A diachronic analysis of the adjective intensifier well from Early Modern English to Present Day English

Abstract

While the use of well as an intensifier of most adjectives had supposedly died out by Early Modern English (Fettig 1934: 186, Mustanoja 1960: 327, Stenström 2000: 188, Ito and Taglimonte 2003: 278), no studies have empirically examined its frequency diachronically. The present study traces its use from Early Modern English (1560) to Present Day English, (2014) using five speech-related corpora in addition to various dialectal sources and audio/video clips. Results indicate that this use was retained in some dialects of English despite not being attested in the Corpus of English Dialogues nor the Old Bailey Corpus, which document predominantly the incipient standard variety. A qualitative analysis of its use reveals a potential diachronic shift in its stress pattern and the scale structure of its intensified heads. Moreover, the present study discusses some of the methodological challenges which arise when using historical corpora to investigate linguistic change, drawing particular attention to sample representativeness and data analysis.

Résumé

Bien que l'on ait cru que l'usage de well ‘bien/très’ comme modificateur intensifiant d'une majorité d'adjectifs avait disparu au temps de l'anglais moderne naissant (Fettig 1934: 186, Mustanoja 1960: 327, Stenström 2000: 188, Ito et Taglimonte 2003: 278), aucune étude n'a examiné de façon empirique sa fréquence diachronique. L’étude actuelle trace l'usage de well depuis l'anglais moderne naissant (1590) jusqu’à l'anglais moderne (2014) en utilisant cinq corpus de langue parlée, en plus de plusieurs sources audio-vidéo de l'anglais dialectal. Les résultats indiquent que cet usage de well a été maintenu dans certains dialectes de l'anglais, malgré son absence du Corpus of English Dialogues et du Old Bailey Corpus, qui documentent principalement la naissance du dialecte standard. Une analyse qualitative de son usage révèle un déplacement diachronique de son modèle accentuel, et la structure scalaire des têtes qu'il modifie. L’étude actuelle expose quelques défis méthodologiques qui se présentent quand on utilise des corpus historiques pour enquêter sur les changements linguistiques, en attirant l'attention sur la représentativité des échantillons et sur l'analyse des données.

1. Introduction

Previous synchronic studies have pointed out that in Present Day British English (PDE) the adverbial well can function as an adjective intensifier (Stenström 2000; Stenström, Andersen and Hasund 2002; Stratton 2018).¹ Some examples of its use include they're well nice (Stenström 2000: 187), I am well glad and this is well knackering (Stratton 2018: 803). While this use of well has been attested throughout various historical stages of the English language, according to Fettig (1934: 15–21) and Mustanoja (1960: 319–328), its highly frequent use declined around the mid-14^th century, but remained as an intensifier of participial adjectives and a limited number of regular adjectives, such as aware and worthy (Stenström 2000: 177). In the present study, regular adjectives are defined as adjectives which are not morphologically derived from verbs (e.g., happy, tall, big) in contrast to deverbal or participial adjectives (e.g., educated, impressed) which are so derived; the latter being what Biber et al. (1999: 530) refer to as “derived adjectives”. Despite this decrease in frequency and constrained distribution, Stenström (2000: 177–190) noted its “revival” among British speakers in a corpus study from the late 20^th century.² Referring to the use of well as an intensifier of regular adjectives as a “revival” suggests that between its decrease in frequency in the mid-14^th century (Mustanoja 1960: 319-327), and its apparent resurgence in the late 20^th century (Stenström 2000), its use as an intensifier of regular adjectives ceased to exist (Ito and Tagliamonte 2003: 278). However, this is a diachronic question which has not yet been investigated. Because post 15^th century grammars and linguistic commentary do not document this intensifying use of well (Kühner 1934: 102-104, Kirchner 1955: 85, Collins COBUILD English language dictionary 1987: 1655, Longman dictionary of contemporary English 1987: 1195), its present-day use is often perceived as an “innovation” (Milroy 1992: 198, Paradis 2008: 322, Aijmer 2018: 84). However, a question which emerges is whether this current intensifying use reflects a continued development from Middle English, or whether it is an innovation on the part of modern language users, resulting in the return of same use of well seen pre-15^th century, but via different paths.

According to Labov (1994: 11), “historical linguistics can be thought of as the art of making the best use of bad data”. Since “historical documents survive by chance and not by design”, historical linguists are limited to the data which are preserved. However, to exacerbate the problem, only a portion of the preserved manuscripts are represented in historical corpora, an acknowledgement which has led to Lauersdorf's recommendation of using “all the data” in a historical analysis (Lauersdorf 2018a: 112; Lauersdorf 2018b: 211–213). While corpora have undoubtedly revolutionized the methodological plane of historical linguistics (Curzan and Palmer 2006, Curzan, 2008, Rissanen, 2012), they can also present various challenges and shortcomings (Joseph and Janda 2003, Rissanen 2008). Even the texts which are included in corpora can often present issues of faithfulness if they have undergone editorial intervention (Kytö and Pahta 2012: 125-127). Further issues can subsequently arise with corpus annotation and quantification (Stratton, in press). In the present study, the diachronic analysis of well is contextualized in relation to these methodological challenges and caveats.

This article addresses two specific questions. Firstly, other than being an intensifier of a limited set of regular adjectives such as aware and able (Stenström 2000), is there any evidence to suggest that well as an intensifier of other regular adjectives really disappeared between Early Modern English (EModE) and Present Day British English (PDE)? Secondly, which methodological challenges arise in investigating its use diachronically, and why is it crucial to account for these before arriving at any conclusions? Section 2 starts by laying out the terminological framework used to describe intensifiers, followed by a review of the literature on well, and a review of the literature on the recycling of intensifiers. Section 3 outlines the methodology by introducing the corpora and sources which were used and provides a description of the data collection process. The results are provided in Section 4, which are subsequently discussed in Section 5 in light of the two research questions. The findings and conclusions are then summarized in Section 6.

2. The intensifier well

The following section divides into three parts. First, 2.1 reviews the terminology used to describe intensifiers. Second, 2.2 reviews the previous literature on the diachrony of the intensifier well, covering its use from Middle English to Present Day English. Third, 2.3 discusses the ebb and flow of linguistic forms. These three subsections subsequently set the stage for the present analysis.

2.1 What are intensifiers?

The use of intensifiers in the English language has been a topic of much linguistic discourse (Stoffel 1901, Bolinger 1972, Paradis 1997, Lorenz 2002, Nevalainen and Rissanen 2002, Ito and Tagliamonte 2003, Tagliamonte and Roberts 2005, Xiao and Tao 2007, Méndez-Naya 2008, Fuchs 2017, Stratton, 2020). Before discussing the intensifier well, it is useful to discuss what intensifiers are; there is, however, little consensus regarding the appropriate terminology. Stoffel (1901) originally referred to intensifiers as “intensive adverbs”, Bolinger (1972: 18) referred to them as “degree words”, and Paradis (1997) referred to them as “degree modifiers”. According to the terminology of Quirk et al. (1985: 590), intensifiers can be subdivided into “amplifiers” and “downtoners”. The former “scale upwards from the assumed norm”, as in that's very interesting, whereas the latter scale “downwards from the assumed norm”, as in that's somewhat interesting. Amplifiers are then further subdivided into “boosters” and “maximizers”. Boosters “denote a high degree, a high point on the scale”, as in that's very good, and maximizers “denote the upper extreme point on the scale” as in he's completely drunk.

Accordingly, the adjective intensifier well is an amplifier; more specifically, a booster. However, in the present study the more generic label ‘intensifier’ or ‘degree modifier’ is used throughout to describe the use of well as an intensifier of adjectives. This practice is consistent with previous scholarship on well (Stenström 2000, Stenström, Andersen and Hasund 2002, Ito and Tagliamonte 2003, Stratton 2018). However, it should be noted that the adverbial well can also function as an intensifier of other parts of speech such as verbs, adverbs and prepositional phrases, the latter two often in metaphorical senses such as well underway and well on track. The fact that in PDE well can intensify regular adjectives as well as verbs may pose a problem for the traditional test used to examine whether a participle is a deverbal adjective. This is because traditionally, participles are considered to have become deverbal adjectives if they can be modified by very as opposed to well. Therefore, I am well confused becomes ambiguous in PDE since it would have been unambiguously considered a passive (i.e., someone/something has confused me well) but because well can intensify adjectives I am well confused could mean ‘I'm very confused’ (i.e., copula + intensifier + deverbal adjective).

2.2 The adjective intensifier well

Stenström (2000) noted the possible “revival” of the adjective intensifier well in a corpus study of British English from the late 20^th century, and a more recent study suggests that this intensifier has increased in frequency over the last two decades (Aijmer 2018). While it is currently used as an adjective intensifier, such use is not a new phenomenon, as it has been attested throughout various stages of the history of the English language. Some examples from Middle English are reported in (1). The examples show well intensifying regular adjectives, as in wel old (1a), wel joyous (1b) and well happy (1c), which is different from the intensification of participial adjectives, as in he is well acquainted with the facts and I was well impressed, which are morphologically derived from verbs.³

According to Fettig (1934:15-21) and Mustanoja (1960: 319-327), the adjective intensifier well became most popular in the 13^th century in the South and South Midlands of England when swiþe ‘very’ lost its popularity. Then, by the mid-14^th century, well reduced in frequency and eventually gave way to the intensifiers ful and riht (Fettig 1934: 186). In the 15^th century, well was reportedly rarely used outside collocations such as wel worth, wel wār and wel content (Fettig: 1934, 186, Mustanoja 1960: 327). To use the words of Fettig (1934: 186), “in Shakespeares Zeit ist wel als Intensiv nur noch selten anzutreffen” (in Shakespeare's time, the intensifier well is found only seldom).⁴ Nonetheless, well continued to intensify a limited number of regular adjectives such as aware and able. These collocations (e.g., well aware and well able) are still considered Standard English today (Stenström 2000: 178). In the present study, this limited set of adjectives which are intensified by well is referred to as the vestigial collocations. This description is used to disambiguate the regular adjectives which well continued to modify after its decline in frequency in the mid-14^th century (e.g., well aware, well able, well worthy, etc.) from the apparently ‘newer’ regular adjectives well intensifies today in PDE (e.g., well cute, well tacky, well jazzy). The term ‘newer’ is borrowed from Aijmer (2018: 86) who describes its 21^st century use as “the new well”.⁵

On the one hand, it is possible that these vestigial collocations have remained throughout the history of the English language because they have become fixed due to their high frequency. As Méndez-Naya (2003: 377) points out, a decrease in frequency of intensifiers results in a reduction of collocates, and the collocates which remain are restricted, and exist only because they have become fossilized throughout time. The vestigial collocations (e.g., well aware, well able, well worthy) belong to a formal register whereas the ‘newer’ use of well belongs to an informal register (e.g., well classy, well cool, well lush, well saucy, well gross and well dodgy).

On the other hand, however, there is some collocational evidence to suggest that scale structure may have also played a role. This is because what adjectives such as aware, able and worthy have in common is that they have a closed scale structure, that is to say, a scale which has a minimum and maximum value. In contrast, adjectives such as intelligent and rich have an open scale structure since they do not have a minimum and maximum threshold. In other words, there is no limit to intelligence and wealth, which is also true for tall, for instance, since there is no maximal degree of tallness. In contrast, words like bent have a closed scale structure, because something is either bent or it is not (i.e., there is a minimum and maximum value). Adjectives which have an open scale structure can be called “relative adjectives” whereas adjectives which have a closed scale structure can be called “absolute adjectives” (Kennedy and McNally 2005, McNally and Kennedy 2002: 85). According to the diagnostics laid out in work by Kennedy and McNally (1999, 2002, 2005), closed scale adjectives permit intensification with completely and fully, whereas open scale adjectives permit intensification only with very. Using slightly different terminology, this difference in scale structure maps onto work by Paradis (1997, 2001, 2008) where absolute adjectives, referred to as “bounded”, can be modified by “totality modifiers”, and relative adjectives, referred to as “unbounded”, can be modified by “scalar modifiers” (Paradis 2008: 321-322). This difference between bounded and unbounded adjectives is conceptually similar to that of open and closed scale adjectives whereby bounded adjectives have “an absolute maximum or…minimum” and unbounded adjectives are “scalar” (Paradis 2008: 322). Therefore, with regard to well in semantic terms, it is possible that after its decline in frequency in the mid-14th century, well continued to intensify only adjectives with a closed scale/bounded structure. This would account for why well continues to intensify adjectives such as aware, in most, if not all varieties of English today, whereas, using well to intensify adjectives which have an open scale/unbounded structure is infelicitous in non-British varieties of English (Kennedy and McNally 1999: 170, McNally and Kennedy 2008: 85).

A prediction of this scalar explanation is that, once delexicalized, decaying intensifiers recede to certain sectors of the semantic space, that is, intensification of adjectives with only a closed scale structure. While this is an empirical question subject to diachronic analysis, intuitively one might expect that, in receding uses, open scale intensification is less frequent than closed scale intensification, since open scale adjectives operate on a gradable scale which is context-dependent (i.e., the meaning changes contextually). Therefore, this context-sensitivity makes open scale adjectives challenging to define because when the standard of comparison is not fixed, the truth conditions may vary (Kennedy 2007: 2). In contrast, closed scale adjectives consist of two clearly defined minimum and maximum values which provide coordination points for resolving issues of semantic uncertainty (Kennedy 2007: 42). Since intensifiers modulate the threshold of a scale, they are sensitive to an open/closed scale distinction (Kennedy and McNally 1999, Kennedy and McNally 2005). Thus, in short, the lack of context needed when intensifying open scale adjectives with well may be a reason why the vestigial collocations remain unaffected in the language, while well as an intensifier of open scale adjectives reduced in frequency to the point where it was thought to have fallen out of use.

In addition to the vestigial collocations, after its decline in frequency in the mid-14^th century, well also continued to modify participial adjectives (Ingersoll 1978: 196). However, as Bolinger (1972: 29) points out, well does not always function as a marker of degree (or “fulfilment”) with participles, as it can sometimes function as an adverb of manner (or “approval”). The examples in (2) illustrate this. In (2a), well is contrasted with poorly or badly and expresses the idea of approval, that is, the book was written well, and we approve of the way in which it was written. Using different terminology, Kennedy and McNally (1999), and Cattell (1999: 61), might describe this “manner” use of well as expressing “quality” as opposed to “quantity”. Whereas, in (2b), well expresses that John was not only mauled by the lion, but to a high degree or intensity, which thus presents a degree reading. When examining the diachronic use of the intensifier well, it is therefore important to bear this distinction in mind, since its manner use is not functionally equivalent to its use as a manner adjunct. However, as Bolinger (1972: 9) points out, there are times when it is difficult to distinguish between the two readings. An example given by Kennedy and McNally (1999: 177) is a well-loaded truck, which can entail that the truck was loaded in a good manner (i.e., the truck was loaded appropriately) but at the same time can also entail that the truck was loaded to a high degree (i.e., the truck was loaded to the point where there is little to no room left).

2.3 Recycling of intensifiers

In the words of Stenström, Andersen and Hasund (2002: 158), it is “intriguing” that the adjective intensifier well is found in British speech in the late 20^th century despite it apparently falling “out of use”. However, the recycling of intensifiers throughout the history of a language is not uncommon (Mustanoja 1960: 319-327, Peters 1994, Ito and Tagliamonte 2003, D'Arcy 2015). It has been argued that one of the reasons for the constant recycling of intensifiers is what Bolinger (1972: 18) refers to as “fevered invention”; this is, according to Tagliamonte (2012: 334) “a process driven by speakers’ desires to be original, demonstrate verbal skills and to capture attention”. “The reuse and recycling of forms is a reason why Modern English has such a wide array of intensifiers” (Romero 2012: 10) which are “locally idiosyncratic” (Barnfield and Buchstaller 2010: 254). If speakers overuse intensifiers, the intensifiers begin to lose their novelty, which is why “speakers must reinvent them from time to time” (Tagliamonte 2016: 92). Peters (1994: 271) points out that the recycling and renewal of intensifiers is particularly prominent after EMoDE. Therefore, the fact that well decreased in frequency after the mid-14^th century (Mustanoja 1960: 319-327) but was found again in PDE (Stenström 2000: 178) is in line with the phenomenon of recycling. As Tagliamonte (2008: 389-391) points out, some intensifiers do not disappear but instead remain at the speakers’ disposal as low- frequency variants which “are available to be co-opted back into the active system” at any time. Barnfield and Buchstaller (2010: 268-271) illustrate this with the intensifier dead which dates back to the 16^th century (Blanco-Suárez 2014: 18), but subsequently went below the observable radar, to be then picked up in the 20^th century. However, it should be noted that although intensifiers are known to fluctuate and compete in frequency, they are not unique in their ability to be recycled, since various linguistic forms and structures wax and wane throughout time (Hickey 2002: 105-128, Mair 2004: 135-136, Buchstaller and Traugott 2006: 363-364).

A relevant question to ask is how it came to be that well was chosen to be “revived” over other functionally equivalent variants. The question of why a particular change takes place in a particular language or variety at a particular time is what Weinreich, Labov and Herzog (1968: 102) refer to as the “actuation problem”. To investigate how and why this change took place, it is necessary to investigate the system of intensifiers as a whole, since it is widely understood that a change in frequency of an intensifier can trigger a change or rearrangement of the system of intensifiers within a speech community due to their multi-dimensional nature (Ito and Tagliamonte 2003, Tagliamonte 2008, Méndez-Naya and Pahta 2010). From a variationist standpoint, to answer what actuated its “revival”, it is also necessary to examine the social factors such as the sex and age of the speaker, which may condition its use (Tagliamonte and Roberts 2005, Tagliamonte 2008, Fuchs 2017). While the question of why this change took place is beyond the scope of the present study, such macro questions should be examined in future research.

3. Methodology

The following section lays out the methodology, divided into two subsections. First, the sources of linguistic data used in the present analysis are delineated and justified in 3.1. Second, the data collection process, which involves a discussion of the data included or not included in the pool of analysis, is reported in 3.2.

3.1 The corpora

Given that apart from the vestigial collocations, the use of well as an intensifier of regular adjectives is not considered Standard English (Stenström 2000: 178), it is likely that its use is more prevalent in spoken language. Moreover, as D'Arcy (2015: 451) points out, “intensification is primarily a dialogic phenomenon, rendering vernacular evidence particularly valuable”. For this reason, the present study focuses predominantly on spoken or speech-related language. As with the growing empirical interest in the use of language in fictional television as a means of investigating synchronic variation and change (Tagliamonte and Roberts 2005, Reichelt and Durham 2016, Stratton 2018), the use of speech-related historical texts has also proven to be a fruitful proxy for studying the spoken language of historical times (Jucker 1995, Culpeper and Kytö 2010, Mazzon and Fodde 2012). While studying spoken language through an orthographic lens is not ideal (Kytö and Walker 2003), use of speech-related or informal types of production “partly remedy this shortcoming” (Claridge 2008: 247). Moreover, previous research has already investigated the competition between the intensifiers full, well and right in written Late Middle English and Early Modern English medical texts (Méndez-Naya and Pahta 2010). Therefore, because of the lack of attention to speech-related texts, and the fact that spoken language may have captured this change more effectively than written language, speech-related corpora were chosen as the main source of linguistic data in this study.

To explore the use of the intensifier well from EModE (1500-1700) to Late Modern English (LModE: 1700–1900) to PDE (1900+), five speech-related corpora were used: the Corpus of English Dialogues 1560–1760 (CED), the Old Bailey Corpus 1720–1913 (OBC), the Diachronic Corpus of Present-Day Spoken English (early 1950s to 1980s and late 1990s, DCPSE), the spoken component of the British National Corpus 1994 (BNC1994), and the spoken British National Corpus 2014 (BNC2014). The CED consists of speech-related texts which can be divided into “authentic dialogue” and “constructed dialogue”. The authentic dialogue consists of face-to-face interactions, trial proceedings and witness depositions, all of which are records of real speech events. In contrast, the constructed dialogue consists of dialogue from drama, comedy, and didactic works.

The OBC consists of authentic dialogue from courtroom proceedings which were taken down in shorthand by scribes. In light of the absence of physical audio recordings, both the CED and the OBC are arguably as near as one can get to the spoken word during this period and offer a rare opportunity to study speech-based language which was used in EModE and in LModE; (for the CED, see Culpeper and Kytö 2010, and for the OBC, see Huber 2008). The DCPSE is a spoken corpus with a variety of spoken genres, such as face-to-face communication, telephone communication, broadcast interviews and discussions, and spontaneous commentary. The DCPSE contains material from two time periods: the London-Lund material (from the 1950s to the early 1980s) and the International Corpus of English (ICE-GB) material (for the 1990s). As for the two BNC corpora, they are collections of orthographically transcribed spoken British English. The original BNC has orthographically transcribed spoken language from approximately 1994, and the Spoken BNC2014 has orthographically transcribed spoken language from approximately 2014. These two corpora were designed in a similar way, given that they are each other's counterparts (Love et al. 2017: 321). A summary of the corpora and their size is reported in Table 1.

Table 1: The corpora used in the present study

In addition to the five speech-related corpora, the data were supplemented with dialectal sources (listed under Dictionaries and Manuscripts) such as the computerized version of Wright's English dialect dictionary (see Markus 2007), in an attempt to explore whether the intensifier was retained in peripheral English dialects. This is because while a good selection of corpora was chosen, these corpora represent mostly south-eastern British English and mostly the incipient standard variety. Furthermore, because the use of well as an intensifier of most adjectives is considered non-standard (Stenström 2000: 178), it is crucial to also examine data from as many different non-standard British dialects as possible. Some other dialectal sources included were: Sternberg's The dialect and folklore of Northhamptonshire (1851), Ferguson's The dialect of Cumberland (1873), Tim Bobin's comic dialogue Tummus and Meary written in the Lancashire dialect (1819), the Diary of Joseph Turrill of Garsington (1863-1876), and The historians of Scotland (1872). However, the use of these dialectal sources is ultimately qualitative as opposed to quantitative, since an attestation or entry in these sources reveals little about frequency. What an attestation of the intensifier well does reveal, however, is its retention in certain dialects of British English. The Freiburg English dialect corpus (FRED), which contains dialectal audio recordings predominantly from the 1970s to the 1990s (so-called “oral histories”), was also included in the present dataset.

3.2 Data collection

When analyzing the frequency of a given linguistic item across corpora, in the traditional corpus approach, tokens are normalized by a common denominator, such as per million words (D'Arcy 2015: 458). However, this method has come under criticism in variationist work since it does not account for the number of times a linguistic form could or could not have occurred (Tagliamonte 2012: 19, Wallis 2013: 7, D'Arcy 2015: 458, Gries 2015: 110, Fuchs 2017: 351). This inclusion of both occurrence and absence of intensifiers within a functionally equivalent context is referred to as “circumscribing the variable context” (Poplack and Tagliamonte 1989: 60) or defining “the envelope of variation” (Milroy and Gordon 2003: 180). With respect to well, this means contrasting the number of times it intensified adjectives (e.g., he is well tall) against the number of times it could have intensified adjectives (e.g., he was ∅ tall). Since “the absence of the unit in question can be an intentional choice to not use it” (Gries 2018: 287), an accountable analysis considers both occurrence and absence. Recent research has found that the decision to use either the number of words in a text/corpus or the number of functionally equivalent contexts in a text/corpus as a means of analyzing frequency can shape the conclusions of an analysis (Biber et al. 2016). In light of this, while work on intensification cogently argues for the variationist approach (Ito and Tagliamonte 2003, Tagliamonte and Roberts 2005, Tagliamonte 2008, D'Arcy 2015), because the quantitative method may significantly affect the outcome, both units of analysis are reported in the present study. The chosen NF (normalized frequency) for the present study was per 100,000 words.

The OBC, the CED, the BNC1994 and the BNC2014 were publicly available online. These corpora were accessed via the CQPweb interface.⁷ A license was required for the DCPSE, and this corpus was accessed via CD-ROM. To search the corpora, different types of POS (Part of Speech) tagging were used according to the quantification method adopted (i.e., frequency normalization vs variationist approach). For frequency normalization, a search query was run to search for all instances where well occurred before an adjective in the corpora. Since the POS tagging varied from corpus to corpus, the queries differed from each other. For instance, in the spoken component of the BNC1994, the query well _AJ0 was necessary, whereas in the OBC and the CED the query well _JJ was necessary. Instances of ambiguity (i.e., when it was not clear whether well was functioning as an adjective intensifier or whether it was as an interjection) were omitted from the present dataset.⁸ To ensure that instances of intensification were not missed due to incorrect tagging, a general search query for the lexical entry ‘well’ was also run in each corpus without any POS tagging. This search query compiled all instances of well regardless of its form or function, which were subsequently checked and contrasted manually with the search of well which did include POS tagging. In doing so, it became clear, for instance, that in the CED woorthie was incorrectly tagged as a noun, which meant that well woorthie did not appear in the initial search query well _JJ. These missed instances were then added to the pool of analysis.

For the variationist approach, a search query was run in order to collect all adjectives in the corpora. The adjectives intensified by well which, due to incorrect tagging, did not appear in the concordance lines, were also added to this pool of analysis, and a sample of the tagged adjectives was checked for tagging accuracy. This procedure was important since automatic taggers are not error-free and are more prone to errors on historical language given that the taggers are typically modeled on modern standard English (McEnery and Hardie 2012: 30, Stratton, in press). Following Ito and Tagliamonte (2003) and subsequent variationist work on the intensification of adjectives (Tagliamonte and Roberts 2005, Tagliamonte 2008, D'Arcy 2015), non-gradable, comparative and superlative adjectives, in addition to negative contexts, were manually removed from the pool of adjectives.⁹ According to Biber et al. (1999: 508), adjectives can be divided into “descriptors” and “classifiers”. Since classifiers are non-gradable (e.g., additional), they were also removed from the pool of adjectives. The remaining adjectives were then coded based on whether they had been intensified by well or not. Therefore, the variationist approach to measuring frequency allowed for a direct comparison of the frequency of the intensifier well proportionate to the number of times adjectives could have been intensified by well per corpus.

4. Results

The following section reports the empirical findings of the use of the intensifier well from 1560 to 2014. The frequency of well as an intensifier of regular adjectives and the vestigial collocations are reported in Section 4.1. The frequency of well as a modifier of participial adjectives (divided into manner and degree) is provided in Section 4.2.

4.1 Well + regular adjectives

A graphical representation of the normalized frequency of the vestigial collocations (e.g., well aware, etc.) and the normalized frequency of well as an intensifier of regular adjectives from 1560 to 2014 is provided in Figure 1. The bold line indicates the frequency of the regular adjectives intensified by well, and the lighter line indicates the frequency of the vestigial collocations. The only instances in the speech-related corpora of well intensifying regular adjectives before the 20^th century were the vestigial collocations. Examples are provided in (3)–(4). Therefore, these five speech-related corpora create the impression that the use of well as an intensifier of other regular adjectives (beyond the vestigial collocations) is a late 20^th century development.

Figure 1: The normalized frequency of the adjective intensifier well from EModE to PDE¹⁰

The NF (normalized frequency) of well as an intensifier of regular adjectives in 1990–1992 was 0.75 per hundred thousand words (or 3 tokens), 1.29 per hundred thousand words (or 43 tokens) in 1994, and just over 2 per hundred thousand words (or 163 tokens) by 2014.¹¹ In contrast, as expected, the vestigial collocations do not disappear from the language. Based solely on the speech-related corpora, the fact that this use of well (as an intensifier of regular adjectives) appears to resurface around the late 20^th century corroborates the findings of Stenström (2000) who observed its apparent “revival” among speakers in the geographical region of London. The frequency measured by number of possible intensifiable contexts is provided in Figure 2. Strikingly, the normalized frequencies in Figure 1 depict a distribution whereby the vestigial collocations suddenly spike in frequency after the time around which the adjective intensifier well was supposed to have decreased in frequency. However, the spike in frequency in the late 18^th century is leveled out when the measure of frequency is based on the variationist paradigm. One reason the normalized frequency spikes from 1761 to 1841 is the significant change in number of words from the CED to the OBC, even though this increase in the number of total words does not account for the ratio of intensifiable contexts. Based on previous literature (Fettig 1934, Mustanoja 1960), the OBC figures stand out as being unexpectedly high. Potential skewing through highly formulaic, possibly legalese language in the OBC could, however, be a contributing factor. Nevertheless, the overall diachronic trend shows a decline in frequency. Figure 2 presents the more anticipated picture, namely that the highly frequent vestigial collocations decrease in frequency after Middle English but increase toward the late 20^th century.¹²

Figure 2: The frequency of the adjective intensifier well from EModE to PDE as measured by intensification rate

In short, the data from the speech-related corpora suggest that other than the use of well in a few vestigial collocations, well did not resurface as an intensifier of regular adjectives with an open scale structure (e.g., well big, well cool, well dirty) until the late 20^th century. If the analysis were to stop here, one might conclude that the answer to the first research question (i.e., did the use of well as an intensifier of regular adjectives really disappear?) is yes. However, by looking further into various dialectal records, it becomes clear that well was in fact retained as an intensifier of regular adjectives in some British dialects. Some attestations from the EDD are reported in (5).

Beyond the EDD, Kühner (1934) documents the sentence “came well wet to the Callander” from 1566; and the sentence “I should think they are well drunk by this time” can be found in the Diary of Joseph Turrill in 1867. The collocation well dry is also attested in the OED and “the letter was carried to Westminster before the ink was well dry” is attested in 1643/1644 in the newspaper Mercurius Aulicus. The fact that well is found intensifying these adjectives is not surprising given that wet and dry have both “relative and absolute uses” (Kennedy and McNally 2005: 370). In the examples found in Kühner (1934) and Mercurius Aulicus, wet and dry have a closed scale structure, (i.e., they have an absolute interpretation). This use is also attested in the OED: A new English dictionary on historical principles (1928: 285), dating back to examples from 1728 such as he was once well warm. Looking into the Lancashire dialect, one finds that welly (a dialectal form of well) was retained as an intensifier of adjectives; two examples from the comic dialogue of Bibbon's 1819 Tummus and Meary, written in the Lancashire dialect, are provided in (6a–b).

As for the variant welly, two points are worth noting. Firstly, welly is thought to have undergone semantic change in some English dialects (Ferguson's The dialect and folklore of Northamptonshire 1851: 121). While it still functioned as an intensifier in the Lancashire dialect, it scaled down the meaning of an adjective (i.e., downtoner) as opposed to scaling up its meaning (i.e., amplifier). Therefore, an example like welly killed is functionally equivalent in meaning to ‘almost killed’. The second point is that moydart, as in (6b), appears to be more of a participial adjective rather than a regular adjective due to the dental suffix <t> which is a feature of past marking in Germanic languages (see German spielte ‘played’ and English walked [wɔːkt]). The OED attests the regional verb moider meaning ‘to confuse or bewilder’ (moider, v. 1) which is in line with Tim Bibbon's translation of (6b). Therefore, (6b) can be interpreted as being ‘confused to a high degree’ (i.e., very confused).

The Brigham Young University (BYU) suite of corpora recently released a TV corpus which contains orthographically transcribed informal British English from 1950 to 2018. Using this corpus, one encounters the intensifier well in the 1980s (7a–e). Because these tokens can be checked with the audio, the prosody makes it clear that well was functioning as an intensifier, not a discourse marker. The examples in (7a–b) show that well was used as an intensifier of adjectives in 1987 and 1988. Anecdotal evidence also supports that well was used this way in the early 1980s. For instance, one online metalinguistic thread reports that well was used frequently in Lancashire from 1980 to 1982.¹⁴ This anecdotal evidence is backed up with the fact that the British punk band The Blood released a song in 1983 titled I'm well sick. In the chorus they sing I'm well sick, I'm well sick, I'm sick as a pig. Not only is it clear in the song from the stress that well is intensifying sick, but the simile I'm sick as a pig indicates the songwriter's wish to intensify the degree of sickness. In the sitcom Hale and Pale, the use of well is mocked in a three-minute skit in an episode aired in 1989. The mocked examples are reported in (7f–g). In summary, while the use of well as an intensifier of regular adjectives was not documented in the five speech-related corpora, which might suggest that it was not used very frequently, its use did not completely disappear from English dialects.

4.2 Well + participial adjectives

The frequency of well as a modifier of participial adjectives from 1560 to 2014 is reported in Figures 3 and 4. As mentioned, because well does not always function as a marker of degree, especially with participles, it is necessary to differentiate its use as a degree modifier from its use as a manner adjunct. However, there are several instances where the reading is ambiguous. After excluding the ambiguous tokens, the distribution of the data indicates that the use of well as both a marker of manner and a marker of degree gradually decreases after Middle English. Figure 3 presents the data using normalized frequencies, and Figure 4 presents the data following the variationist approach.

1. (8) 16^th century

   1. Manner:

   2. (a) they were well brought vp in the Court

   3. (b) before the musicke was well tuned

      Degree:

   4. (c) they are well experienced

   5. (d) I am well satisfied

      Ambiguous:

   6. (e) it hath been well debated and consider'd by us all

Figure 3: The normalized frequency of well as a modifier of participial adjectives from EModE to PDE

Figure 4: The frequency of participial adjectives intensified by well from Early Modern English to PDE

In the beginning of the EMoDE period, well was frequently used, with participles, as both an adverb of manner and a marker of degree (see Figures 3 and 4). In (8a), well has a manner reading because it describes how the subject (‘they’) was brought up, namely ‘well’ as opposed to ‘badly’. The same is true for (8b) since music can be tuned ‘well’ or ‘poorly’ meaning that well expresses the quality of the tuning (i.e., the manner). However, in (8c), well functions as a marker of degree since it is not possible to modify experienced in terms of binary approval. Instead, well intensifies the degree to which ‘they’ are ‘experienced’, be it somewhat experienced, very experienced, or extremely experienced (here they are well experienced). The same is true for (8d). If someone is satisfyd ‘satisfied’, they are already content. Therefore, well intensifies the degree of satisfaction (e.g., ‘highly satisfied’).

For well to intensify a participle, the verb from which it comes has to be gradable, that is, capable of being understood in terms of “more” or “less”. This is especially clear with stative verbs such as to love, to please, to know and to appreciate (OED, s.v. well. adv and n4, 13c, d). In contrast, event verbs such as to arrive are typically non-gradable. The reason stative verbs are typically gradable is that their scales are usually abstract. As McNally and Kennedy (2008: 242) ask, is there a maximal degree of liking or loving? Moreover, according to McNally and Kennedy (2002: 6), participles which typically permit modification can be modified by fully. The fact that one can say he was fully satisfied confirms the degree reading of well in (8d) and moreover indicates closed scale intensification. There were, however, examples where the function of well was ambiguous. This is the case in (8e) because something can be debated ‘well’ or ‘poorly’, but something can also be debated to high or low degree. For further early 16^th century attestations of unambiguous examples of well as a degree marker, such as we were wel affrayd then (see OED: A new English dictionary on historical principles,, vol. 10, 1928: 284).

1. (9) 17^th century

   Manner:

   1. (a) The trees are well placed

   2. (b) A well reputed Gentleman

   3. (c) he's a very well bred man

      Degree:

   4. (d) who was well contented to haue taken a nappe

      Ambiguous:

   5. (e) you seem a person well qualified

   6. (f) Thou shalt be well rewarded for thy labour

   7. (g) as I am by my counsell well informed

The degree use of well with participles continues until its present-day use and can thus be seen in the 17^th century with examples such as well contented and well satisifed. Examples in (9e–g) are ambiguous. For instance, in (9e), people can be qualified in an appropriate manner and there are different degrees to which someone can be qualified. In (9f), well is ambiguous because, on the one hand, you can be rewarded ‘well’ or ‘poorly’ depending on how much of a reward is given. If the reward involves receiving a large sum of money, one is well rewarded (i.e., being ‘well rewarded’ is predicated to an individual, namely ‘thou’). If the reward is simply a pat on the back or some applause, then the reward is likely a poor one (i.e., approval = manner). On the other hand, being rewarded is something which is inherently positive. If someone is rewarded, one would not expect it to be necessary to mention that they were rewarded in a positive manner (i.e., ‘approval’) but instead well would likely refer to the degree to which one was rewarded. This is because you can also be rewarded to different degrees (e.g., given thousands of dollars, or millions of dollars, or billions of dollars). Well is also ambiguous in (9g) since someone can be informed in the appropriate or correct manner, but can also be informed to a high degree (i.e., not just informed, but well informed).

1. (10) 18th century

   Manner:

   1. (a) let me see if it be well made

   2. (b) she is a good honest well behaved girl

   3. (c) there was a well dressed woman

      Degree:

   4. (d) I am well assured that this is my property

   5. (e) I am well satisfied

   6. (f) I am well convinced

      Ambiguous:

   7. (g) Once we were both so well heated

   8. (h) I have heard Frankland say she was to be well rewarded for this fact

2. (11) 19^th century and 20^th century

   Manner:

   1. (a) I believe he was well equipped

   2. (b) That was a well timed run

   3. (c) Because the story is so good and so well written

      Degree:

   4. (d) you seem well acquainted with the technical phrases

   5. (e) Park Lane is a place pretty well frequented

      Ambiguous:

   6. (f) I think we should be [er be] well prepared to foot the bill

   7. (g) I can say to my honourable friend, the member for Rydale who takes such a close interest and is so well informed [er] on these matters

   8. (h) two well baked loaves, please

3. (12) 21^st century

   Manner:

   1. (a) he's a well behaved kid and he gets involved with a lot of stuff

   2. (b) oh I'll be well behaved today

   3. (c) you know it looked well kept

   4. (d) I'm like oh that's well written I like that

      Degree:

   5. (e) I was like well impressed cos it's taken her twenty years to buy me a book

   6. (f) if my parents got a divorce I'd be well upset

   7. (g) …bet you're well spoilt

   8. (h) Oh shit man was well worried

   9. (i) [name] and I are going to be well screwed by the time we may eventually end up having children

      Ambiguous:

   10. (j) they'll be well drained

   11. (k) make sure you're well hydrated beforehand

   12. (l) I am well confused with all of this stuff man

In the 18^th century, both the manner and degree use of well decreases in frequency, but the level of ambiguity remains. To focus on only one example from the 19^th century, the function of well in well baked loaves (11h) is ambiguous because it could be referring to the degree to which the loaves are baked, or it could be referring to the acceptability of the baking. Interestingly, before the 20^th century, well appeared to have a preference for intensifying mostly closed scale participial adjectives (e.g., satisfied, assured), but by the late 20^th and early 21^st century, well is frequently found intensifying deverbal adjectives with an open scale structure (e.g., impressed, upset, terrified, worried).¹⁶ McNally and Kennedy (2002: 6) state that “participles associated with open-scales, such as worried, do not permit modification by proportional modifiers, nor do they permit modification by well”. While this may true for North American English, this is no longer true for British PDE. The open-scale examples in (12e–i) illustrate this. Therefore, to summarize the diachronic data regarding scale structure, well appears to transition from an intensifier of mostly closed scale adjectives (both deverbal and non-derived adjectives) to an intensifier which widely permits open scale intensification.

5. Discussion

According to the five speech-related corpora, other than modifying a limited number of adjectives at a low frequency from 1560–1913, the use of well as an intensifier of other regular adjectives was not attested until the late 20^th century, which corroborates the initial observation of Stenström (2000: 188). However, when looking into dialectal sources it becomes clear that well as an intensifier of regular adjectives (i.e., adjectives other than aware, able, capable, etc.) had in fact been retained. This might suggest that although its use as an intensifier of regular adjectives did not completely disappear in all British dialects, it was – despite its lack of attestation in the speech-related corpora – used at a low frequency. The fact that its use was retained in some peripheral dialects is not unusual since socially, politically or geographically isolated dialects tend to be less affected by linguistic changes in mainstream communities, thus preserving relic features (Tagliamonte 2013: 5).

The absence of such use in the speech-related corpora relates to a widely acknowledged methodological challenge in corpus and historical linguistics, namely that the lack of attestation in a corpus (or corpora) does not necessarily mean that something does not or did not exist (Rissanen 1989). Since researchers in historical linguistics are largely restricted to the texts and sources which are at their disposal, using these as the sole source of evidence of a language at a given point in time can be misleading, since these texts are not representative of all varieties. This was the case with well in the speech-related corpora which are available for historical English. Had it been assumed that these corpora were representative of the historical periods in question, one might have arrived at a very different conclusion. Given that the selected corpora mostly represent the incipient standard variety of English, other varieties, in which the use of well as an intensifier of regular adjectives was retained, were not represented. In this respect, the present findings illustrate the importance of Lauersdorf's injunction to use “all the data” in a historical analysis (Lauersdorf 2018a: 112, Lauersdorf 2018b: 211-213).

Stenström (2000: 177) reported a conversation led by an adolescent who explicitly discussed his intensifying use of well, recounting their mother's confusion with its use, as in (13). The very fact that many speakers were unfamiliar with this use in the late 20^th century, and the fact that in non-British varieties of English this use of well is infelicitous (Kennedy and McNally 1999: 170), suggests that many British speakers in the 20th century were not exposed to this usage. In Bolinger's 1972 monograph, Degree Words, some infelicitous examples of the use of well as an adjective intensifier were provided such as it is well unlikely (41). The majority of these examples are now considered felicitous today in British English, according to the two BNC corpora and native speaker intuitions. Aijmer (2018: 84) also reported that speakers “experienced” this “new” use of well as something which is innovative even though it was used in Old and Middle English. This perception of novelty is therefore an example of the “recency illusion”, whereby recently noticed linguistic forms or structures are thought of as new despite having existed for centuries (Zwicky 2005).

1. (13) “My mum says, I go yeah that's well nice, and she goes erm well nice…(?)… That's it yeah I know I'm always saying well, well cool and I keep on saying that, I've said it like, about so many things when we're home, and she goes, what is this you always saying well with everything” (Stenström 2000: 177)

Since the data in present study indicate that its intensifying use was retained in some dialects, it is possible that this retained dialectal use spread in the 20^th century to the point where speakers who did not have it in their repertoire experienced it as something new. This synchronic novelty would have added a layer of originality in speech communities where it was uncommon to use well in this way, a somewhat expected development in light of what is known about the motivations for changes in the intensifier system when current variants lose their intensifying effect (Stoffel 1901: 02, Tagliamonte 2008: 391). However, at the same time, generally speaking, it is not common for diffusion to work from remote varieties outward since linguistic change typically spreads from urban center to urban center, and may or may not spread to rural sectors (Trudgill 1972, 1974; Chambers and Trudgill 1980: 176, Labov 2001: 285). For this reason, one also has to entertain the possibility that the current intensifying use of well is not the same as its previous Old and Middle English attested form. In other words, the speakers who are found using it in the 20^th and 21^st century may have developed its use through some mechanism such as analogy, such that the vestigial collocations give rise to “new” ones, for example, well aware → well bad (Stenström, Andersen and Hasund 2002: 158-159; Aijmer 2018: 85) and not due to the dialectally retained use of well spreading outward. In his article on ebb and flow, Hickey (2002: 125) warns researchers away from readily assuming that the use of present-day forms which existed historically were “transmitted unchanged throughout history”. If this analogical mechanism is the case with well, the present study might provide new insight with respect to the longitudinal ebb and flow of intensifiers. This notion is backed up by the several scholars who have explicitly suggested that the 20^th and 21^st century intensifying use of well is in fact an innovation among adolescents at least from a synchronic standpoint (Milroy 1992: 198, Paradis 2008: 322, Aijmer 2018: 84). While the origin of the current use of well is not entirely clear, there is some prosodic evidence to suggest that its present-day use is not entirely the same as its previously attested one.

The prosodic evidence is as follows. One cannot help noticing that the stress in 21^st century utterances such as that's well funny is different to the stress in instances such as he's well aware [of the facts]. When well intensifies regular adjectives other than the vestigial ones, it is stressed (e.g., he's wéll hard or that's wéll stupid) whereas in the residual use (e.g., he's well ‘aware) it is the adjective which is stressed, not well. The OED mentions this too: “in this use sentence stress usually falls on well rather than the following adjective” (well, adv. and n. iv16c.). By “in this use” the OED refers to the use of well as an intensifier. Therefore, in British English stress can now determine the function of well when it appears adjacent to a participial adjective. For instance, in the example he is well educated, if educated is stressed (i.e., he's well éducated) it is understood as ‘he is educated in a good manner’. In contrast, if it is well that is stressed (e.g., he's wéll educated), well has a degree reading (i.e., he is very educated). In FRED, well in the collocation well educated (e.g., they was well educated you see [Male NNT_003, Nottinghamshire, DOB 1902]) always expressed manner and was unstressed. The fact that when well intensifies the “newer” regular adjectives it is stressed, whereas when it intensifies adjectives such as aware and able it is not stressed, provides some support to the notion that the “revived” use of well did not emerge due to dialectal diffusion. If its present-day use did emerge due to spreading, one might expect the stress to be the same. This difference in stress may therefore show “layering”, that is, the co-existence of old forms alongside new ones (Hopper 1991: 22-23). While one cannot completely rule out the possibility that well was always stressed when functioning as an intensifier of regular adjectives, the fact that it is not stressed in non-British varieties of English where well can intensify adjectives such as aware and worthy (i.e., the vestigial collocations) suggests that the change in stress is innovative. Nevertheless, the present-day use may be an overlapping product of diffusion and an innovative extension.

Stratton (2018) analyzed the intensifier well in the British TV show The Inbetweeners. In the examples given, well was stressed (e.g., this is wéll knackering, that was wéll Jimmy Savel, that's wéll racist and you'll be wéll jel). This can be verified by listening to the audio-video recordings from the show. It is also worth noting that the present-day use of well tends to be stressed when intensifying prepositional phrases (e.g., that's wéll out of order). One way of testing whether well was always stressed would be to find texts which rely on poetic stress or contain a metrical tradition which would indicate whether it was stressed or not. However, to date, texts with such a metrical pattern in which well was also attested as an intensifier have not been found. Nevertheless, it is clear from the stress in the examples in (7), which were taken from audio-video recordings of British TV shows from the 1980s, that well was stressed, not the adjective (e.g., it was wéll brill and it was wéll naughty).

In examining the diachronic use of well, several methodological challenges arose. For instance, while POS tagging can identify instances where adverbs appear adjacent to adjectives, it reveals little about the function of the adverb itself and does not provide the suprasegmental information necessary for determining whether well was functioning as an intensifier. This is a challenge for studying historical phonology in general, since researchers are subject to studying phonology through an orthographic lens and, depending on the time period, have little to no available phonetic recordings to work with (Stratton, in press). As Curzan and Palmer (2006: 25) point out, computers can search texts much faster and much more reliably than humans can locate valuable linguistic data, but ultimately computers cannot read data qualitatively in the same way humans can. In the case of well, assuming the output from a computer search to be correct without manual qualitative division of its two functions (degree and manner), would have given inaccurate and skewed frequency data.

Another methodological issue which appeared in the present study was the question of quantification. Despite the fact that the variationist paradigm has become the norm in synchronic variationist analyses, many historical and corpus-based studies continue to use frequency normalization as the baseline for comparison. As the data in Figures 1 and 2 indicate, the frequency of well differs depending on the approach adopted. As useful as frequency normalization can be over absolute frequency measurements, it can skew the data. Frequency either increases or decreases based on the size of the corpora (e.g., in the OBC) since the number of words are used as the baseline for comparison, as opposed to the number of accountable contexts in which a variant could or could not have occurred. Just because a corpus contains more words does not necessarily mean there are more intensifiable contexts. Therefore, in line with previous recommendations (Tagliamonte 2012: 12), I advocate the use of the variationist approach when measuring frequency in historical studies, which contributes to recent discussions of this nature (Biber et al. 2016).

6. Conclusion

In analyzing the diachronic frequency of well, the present study discussed some of the broader methodological challenges of historical linguistics. That a lack of attestation does not necessarily equate to a lack of existence was clearly the case with well. Although five widely used corpora were selected, these did not capture its retained dialectal use. This thus illustrates the importance of using “all the data” in a historical analysis (Lauersdorf 2018a, 2018b). Following previous research (Biber et al. 2016), the present study also compared frequency normalization with variationist quantitative methods and found differences in distribution. Even though the same data were analyzed, the measurements sometimes created different trajectories, with frequency normalization distorting the picture somewhat.

As for the more local question of well, despite its decrease in frequency in the mid-14^th century, its use as an intensifier of regular adjectives did not completely disappear from all dialects of British English. Nevertheless, although it was retained, one cannot make definitive claims regarding its frequency in dialectal sources. One might hypothesize however, based on its absence in the speech-related corpora, that this use of well was not very frequent. Notwithstanding, regardless of whether it was retained at a high or low frequency, “once a word has evolved to have an intensifier function it remains in the reservoir of forms that a language user may deploy to boost meaning from that point onwards regardless of whether it actually becomes one of the favoured forms or not” (Tagliamonte 2008: 391). While it is possible that its present-day use is a product of the dialectal retention spreading outward, this type of diffusion is not particularly common. The alternative explanation paints a picture of innovation where the mainstream distributionally-restricted use developed new functions in the 20^th and 21^st century. However, one wonders whether multiple causation is at play, since both spreading and innovation could have served as contributing factors in its present-day development.

Irrespective of how well emerged, it is clear that its use changes in frequency. If its frequency from Middle English were plugged into the data from the present study, one would potentially observe a diachronic U-shape distribution. However, admittedly this would not account for its retention in dialectal uses. In addition to a change in frequency, the distribution of its intensified heads indicates that intensification of both open and closed scale adjectives (both deverbal or non-deverbal) becomes permissible. Given that the stress pattern lines up with the scale structure differences, it is possible that these are recent developments to the mainstream use of well as opposed to a continuation of the retained one. In the absence of phonetic recordings, it is unclear whether the current differences in stress were always a property of well. Even though the acknowledged methodological challenges in historical linguistics leave some questions unanswered, the findings from the present study contribute to the growing body of literature on intensifiers, and more specifically lend support to the notion that intensifiers can rise and fall at repeated intervals throughout time.