (Sylvester, Tiddeman & Ingham [2022: 260]). In other words, data from lexical domains associated with everyday-life contexts and activities show that native and borrowed words tended to co-exist throughout the Middle English period.
- 3 “Am auffälligsten ist, daß nur wenige der besprochenen Begriffe eine konstante Bezeichnung behalten (...)
11Finally, the impact of borrowing on the lexicon of both the church and the manor has been investigated, albeit to different extents. Among several studies of religious lexis that have been carried out, a couple are highlighted here: a general conclusion of Käsmann’s [1961: 31] study focusing on the religious lexis of the first half of the Middle English period is that the field changes considerably throughout the centuries, both with and without the influence of loanwords3. Additionally, Chase [1988], whose study served as the basis of the Historical Thesaurus of English Section 03.08 faith, observes that the proportions of source languages vary considerably according to sub-fields: for example, he points out that the proportion of French material is higher among more abstract concepts like faith or catholicity, as well as among the vocabulary concerning church government / administration and rites and ceremonies (Chase [1988: 493]). Higher proportions of English-origin material can be found among terms for canonical hours and consumables (Chase [1988: 493-494]).
12As far as the lexis associated with the manor is concerned, the scarcity of studies focusing on the effects of language contact in the medieval countryside is undoubtedly connected to the linguistic situation of rural areas. Two studies are worth mentioning, namely, Rothwell’s [2009] investigation of rural vocabulary highlighting the scarcity of French-origin words on the one hand; and Ingham’s [2009] investigation of code-switching in manorial accounts on the other. In particular, the latter reveals the use of bilingual discourse among estate professionals, shedding light on the multilingual situation characterizing medieval manors and calling for further investigations of this evidence.
13Based on the approaches and observations outlined above, we are focusing our analysis on nouns, their etymological origins, their semantic specificity, as well as their rates of obsolescence. The overlap of social hierarchy and language is, at least superficially, similar in both domains, making them an interesting pair for comparison. Additionally, different sub-domains within a larger domain may exhibit varying patterns, making a comparison across sub-domains a useful starting point for an initial study and a basis for future research.
14Our dataset consists of Middle English nouns belonging to two lexical domains associated with the two socio-historic contexts previously discussed, the church and the manor. Specifically, we collected terms denoting places and objects (i.e., tools and other items) pertaining to medieval churches and manors. Within each domain, referred to as church and manor respectively, we further separated the data into the sub-categories locations and objects.
15The lexical data belonging to the domain church were drawn from the Historical Thesaurus of English (HTE): specifically, all the words that were used in Middle English to refer to a place or object were extracted from the category 03.08 faith and its sub-categories. As a result, 162 words denoting places and 98 words denoting objects were collected, for a total of 260 words.
- 4 The resource is the result of Gloria Mambelli’s PhD research project (Mambelli [2024]).
16The lexical data belonging to the domain manor were extracted from two separate thesauri focusing on the vocabulary associated with non-elite strata of medieval society, both modelled on the conceptual structure provided by HTE: an ongoing project collecting manorial vocabulary in Anglo-French, Medieval Latin, and Middle English4; and the Bilingual Thesaurus of Everyday Life in Medieval England (BTh), which assembles Anglo-French and Middle English lexis relating to seven occupational domains. Only the English lexical data contained in these resources, which are derived from the Middle English Dictionary (MED), were considered for this study. Specifically, the data extracted from the former consist of terms denoting the locations that made up a manorial estate, whilst those from the latter are terms for locations connected to animal husbandry and for tools employed in farming. As a result, the domain contains a total of 348 items, consisting of 209 lexical items denoting locations and 139 denoting objects. The data collected for both domains are summarized in Table 1.
Table 1. Overview of the size of the dataset (including subdomains)
|
church
|
manor
|
|
locations
|
162
|
209
|
|
objects
|
98
|
139
|
|
Total
|
260
|
348
|
17For each lexical item in the two domains, we also collected information on the etymology from the Oxford English Dictionary (OED) and MED. Since OED entries are still being updated from the second to third edition, some of the entries’ etymology sections were either very short, or unclear, or both, in which case MED was used to supply this information. We included all languages that were given as definite or potential source languages by the dictionaries, i.e., if a word is of both Latin and English origin, it was classified as mixed (see below for more detail).
- 5 The CLs are preferable to the HLs for the purpose of comparison because a more highly lexicalized f (...)
18This study makes use of the Category Level (CL) system devised by Sylvester, Tiddeman & Ingham [2020] to measure technicality on the basis of hyponymic and meronymic relations. At the basis of this system is the idea that technicality corresponds to greater specificity of meaning, as suggested by Wright [1995]: hence, a semantic hierarchy is required in order to measure technicality. For this purpose, Sylvester, Tiddeman & Ingham [2020] made use of the hierarchical structure of HTE, in which “lexical items are categorized according to two relations: hyponymy (x is a kind of y) and meronymy (x is a part of y)” (Sylvester [2018: 255]). In such a semantic hierarchy, hyponyms are grouped below hypernyms, and can, in turn, be hypernyms to even more specific senses. Rather than simply relying on hierarchical levels (HLs), which “are not a consistent or quantitative indicator of technicality in individual cases and at any specific level” (Sylvester, Tiddeman & Ingham [2020: 4]), the CL system involves further sub-divisions of HLs into levels of specificity, which make it possible to identify increasing levels of technicality within a specific sense5. These levels of specificity are numbered in ascending order: the most general sense (i.e., the top level that appears in the semantic hierarchy of HTE) is the hypernymic level CL0; the next, slightly more specific sense is the hyponymic level CL1 (which is, in turn, hypernymic in relation to CL2), and so on, until CL4, which is the most specific level used in the HTE’s hierarchy. For example, the word poppy, in the sense ‘ornamental finial on the end of a seat in church’, is classified as follows (Table 2):
Table 2. The term poppy in the HTE classification
|
03 Society
|
|
|
03.08 Faith
|
|
|
03.08.05 Artefacts
|
|
|
03.08.05.07 Furniture
|
|
|
03.08.05.07.08 Seat
|
[CL0]
|
|
sitting c1400-
|
|
|
Bishop’s
|
[CL1]
|
|
bisceopstol OE
|
|
|
see 1297-
|
|
|
chair 1480- obs. exc. History
|
|
|
parts of
|
[CL1]
|
|
at end of
|
[CL2]
|
|
ornamental finial on
|
[CL3]
|
|
poppy 1428-
|
|
- 6 Rather than building a new semantic hierarchy, as done by Sylvester, Tiddeman & Ingham [2020], the (...)
19On the basis of the CL system, we can easily assign the term poppy to CL3 by looking at how HLs and their sub-divisions are presented in HTE. However, not all thesauri use the same classification system. For instance, although BTh is derived from HTE in terms of taxonomic categories, which are arranged according to the semantic relations of hyponymy and meronymy, its conceptual classification is based on associative rather than semantic fields. Moreover, unlike HTE, BTh categorizes lexis according to semantic roles based on the concept of ‘activity’, i.e., under category headings such as ‘Agents’, ‘Instruments’, ‘Specialized locations’, etc. (Sylvester, Marcus & Ingham [2017: 310-311]). Likewise, the macrostructure of the thesaurus assembling manorial vocabulary from which part of the data were extracted is based on the semantic role approach since its classification is modelled on that of BTh. When investigating BTh lexical material, Sylvester, Tiddeman & Ingham [2020] had to rearrange it “into an extensive semantic hierarchy, modelled on the taxonomic categories devised for the Historical Thesaurus of English” (Sylvester, Tiddeman & Ingham [2020: 2]) in order to apply the CL system. Similarly, if, on the one hand, the application of CLs to the HTE classification was straightforward in view of the HTE organizational system, further work was required in order to apply the CL system to the other two resources6. As an example, let us look at the classification of the terms for ‘stable’ in BTh, which we assigned to CL1 (Table 3):
Table 3. The category ‘stable’ in the BTh classification
- 7 Attestation dates displayed in BTh refer solely to the period 1150-1450.
|
Farming
|
|
|
Animal Husbandry
|
|
|
Specialized locations
|
|
|
Locations related to keeping / managing horses
|
[CL0]
|
|
Haras 1303-14507
|
|
|
Stable
|
[CL1]
|
|
Stable 1275-1450
|
|
|
Stottes stable 1377-1397
|
|
|
Won(e) 1225-1450
|
|
20No lexical material appears under the headings ‘Farming’, ‘Animal Husbandry’, and ‘Specialized locations’, which are used as HL markers. There are several categories below ‘Specialized locations’, one of which is ‘Locations related to keeping / managing horses’. In this case, it is relatively easy to assign this hypernymic category to CL0 and, consequently, the hyponymic category below, i.e., ‘Stable’, to CL1. However, not all the categories below ‘Specialized locations’ contain lexical material: for instance, no words are found in the category ‘Locations related to bee keeping’, since all the lexical material is found one level below, under the heading ‘Beehive’. In this case, ‘Beehive’ was assigned to CL0 even though it appears at the same level of ‘Stable’ in the BTh hierarchy, on the basis of the criterion that only levels containing lexical material are considered. In addition, only the BTh categories containing Middle English material were considered: those containing solely Anglo-French material were discarded.
21Another example is the classification of the BTh category ‘Ploughing equipment’ and its hyponymic sub-categories. As illustrated in Table 4, ‘Ploughing equipment’ is found one level below ‘Tools and implements in general’. On the basis of this, the former may be assigned to CL1 and the latter to CL0. In fact, if we search the two aforementioned categories in HTE, we find that they correspond to two different hierarchical levels in the HTE semantic hierarchy, and they would be thus both classified as CL0. Consequently, based on a comparison with HTE, both BTh categories (‘Tools and implements in general’ and ‘Ploughing equipment’) were assigned to CL0.
Table 4. The category ‘ploughing equipment’ and its sub-categories in the BTh classification
|
Tools and implements in general
|
CL0
|
|
Iren 1150-1450
|
|
|
Tol 1150-1450
|
|
|
Ploughing equipment
|
CL0
|
|
Plough lenge 1450
|
|
|
[4 more items]
|
|
|
Associated equipment
|
CL1
|
|
Aker-staf 1230-1450
|
|
|
[19 more items]
|
|
|
Plough
|
CL1
|
|
Plough 1200-1450
|
|
|
[2 more items]
|
|
|
Kinds of plough
|
CL2
|
|
Cart plough 1325-1450
|
|
|
Molle-plough 1362-1450
|
|
|
Parts of a plough
|
CL2
|
|
Aker-staf 1300-1350
|
|
|
[33 more items]
|
|
|
Plough-beam
|
CL3
|
|
Bem 1350-1450
|
|
22Application of the CL system to different onomasiological resources potentially results in some limitations, considering the different classification aspects that had to be taken into account when identifying levels of technicality in the three thesauri from which our data were extracted. Nevertheless, as argued by Sylvester, Tiddeman & Ingham [2020: 4], even though “it is impossible to create an entirely objective and quantitative system to measure technicality”, the CL system represents “the best available method of comparing technicality across different domains”.
- 8 Classical Latin and the pre- and post-Conquest insular varieties of Latin are distinguished in MED.
- 9 Cf. Dance, Durkin, Hough & Pagan [2023: 109], who note that, even though many first attestations of (...)
23In addition to the level of specificity, each lexical item extracted from the aforementioned thesauri was recorded in our dataset alongside its source language(s). In cases in which an item derives from two or more languages, its mixed origin was taken into account. However, considering that the dataset is rather small, keeping each individual language and each combination separate was not feasible. Furthermore, as argued by Durkin [2014: 236-237], it is often difficult, if not impossible, to decide whether a word was borrowed from either French or Latin, and “in most of these cases composite origin from both languages seems the likeliest scenario”. This is due to the transparently close relationship and similarity of the cognates reinforcing each other (cf. Durkin [2020: 174]), such as L templum and Fr. temple or L oratorium and Fr. oratorie. Therefore, we decided to combine the different varieties of French (Anglo-Norman, Central French, Old French) and Latin (Anglo-Latin, Latin, Medieval Latin)8 that are mentioned in the dictionaries into one single category, labelled “Fr / Lat”, collecting all the lexemes of Romance origin. The next etymological category is that of “mixed” English-Romance words, containing all those words, compounds, and phrases that combine English with French and / or Latin elements. The final, smallest category is for words of “other” origin, including words of other mixed origin, such as (Old) English and Old Norse combined, and words borrowed from languages other than French and / or Latin (i.e., Old Norse and Middle Dutch). The most common donor language after French and Latin is Old Norse. Borrowings (usually from Latin and Old Norse) that are already attested in Old English were counted as English, since they would likely have been assimilated by the beginning of the Middle English period9.
24This section outlines the results obtained by analysing each domain individually and by comparing them in order to identify differences and commonalities. Firstly, the domain church is analysed in terms of language distribution according to CLs, both overall and with a focus on the two subsets locations and objects individually. Secondly, an analysis of latest attestation dates of senses is carried out on the domain in order to investigate survival and obsolescence rates throughout the Middle English period on the basis of source languages. The same analyses are then carried out on the domain manor and, subsequently, a comparison of the results from both domains is presented.
- 10 “Die nachweisbaren Auswirkungen des lateinischen Einflusses auf das Me. bleiben bis in die zweite H (...)
25The majority of the items in the church domain (Table 5), are derived from French and / or Latin (150 items, 57.69%), such as consistory (‘bishop’s court’), sanctuary, or frontier (‘cloth at the front of the altar, over the upper part’). However, it is worth noting that within this group, the more exact distribution is as follows: 60.6% French (91 items), 20.7% both or ambiguously French / Latin (31 items), and 18.7% Latin (28 items). A relatively large number of French loans was to be expected, given the prevalent use of French by the clergy in a variety of contexts, including in informal situations and in rural environments (cf. the number of fourteenth-century rural clerics giving testimony in French reported in Short [2009]), but it is somewhat surprising that Latin borrowings are so much less frequent than French borrowings. Käsmann [1961: 16], too, has observed that the number of Latin borrowings is not proportional to the importance it had in medieval England. He speculates that this is partially due to the use and popularity of French in other areas, which facilitated adoption of French loans over Latin ones, and that French forms were often adopted as a sort of compromise, since their relation to the Latin etymon was transparent in many cases10. Additionally, as pointed out by Chase [1988] and discussed in Section 2, the proportions of source languages can vary greatly between sub-domains. Another reason for the relatively lower impact of Latin might be attributed to the different perceptions of language boundaries in later medieval England. As argued by Trotter, mixed-language use is evidence of the “lack of the tidy demarcation between languages” at the time [1996: 29]. It is possible that “the process of merger between [...] Middle English and Anglo-French” [2010: 53] was not fully reflected in the interaction between Latin and the vernaculars, which may have been perceived as more separate.
26The categories consisting of English and mixed items are of comparable size, with 57 and 48 words respectively, or 21.92% and 18.46% of the total. Among English-origin words we find words such as fat ‘holy water vessel’ and church, which have been in use since before the Conquest, but also words such as churchyard and rood-beam, which are new Middle English formations. The mixed category consists mostly of such new Middle English formations, for example anchor-house and shriving-pew, but also includes a handful of pre-Conquest Latin loans which would likely have been perceived as native words by the beginning of the Middle English period but were subsequently reinforced or influenced by their French cognates, such as L cella and Fr. cel(le) or L altare and Fr. auter, alter. There are also a few mixed items that existed already prior to the Conquest, e.g., chapter-house (< capitelhus), font-vat (< fantfæt) and munec-lif (< munuclif). There are only five words of other origin, all of them on the hypernymic level (CL1), which are described below.
Table 5. Distribution of source languages in church
|
Source language
|
Number of items
|
Percentage
|
|
En
|
57
|
21.92%
|
|
Fr / Lat
|
150
|
57.69%
|
|
Mixed
|
48
|
18.46%
|
|
Other
|
5
|
1.92%
|
|
Total
|
260
|
100%
|
Table 6. Distribution of lexical items in church, per CL and per language category. Percentages of the overall total are given
|
En
|
%
|
Fr / Lat
|
%
|
Mixed
|
%
|
Other
|
%
|
Total
|
%
|
|
CL0
|
29
|
11.15%
|
88
|
33.85%
|
31
|
11.92%
|
0
|
0.00%
|
148
|
56.92%
|
|
CL1
|
26
|
10.00%
|
57
|
21.92%
|
16
|
6.15%
|
5
|
1.92%
|
104
|
40.00%
|
|
CL2
|
2
|
0.77%
|
4
|
1.54%
|
1
|
0.38%
|
0
|
0.00%
|
7
|
2.69%
|
|
CL3
|
0
|
0.00%
|
1
|
0.38%
|
0
|
0.00%
|
0
|
0.00%
|
1
|
0.38%
|
|
Total
|
57
|
21.92%
|
150
|
57.69%
|
48
|
18.46%
|
5
|
1.92%
|
260
|
100.00%
|
- 11 Even so, the general distribution is the same in all languages, with the large majority of items at (...)
27When we consider the distribution across CLs (Table 6), there are many more items on the hypernymic levels than on the hyponymic levels: 148 at CL0, 104 at CL1, seven at CL2 and only one at CL3. There are no lexical items at CL4 in this dataset. This overall decrease in the number of items with increasing specificity is clearly mirrored by all language groups except English, where we find roughly the same number of items at CL0 and CL111. This general pattern can also be observed in the two sub-domains locations (Table 7) and objects (Table 8): CL0 and CL1 are the largest categories by far, CL0 making up between 55% and 58% of the total and CL1 around 40%, both overall and in the two sub-domains. The Fr / Lat group is the largest among the language categories (between 57-58% of the total, both overall and in the sub-domains), followed by English and mixed. The English category is slightly larger overall with 57 words, compared to 48 words in the mixed category, and in the locations sub-domain with 37 words, compared to 26 words in the mixed category. In the objects sub-domain, however, there are slightly more mixed items (22) than English items (20). Furthermore, the number of locations items decreases from CL0 to CL1 among Fr / Lat and mixed items but is roughly the same among English items, following the overall trend. In the objects subset, however, there are slightly more items at CL1 than CL0 among items of mixed etymology.
28Finally, there are two major differences between the locations and objects sub-domains. Firstly, there is one word at CL3 in the objects sub-domain (poppy, used as an example above), whereas we only find words up to CL2 in the locations sub-domain. Secondly, all items of other origin can be found in the locations sub-domain, and none among the objects: cluse (from Middle Flemish; 03.08.05.05.02.01-01 ‘parts of a monastery; cell’), kirk-garth (03.08.05.02-08, ‘land, churchyard’; garth from Old Norse, kirk either with unpalatalized initial k-, or re-borrowed from Old Norse), wand-church and wand-kirk (03.08.05.03.03-15, ‘church / place of worship; made of wattle-work’; wand from Old Norse), and sanctuary garth (03.08.05.02-08, ‘land, churchyard’; garth from Old Norse, sanctuary from French / Latin).
Table 7. Distribution of lexical items in the locations subset in church, per CL and per language category
|
EN
|
%
|
Fr / Lat
|
%
|
Mixed
|
%
|
Other
|
%
|
Total
|
%
|
|
CL0
|
18
|
11.11%
|
54
|
33.33%
|
22
|
13.58%
|
0
|
0.00%
|
94
|
58.02%
|
|
CL1
|
17
|
10.49%
|
38
|
23.46%
|
3
|
1.85%
|
5
|
3.09%
|
63
|
38.89%
|
|
CL2
|
2
|
1.23%
|
2
|
1.23%
|
1
|
0.62%
|
0
|
0.00%
|
5
|
3.09%
|
|
Total
|
37
|
22.84%
|
94
|
58.02%
|
26
|
16.05%
|
5
|
3.09%
|
162
|
100.00%
|
Table 8. Distribution of lexical items in the objects subset in church, per CL and per language category
|
EN
|
%
|
Fr / Lat
|
%
|
Mixed
|
%
|
Total
|
%
|
|
CL0
|
11
|
11.22%
|
34
|
34.69%
|
9
|
9.18%
|
54
|
55.10%
|
|
CL1
|
9
|
9.18%
|
19
|
19.39%
|
13
|
13.27%
|
41
|
41.84%
|
|
CL2
|
0
|
0.00%
|
2
|
2.04%
|
0
|
0.00%
|
2
|
2.04%
|
|
CL3
|
0
|
0.00%
|
1
|
1.02%
|
0
|
0.00%
|
1
|
1.02%
|
|
Total
|
20
|
20.41%
|
56
|
57.14%
|
22
|
22.45%
|
98
|
100.00%
|
- 12 Among the obsolete items are 16 of French, 5 of ambiguous French/Latin, and 6 of Latin origin.
29In terms of survival and obsolescence, there are 27 French / Latin items, 14 English items, 13 mixed items and three items of other origin that have become obsolete by the year 1500 (see Figure 1). For the Fr / Lat category, this amounts to a loss of 18%, for the English and mixed categories, this amounts to 24.5% and 27%, respectively. There are only five words of other origin. Of these, cluse, wand-kirk and wand-church become obsolete by the late fifteenth century, sanctuary-garth disappears around 1624, and only kirk-garth is still in use today, albeit only in the restricted context of northern English varieties (OED [s.v. kirk-garth]). Among the English items that have become obsolete by 1500, we find, e.g., wax-house (‘part of the monastery where wax candles were made’), which is last attested ca. 1473, or elvat < OE elefæt (‘ampulla, chrismatory’), which is last attested ca. 1450. Among the French / Latin12 items are for example novicery (‘parts of the monastery: noviciate’) which is last attested ca. 1496, or purpitle (‘(choir) screen’), last attested ca. 1453. Finally, in the mixed category, we find, e.g., overstory (‘clerestory’, only attested twice, around 1490) or font-vat < OE fantfæt (last attested ca. 1220). In all language groups, roughly half of the obsolete items denote locations, half denote objects. These numbers and percentages refer to words or compounds used in particular senses; some of them stay in use slightly longer in other senses. This means that if we were only to count the items that fell completely out of use by the end of the Middle English period, the numbers would be slightly lower among all language groups.
Figure 1. Survival and obsolescence in church
30Let us turn now to the domain manor. As illustrated in Table 9, native vocabulary occupies considerable space in the domain, where 220 lexical items, i.e., 63.22%, are of English origin. These include words such as erthe (‘arable land’), sadel (‘saddle’), wold (‘woodland’), and compounds such as brenning-iren (‘branding iron’), lond mede (‘meadow land’) and pigges-hous (‘pigsty’). The 72 Romance-origin items, including terms deriving solely from French (e.g., wast(e) ‘wasteland’), solely from Latin (e.g., vaccarie ‘cow pasture’), and from French and / or Latin (e.g., tenaunci ‘tenant land’, from AF tenaunci and ML tenantia), account for 20.69% of the domain. A relatively low proportion of lexical material is of mixed and other origin (7.76% and 8.33% respectively). The former category contains mixed-origin terms such as restif lond (‘fallow’, the first element from Old French and the second one from Old English), while the latter includes mostly words from Old Norse, such as eng ‘meadow’, as well as a few from Middle Dutch among other sources, namely, pike ‘fork’ (from Old French, Old English, Middle Dutch), rive ‘rake’ (from Middle Dutch) and kippe-line (denoting a part of a harness, made of an element from Middle Dutch and a native one).
Table 9. Distribution of source languages in manor
|
Source language
|
Number of items
|
Percentage
|
|
En
|
220
|
63.22%
|
|
Fr / Lat
|
72
|
20.69%
|
|
Mixed
|
27
|
7.76%
|
|
Other
|
29
|
8.33%
|
|
Total
|
348
|
100.00%
|
31We turn next to the distribution of lexical material and source languages across CLs. As in the case of church, the CLs considered for the present analysis range from CL0 to CL3 since no items are found at the most specific level, CL4. The figures are tabulated in Table 10, which, similarly to church, primarily shows a higher proportion of material clustered at the hypernymic levels: 185 items (53.16%) at CL0 and 116 (33.33%) at CL1. All levels show a preponderance of native vocabulary, even though a considerably higher number of native items (120, i.e., 34.48% of the total) is found at CL0. Similarly to what we observed for church, the number of items, regardless of source language, decreases as we move down the hierarchy, except for the mixed category, where 9 items (2.59%) are found at CL0 and 15 (4.31%) at CL1. This can be explained by the presence in this category of numerous compounds, which tend to be more specific and are thus clustered at a lower level. CL3 is the least populated level, containing only the native item bem (‘plow beam’, MED, s.v. bem, n. 3(b)), followed by CL2, where only 13.22% of the material is found.
Table 10. Distribution of lexical items in manor, per CL and per language category
|
En
|
%
|
Fr / Lat
|
%
|
Mixed
|
%
|
Other
|
%
|
Total
|
%
|
|
CL 0
|
120
|
34.48%
|
40
|
11.49%
|
9
|
2.59%
|
16
|
4.60%
|
185
|
53.16%
|
|
CL 1
|
62
|
17.82%
|
28
|
8.05%
|
15
|
4.31%
|
11
|
3.16%
|
116
|
33.33%
|
|
CL 2
|
37
|
10.63%
|
4
|
1.15%
|
3
|
0.86%
|
2
|
0.57%
|
46
|
13.22%
|
|
CL 3
|
1
|
0.29%
|
0
|
0.00%
|
0
|
0.00%
|
0
|
0.00%
|
1
|
0.29%
|
|
Total
|
220
|
63.22%
|
72
|
20.69%
|
27
|
7.76%
|
29
|
8.33%
|
348
|
100.00%
|
32Let us now focus on the vocabulary denoting locations and objects and examine these two sub-domains. Table 11 shows that the main trends identified in the domain overall are mirrored in locations: as far as all source languages are concerned, a higher proportion of material is clustered at the hypernymic levels (CL0 and CL1) and proportions decrease the further down the hierarchy we progress, including those displayed by the mixed category. However, no items are found at CL3 since the most specific term found in the domain manor is a term denoting an object, and only two native items (grove and outwode) and one Norse-derived term (hag), all denoting specific types of woodland, are found at CL2.
33Interestingly, the sub-domain objects shows different trends (Table 12). Firstly, rather than showing a regular decrease from CL0 to CL3, the majority of language categories (all except English) display a higher proportion of lexical material at CL1, where 42.45% of the material is found. Secondly, the native vocabulary shows a steady increase in density from CL0 to CL2 (27 items at CL0, 31 at CL1, 35 at CL2). Finally, whilst only three items (1.44%) are found at CL2 in locations, a relatively high proportion of material (30.94%) is found at the same level in objects.
Table 11. Distribution of lexical items in the locations subset of manor, per CL and per language category
|
En
|
%
|
Fr / Lat
|
%
|
Mixed
|
%
|
Other
|
%
|
Total
|
%
|
|
CL 0
|
93
|
44.50%
|
38
|
18.18%
|
7
|
3.35%
|
11
|
5.26%
|
149
|
71.29%
|
|
CL 1
|
31
|
14.83%
|
19
|
9.09%
|
5
|
2.39%
|
2
|
0.96%
|
57
|
27.27%
|
|
CL 2
|
2
|
0.96%
|
0
|
0.00%
|
0
|
0.00%
|
1
|
0.48%
|
3
|
1.44%
|
|
CL 3
|
0
|
0.00%
|
0
|
0.00%
|
0
|
0.00%
|
0
|
0.00%
|
0
|
0.00%
|
|
Total
|
126
|
60.29%
|
57
|
27.27%
|
12
|
5.74%
|
14
|
6.70%
|
209
|
100.00%
|
Table 12. Distribution of lexical items in the objects subset of manor, per CL and per language category
|
En
|
%
|
Fr / Lat
|
%
|
Mixed
|
%
|
Other
|
%
|
Total
|
%
|
|
CL 0
|
27
|
19.42%
|
2
|
1.44%
|
2
|
1.44%
|
5
|
3.60%
|
36
|
25.90%
|
|
CL 1
|
31
|
22.30%
|
9
|
6.47%
|
10
|
7.19%
|
9
|
6.47%
|
59
|
42.45%
|
|
CL 2
|
35
|
25.18%
|
4
|
2.88%
|
3
|
2.16%
|
1
|
0.72%
|
43
|
30.94%
|
|
CL 3
|
1
|
0.72%
|
0
|
0.00%
|
0
|
0.00%
|
0
|
0.00%
|
1
|
0.72%
|
|
Total
|
94
|
67.63%
|
15
|
10.79%
|
15
|
10.79%
|
15
|
10.79%
|
139
|
100.00%
|
34The analysis of latest attestation dates (cf. Figure 2) reveals that 81 native lexical items (36.82%), 26 Romance-origin items (36.11%), 13 mixed-origin items (48.15%), and 13 other-origin items (44.83%) became obsolete in the sense in question by the end of the Middle English period. Among these, we find native terms for locations such as bernet, in the sense of ‘a tract of wasteland cleared by burning’ (MED [s.v. bernet]), and kete, denoting ‘some kind of shed or outhouse for cattle or sheep’ (MED [s.v. kete]), as well as words for tools such as the native term bil, in the sense of ‘a digging implement; a mattock or pickaxe’ (OED [s.v. bil]) and the French-origin term croupe, in the sense of ‘crupper’ (MED [s.v. croupe]).
35As far as native and Romance-origin items are concerned, a relatively higher proportion of lexical material survived past 1500 in the relevant sense in both categories (63.18% and 63.89% respectively), including terms such as medwe ‘meadow’, of native-origin, and pasture, from both French and Latin, which are still in use at present. Conversely, the proportions of mixed- and other-origin items surviving after the Middle English period and those becoming obsolete by 1500 are very similar: 13 mixed-origin items (48.15%) fell out of use while 14 (51.85%) survived, and 13 other-origin words (44.83%) became obsolete while 16 (55.17%) survived. In sum, native and Romance-origin vocabulary, on the one hand, and mixed- and other-origin items, on the other, exhibit similar patterns in terms of survival and obsolescence, the former displaying a loss of 36% and the latter of 44-48%.
Figure 2. Survival and obsolescence in manor
36In terms of source languages, there is a clear difference between church and manor (Figure 3): unsurprisingly, church is heavily influenced by Romance borrowings, and more precisely by loanwords deriving solely from French (which account for more than half of the Fr / Lat group). Conversely, manor shows a very strong preponderance of native terms, whose number is considerably lower in church. Interestingly, the predominant source languages show similar proportions in the two domains: Romance lexical items account for 57.69% of church and English for 63.22% of manor. Moreover, the proportion of native items in church (21.29%) mirrors the proportion of Romance vocabulary in manor (20.69%).
37The other and mixed language groups are somewhat different: there are almost as many words of mixed etymology (18.46%) as of exclusively English etymology (21.29%) in church, and only a small number of words from other languages (2%). The manor domain, on the other hand, displays a similar proportion of mixed- and other-origin lexical items (7.76% and 8.33% respectively).
Figure 3. Distribution of source languages in church and manor
38In terms of distribution of lexical material across CLs according to the system outlined previously, levels of technicality range from 0 to 3 in both domains (there are no items at the most specific level, CL4). In the church domain, over 90% of all items are at CL0 and CL1, with only a single word at CL3. Similarly, the majority of lexical items in manor are clustered at CL0 and CL1, and only a single word is found at CL3. However, the two domains show different densities at CL2, where manor has 46 items (13.22%) and church only 7 (2.69%) (see Figure 4).
Figure 4. Comparison of item distribution across CLs between church and manor
39As far as the distribution of source languages per CL is concerned, Figure 5 shows that the dominant language in each domain (i.e., French / Latin in church and English in manor) generally remains predominant across CLs, including CL3, where only a single item is found in both domains. If we look at proportions, we find that they are relatively similar in church at CL0, 1 and 2, although there is a slight increase in the proportion of English material with increasing technicality. manor does not show the same consistency across CLs 0-2: although the English material accounts for the highest proportion at each CL, the percentage is just over 50% at CL1, whereas CL0 and especially CL2 display higher percentages.
Figure 5. Distribution of lexical items in church and manor, per CL and per language category
40With regard to the vocabulary for locations (Figure 6), we can firstly observe that it exhibits a similar pattern to that identified in the domains overall in terms of distribution of source languages across CLs. Lexical items are still clustered at the hypernymic levels (CL0 and CL1), with a higher proportion of Romance material in church and of native material in manor. The proportion of English items in church slightly increases with each step of increased technicality, and the comparatively low (but still over 50%) proportion of English items at CL1 in manor can be seen here as well. As previously noted, the single word found at CL3 in both domains belongs to the sub-domain objects, therefore no items are found at this level in locations. The number of words found at CL2 is much lower than at CL0 and CL1 in both domains, making any general comparisons and conclusions difficult. Based on the few terms we have, we can observe that other source languages disappear in church, and Romance-origin as well as mixed items disappear in manor at this level.
Figure 6. Distribution of lexical items in the locations subsets of church and manor, per CL and per language category
41In the objects subdomains (Figure 7), we still find the respective dominant languages representing the largest proportions at all CLs in both domains. In the church domain, the overall and proportional number of English-origin terms is lower than in the locations subdomain, staying below 25% at CL0 and CL1, and entirely absent at CL2 and CL3. In manor, on the other hand, we still observe the slightly lower proportion of English at CL1, while the proportion of English is much higher at all other levels. Interestingly, the number of items decreases along with increased specificity in church, whereas manor shows a considerably different picture: CL1 is the most populated level, followed by CL2, which reflects the distribution of material in the domain farming identified in Sylvester, Tiddeman & Ingham [2020: 7]. The most striking difference between the two domains is found at CL2, which is much more populated in manor: 43 items are found in manor, while only 2 items (both of Romance origin) are found in church at the same level.
Figure 7. Distribution of lexical items in the objects subsets of church and manor, per CL and per language category
42Finally, as illustrated in Figure 8, comparison of survival and obsolescence across source languages in the two domains reveals a general tendency to survive past 1500, considering that the proportions of surviving items exceed 50% in the majority of language categories in both domains. Only other-origin lexis in church deviates from this pattern, displaying a higher percentage of material becoming obsolete by 1500; however, the limited number of items within this language category makes it difficult to draw reliable conclusions relating to this specific category. Comparison of the two domains shows that lexical material belonging to the domain church displays higher survival rates, at least as far as native, Romance-origin, and mixed-origin words are concerned. A higher proportion of surviving native and Romance material, compared to that of obsolete items, was identified in manor as well; however, percentages are relatively lower if compared to those in church.
43We should note the presence of hapax legomena, i.e., terms that only appear once or twice in a single text in the dataset. In the case of church, these are mostly French (and / or Latin) borrowings such as architemple ‘principal place of worship’ (attested 1297) or colloque ‘part of monastery: parlour’ (attested 1482), but also a few new formations in Middle English such as water-stone ‘holy water stoup’ (attested 1379) or mind-place ‘shrine of a saint’ (attested ca. 1449). Meanwhile, most of the manor items with a single attestation are native compounds such as bou erthe ‘pasture land’, eringe land ‘arable land’, or ele bed ‘eel pond’. The importance and longevity of these hapaxes are difficult to interpret: first, we do not know if they are simply nonce-formations or if all other attestations happen to have disappeared between the Middle Ages and now. If they are nonce-formations, they can give us an idea of the kind and degree of everyday lexical innovation at work during the Middle English period. Secondly, since these words have their first and last attestations within the Middle English period, they raise both the overall number of items and the number of obsolete items. This increase might make the Fr / Lat and mixed categories, in the case of church, appear larger and more innovative compared to English-origin items, where the increase due to new ME formations is somewhat mitigated by the portion of words already in use since the OE period. Similarly, the number of native hapaxes in manor may skew the results relating to obsolescence among the native material. Nevertheless, in spite of this caveat, it is interesting to observe that our data do not show considerable variation in survival / obsolescence rates across source languages in each domain. In particular, English and Romance lexis show similar rates in each domain, suggesting that we are not simply observing Romance loans ousting native vocabulary; after all, some proportion of the vocabulary becomes obsolete in every language group, even in the Romance category, and if the influx of French and / or Latin loanwords were the main cause of obsolescence among native items, we would expect a (bigger) difference in obsolescence rates.
Figure 8. Survival and obsolescence in church and manor.
44Overall, we can observe that both domains are characterized by a dominant source language, which accounts for over half of the domain: English-origin items dominate in manor, whereas church shows a predominance of Romance-origin items, the majority of which derive from French. The proportion of Romance loanwords in manor (20.69%) is close to that identified by Ingham, Sylvester & Marcus [2019], and Sylvester, Tiddeman & Ingham [2020: 12] in the domain farming (19-25%), with which the dataset investigated in this paper shares a number of lexical items (specifically those relating to farming objects and animal husbandry locations). The much higher proportion of 57.69% Romance loans in church, on the other hand, is comparable to the proportion of 52% French and French / Latin loans in the domain hunting reported by Sylvester, Tiddeman & Ingham [2020: 18], which “was deliberately selected to add culturally elite lexis” (Sylvester, Tiddeman & Ingham [2020: 26]) and whose high proportion of French-origin items is thus unsurprising.
45While the domain church appears to be highly receptive to borrowing, the predominance of native lexis in manor seems to have its roots in the sociolinguistic context with which the domain is associated, specifically the “numerical imbalance between speakers of the two languages” (i.e., English and French respectively) in the medieval countryside, which “inevitably affected the overall linguistic situation obtaining on the estates” (Rothwell [2009: 384]). As we know, rural areas were characterized by a predominance of English monolingual speakers: although administrators such as stewards and bailiffs, whose profession required “a good working knowledge of French” (Ingham & Marcus [2016: 152]), might have been “the bilingual innovators acting as the principal agents of potential language change” (Matras [2009: 165]) in manorial environments, their bilingual competence was not required when interacting with the monolingual labourers who shared their native language. Nevertheless, our data show that a portion of Romance-origin items (20%) entered this domain during the Middle English period. Although not substantial, such a proportion shows how French and Latin contributed to shaping not only high-status domains pertaining to the life of the ruling elite, but also those associated with predominantly anglophone environments, which are more likely to show a certain degree of resistance to borrowing. In a religious context, on the other hand, the dissemination of borrowed material would have been facilitated by more direct contact between English native speakers and the bilingual clergy, and especially those belonging to the mendicant orders, as suggested by Ingham [2018a].
46Even though, as described in Section 3, the material under investigation was extracted from thesauri presenting different onomasiological classifications, the application of the CL system allowed us to investigate and compare the distribution of lexical material across the two domains, proving that CLs represent effective indicators of technicality in the comparison of different domains. Firstly, the analysis of the distribution of vocabulary across CLs brought to light the fact that only lexical items denoting objects are found at the most technical level, i.e., CL3, with the caveat that only one single item in each domain was found at this level. This possibly suggests that words for objects and tools are more likely to show greater specificity of meaning, even though a larger dataset is required in order to test this hypothesis. With respect to CL2, the analysis unveiled a considerably higher proportion of material in the objects subset of manor when compared to the corresponding subset in church: this result points at a greater technicality in the vocabulary for objects relating to manorial contexts as opposed to religious ones. However, this may be due to the different sources from which the data were extracted (i.e., HTE, BTh, and an ongoing project, all displaying slightly different classification structures) and, in particular, to the high number of co-hyponyms, e.g., those referring to the parts of a plough, contained in manor, which may contribute to skewing the overall result. This is also not meant to suggest that Christian rituals did not require very particular objects, but rather that there are other areas that required a higher degree of specialization: a cursory glance at the rest of the Middle English religious lexis (Section 03.08 faith in HTE) reveals a high number of items at CL2-4 relating to feasts and holy days, Christian symbols in general and the cross in particular, monastic life, and ecclesiastical hierarchies.
47As far as source languages are concerned, their distribution does not seem to vary considerably across CLs: Romance-origin items tend to dominate in church at all levels, while manor shows the opposite trend, i.e., a preponderance of native material at all levels. The proportion of native items increases at the hyponymic levels in manor, especially in the objects subset, hence mirroring Sylvester’s [2018: 261] findings concerning the predominance of native terms and consequent “resistance to borrowing lexis” at the hyponymic level. This is not the case in church, where the preponderance of borrowed material remains constant at all levels. In particular, the objects subset of church shows solely Romance-origin items at the hyponymic levels, where native items are entirely absent. The trends identified in this domain seem to mirror those displayed by the domain hunting, investigated alongside others by Sylvester, Tiddeman & Ingham [2020]: in addition to a relatively high proportion of Romance-origin vocabulary, the authors found it to exhibit “the reverse pattern from the corpus average” [2020: 16], i.e., the proportion of Romance borrowings increases with technicality. They suggest that the opposite trends may be the “result of the different text types from which the vocabulary items were drawn, reflecting different target audiences for that lexis” [2020: 28]. The primary aim of manorial accounts, and administrative texts in general, was efficiency rather than dissemination, and this is why code-switching often occurs in such texts (Ingham [2009], [2013] and [2018b]; Ingham & Marcus [2016]; Wright [2000] and [2013]). While those responsible for drawing up accounts were literate and bilingual, not all manorial officers were, and, in particular, manorial administration was tied to a predominantly anglophone environment, dominated by English monolingual labourers. Ingham’s investigations of the presence of English terms in Latin-matrix manorial documents have shown that the fact that discourse between bilingual officials and the English monolingual labourers was conducted in English “may well have favoured the use of English items, such as familiar places, activities, and artefacts” in French discourse as well [2009: 85]. This possibly explains the relatively low proportion of borrowed lexis in the vocabulary for manorial locations and objects.
- 13 See also various articles by Timofeeva [2018a, 2018b, 2020], who characterizes particular lexemes a (...)
48The range of what can be called “religious” texts (and thus serve as sources for religious lexis) is much bigger, as is their number, and the intended audience is also more varied than that of manorial accounts and administrative documents: there are specialized texts that address very particular religious people or groups, such as the Ancrene Wisse or various monastic rules. There are also texts for “religious instruction” of the laity, such as the Cursor Mundi, which “is very likely to have been intended as source material for one of [sic] more members of a clerical community to use in spoken form when preaching” (Ingham [2018a: 213]); or the mnemonic decalogues discussed by Timofeeva [2020]. While the lexis of the domain of hunting “comes almost exclusively from manuals written for the English aristocracy” (Sylvester, Tiddeman & Ingham [2020: 27]) and the occupational vocabulary is derived largely from administrative texts, the greater variety found among religious writings and their intended audiences makes it difficult to apply the explanation offered by Sylvester and colleagues. They also discuss the “proximity constraint” suggested by Matras [2009], which assumes “greater stability of concepts pertaining to the immediate surroundings: orientation in space, time and quantity, the private domain of mental and physical activity, and the nearest human environment (body and close kin). Concepts that involve negotiation of activity with others are, by contrast, more prone to borrowing” (Matras [2009: 169]). This helps explain why the technical vocabulary of occupational domains, which includes the professionals’ “tools of the trade”, were more resistant to borrowing (Sylvester, Tiddeman & Ingham [2020: 27]) and this is confirmed by the results of the manor domain. The fact that we find a lot more borrowed material in church can be explained in a variety of ways with respect to Matras’ proximity constraint: on the one hand, if we take receptiveness to borrowing as an indicator that locations and objects of the church are distal, this would mean that these “tools of the trade” are not as private as those belonging to professionals or farm workers. There are, for example, terms for church furniture, (re)movable parts of altars and shrines as well as different kinds of screens or dividers in the objects subset. The latter involve negotiations of spatial and visual boundaries, and objects used in common church rituals, such as baptism, also involve “negotiation of activity with others”. On the other hand, there are many borrowed terms for locations that would only have been accessible to a select few, such as the different parts of a monastery, or areas in and around a church closed to the public, such as clerical residences. If, however, we centre the clergy in this model, the predominance of Romance loans should come as no surprise, considering that these sub-domains are proximal to the clergy. Since French and Latin would have been more central to their religious education and professional practices than English, the use of Romance-based lexis would likely have been more natural than the use of English-origin terms13.
49Finally, the analysis of latest attestation dates of senses allowed us to investigate and compare obsolescence and survival rates in relation to source languages in both domains throughout the ME period. In general, the majority of items survive past 1500 in both domains, although higher proportions of surviving material were identified in church, which may be possibly linked to the greater amount of religious texts available and the fact that farming techniques can be considered likely to change faster than church customs. The native English material becoming obsolete does not exceed the surviving one in either domain. This suggests that the influx of French- and / or Latin-origin loanwords did not push native terms out, possibly confirming Sylvester & Tiddeman’s [2023], and Sylvester, Tiddeman & Ingham’s [2022; 2023] findings concerning a tendency of native terms and loanwords to co-exist within the same semantic space throughout the Middle English period rather than competing and resulting in widespread lexical replacement. However, considering that this study did not involve the comparison of synonyms specifically, further investigations aimed at comparing sets of synonyms from different source languages are required in order to shed more light on patterns of synonymy and replacement in the two domains in question. Furthermore, the observation that manor exhibits relatively lower survival rates compared to church calls for further investigations of the available textual evidence connected to the two domains, considering that a higher number of religious texts may explain the higher percentage of surviving lexis in the domain church.
50The analysis of the two lexical domains has revealed some key findings regarding the distribution of lexical material, source languages, and levels of technicality, as well as some insight in terms of survival and obsolescence.
51Firstly, the distribution of source languages in each domain seems to reflect sociolinguistic factors, as competence in French and Latin was more common among the clergy; the latter, as argued by Ingham [2018a], contributed to the diffusion of French-origin items in Middle English. On the other hand, the relatively lower spread of loanwords in the domain manor is accounted for by the larger number of English monolingual speakers in the countryside [Rothwell 2009; Short 2009] and confirms Ingham’s [2009] findings concerning a preference for English terms when referring to places and artefacts on manorial estates. However, the proportion of Romance-origin items found in this domain suggests that contact with Latin and French must have occurred in rural contexts as well, possibly due to the presence of multilingual individuals responsible for estate management.
52Secondly, as far as technicality is concerned, we found consistency across domains and levels of specificity in that the lexical material is mostly clustered at the hypernymic levels in both domains, with minimal representation at the most specific level, and the dominant source language (French / Latin in church and English in manor) remains preponderant across CLs. Although the analysis of vocabulary for locations and objects within the domains does not show substantial variation in terms of distribution of source languages, the higher proportion of technical vocabulary found in the objects subset of manor unveils a relatively wider pool of variants denoting instruments and tools used on manorial estates. These results show that it is possible to investigate specificity of meaning across different domains by looking at the distribution of lexical material at different levels of technicality. When it comes to measuring technicality, however, we recognize that further research is necessary to fully test the applicability of the CL system to a wider range of onomasiological resources presenting different conceptual classifications.
53Finally, with regard to survival and obsolescence rates, the results of this study reveal a general tendency for lexical material to survive past 1500, albeit relatively higher in church. Further investigation taking the interaction between synonyms of foreign and native origin into account might confirm that, as found by Sylvester & Tiddeman [2023], and Sylvester, Tiddeman & Ingham [2022, 2023], the influx of loanwords in these domains generally did not result in significant obsolescence rates in the native lexis, contrary to traditional views of the outcomes of borrowing in Middle English.
54In view of the limited amount of data employed in this study, we acknowledge that further investigations, possibly covering larger datasets and more lexical domains, may be helpful in expanding on these findings and shedding further light on the effects of contact-induced change in domains associated with different social contexts of late medieval England. In addition, further research attempting to distinguish between French and Latin etymology may be able to provide a more detailed view of the effects of borrowing in these major sociolinguistic environments.