The Adverb of Degree in Old and Early Middle Dutch Lourens Jacob Visser

The Adverb of Degree in Old and Early Middle Dutch Lourens Jacob Visser

S2206218

l.j.visser.2@student.rug.nl Breezand 19 8531 PN Lemmer University of Groningen

Faculty of Arts

MA Thesis Language and Cognition Supervisor: Prof. Dr Jack Hoeksema

June 1, 2019

Word count: 19,042

Abstract

Adverbs of degree serve to strengthen (high degree), weaken (minimal degree), or reinforce a proposition (absolute degree). Various syntactic and semantic restriction apply for when an individual adverb can be used, typically depending on the extent of grammaticalisation of the word (Klein, 1998). During the grammaticalisation process, adverbs first tend to expand their context and then become more specialised (Bolinger, 1972; Klein, 1998). Even though Modern West Germanic languages have proper documentation, material for the older stages is more limited. The present study sought to describe how the different adverbs were used in the earliest stages of Dutch and which factors predicted the choice of adverb. It was expected that adverbs that still exist would show fewer restrictions, and that those that no longer do would be more specialised. Based on data from two corpora, this turned out mostly true, though the adverb “zeer” specialised in register, and “utermaten” did not show specialisation over time.

It was also expected that the adverbs would be distinguishable, as Humboldt’s Principle holds that a one-to-one relationship between form and meaning is preferred (Koefoed, 1978). The adverbs of high degree in Middle Dutch showed different tendencies in usage, but those of minimal degree could not be distinguished as well. These results provide evidence for the grammaticalisation patterns described above, but they do not show that Humboldt’s Principle held true here.

Keywords: Adverb of degree, intensification, Dutch, historical linguistics, corpus

linguistics.

Table of contents

1. Introduction ... 5

2. Background ... 6

2.1. The adverb of degree ... 6

2.1.1. General description. ... 6

2.1.2. Syntactic conditions. ... 11

2.1.3. Semantic conditions. ... 13

2.1.4. Sociolinguistic conditions. ... 16

2.2. Old and Middle Dutch ... 17

2.2.1. Old Dutch... 17

2.2.2. Middle Dutch. ... 19

2.2.3. Available sociolinguistic factors. ... 19

2.2.4. Adverbs of degree in Old and Middle Dutch... 20

2.3. Adverbs of degree in other Old Germanic languages ... 21

2.3.1. Swīþe and ful. ... 22

2.3.2. Sāre. ... 24

2.3.3. Fela. ... 25

2.4. Research question and hypothesis ... 26

3. Method ... 27

3.1. Materials ... 27

3.2. Procedure ... 27

3.3. Analysis ... 29

4. Results... 30

4.1. Adverbs of high degree ... 30

4.1.1. Hard. ... 30

4.1.2. Zeer. ... 34

4.1.3. Veel. ... 37

4.1.4. Swītho. ... 41

4.1.5. Comparison. ... 42

4.2. Adverbs of extremely high degree... 44

4.2.1. Mahtigan. ... 44

4.2.2. Grotelijk. ... 44

4.2.3. Utermaten. ... 45

4.3. Adverbs of minimal and quasinegative degree ... 47

4.3.1. Bore. ... 47

4.3.2. Een deel. ... 49

4.3.3. Iewet. ... 51

4.3.4. Weinig. ... 53

4.3.5. Comparison. ... 53

5. Discussion ... 55

5.1. The usage of the individual adverbs of degree... 55

5.1.1. Adverbs of high degree. ... 55

5.1.2. Adverbs of extremely high degree. ... 57

5.1.3. Adverbs of minimal and quasinegative degree. ... 58

5.2. Distinguishing factors ... 60

5.2.1. Adverbs of high degree. ... 60

5.2.2. Adverbs of minimal degree. ... 60

6. Conclusion ... 61

6.1. Summary ... 61

6.2. Limitations ... 62

6.3. Future research ... 63

7. References ... 65

1. Introduction

Adverbs of degree are among the most frequent words in the lexicon of most languages (Hoeksema, 2011), and they can have a variety of functions. For example, they can be used to strengthen a particular adjective, to weaken it, or to affirm it, all to varying degrees.

For example, a person can be very tall, extremely tall, perhaps a bit tall, or not at all tall. These examples demonstrate the existence of a scale, and different adverbs of degree can push it up or down. Strictly speaking, any adverb that modifies an adjective has a degree modifying function (Bolinger, 1972), though adverbs of degree are by no means limited to modifying adjectives alone. Over the years, various studies have been performed to document the usage of the different adverbs of degree in the Modern Germanic languages. For examples, Klein (1998) described them for Dutch, van Os (1988) documented them for German, and Bolinger (1972) did so for English. In each case, various restrictions apply about which adverbs can be used with which phrases, while others can only be used with particular kinds of adjectives. By contrast, material describing them for earlier stages of these languages is less extensive, with English being the main exception, as various studies exist documenting adverbs of degree in various earlier stages of the language (i.e. Ito & Tagliamonte, 2003; Méndez-Naya, 2003;

Peters, 1994). Typically, adverbs of degree are highly prone to change, as new adverbs of degree frequently emerge through grammaticalisation of lexical words (Klein, 1998; Peters, 1994), the usage of existing ones also tends to change over time (Bolinger, 1972; Klein, 1998), and their intensity tends to change over time as well (Bolinger, 1972; Klein, 1998; Lorenz, 2002; Méndez-Naya, 2003). However, these tendencies may be a relatively recent phenomenon, and it may not have held true in medieval times (Hoeksema, 2011). All of this suggests that it is highly relevant to study the adverb of degree from a historic perspective, because they are highly frequent and highly changeable. Currently, no extensive study exists that investigates the usage of these adverbs for the earliest stages of Dutch, aside from what has been documented by the historical dictionaries ONW (2015) and VMNW (2015). The present study will therefore seek to describe how the different adverbs of degree were used in the Old and Early Middle Dutch periods.

This study is structured in the following way: section 2 will provide a theoretical background to adverbs of degree in general, as well as an overview of Old and Middle Dutch.

It will first provide a classification system for the adverbs of degree and discuss

grammaticalisation patterns that affect them. It will then describe language internal and external factors that influence the choice of adverb. After that, a brief description of the Old and Middle Dutch languages will be given, as well as a description of the available sociolinguistic factors there, and a description of the adverbs of degree in those languages contrasted with their Modern Dutch equivalents. Then it will discuss a selection adverbs of degree in related Old Germanic languages. Finally, it will outline the aim of the present study by presenting two specific research questions and their respective hypotheses. Section 3 will then provide a description of the used corpora and describe the used research method. Then section 4 will describe the adverbs of degree in detail, and the adverbs of the same degree will be compared. Section 5 will seek to interpret the results based on the background literature, and section 6 will conclude, discuss the limitations, and provide suggestions for future research.

2. Background 2.1. The adverb of degree

2.1.1. General description.

Adverbs of degree come in many different varieties and with various different

functions, and, as such, it is necessary to discuss a classification system. Various different

systems have been proposed over the years, but the most elaborate one is proposed by Klein

(1998), which itself is an adaptation of the system proposed by van Os (1988), and is shown in

Table 1:

Table 1.

A table showing the different degrees, adapted from Klein (1998, p. 20).

Class Degree Examples

I Absolute completely, absolutely

II Approximate almost, nearly

III Extremely high extremely, awfully

IV High very

V Moderate rather, pretty

VI Minimal somewhat, a bit

VII Quasinegative little, hardly

VIII Negative not, not a bit

In this system, adverbs of absolute degree affirm the modified phrase, while those of approximate degree indicate that the scale of the modified word is close to its endpoint (Klein, 1998). Adverbs of high degree shift the scale upwards, while those of extremely high degree are similar, but they do this to a greater extent and tend to have a stronger emotive value (Klein, 1998). Adverbs of moderate degree also shift the scale upwards but to an even lesser extent (Klein, 1998). On the other hand, adverbs of minimal degree are used to indicate that a particular quality is present, but the scale is shifted downwards, while those of quasinegative degree also shift the scale downwards, but the quality is present to a lesser degree than expected (Klein, 1998). The difference between the latter two can therefore be considered fuzzy. Finally, adverbs of negative degree indicate the complete absence of a particular quality (Klein, 1998). An alternate classification is used by Bolinger (1972) is his classic study, which uses four categories:

• Boosters

• Compromisers

• Diminishers

• Minimisers

In this system, boosters correspond to adverbs of high and extremely high degree,

compromisers to adverbs of moderate degree, diminishers to the quasinegative degree, and

minimisers to adverbs of minimal degree. It is also possible to distinguish these adverbs by

their lexical value or lack thereof, as certain ones have acquired a purely functional usage due to grammaticalisation (i.e. Bolinger, 1972), which is why it is necessary to discuss the common grammaticalisation patterns that affect them.

Adverbs of degree are generally derived from lexical words through grammaticalisation. In general, grammaticalisation describes the change from a lexical word to a grammatical one, which happens through a process of reanalysis (Millar, 2013). Lexical words tend to lose their original morphological and syntactic properties and adopt those of a

“lower” or more functional category by decategorisation, which tends to happen especially to high frequency words (Hopper & Traugott, 2003). Words that are prone to undergo grammaticalisation and become adverbs of degree are shown in the following list based on Peters (1994, p. 269) and Klein (1998):

• Local/dimensional adverbs

• Quantitative adverbs

• Qualitative adverbs

• Emphasisers/Modal adverbs

• Taboo words

When lexical words grammaticalise to adverbs of degree, the original lexical meaning will typically still be present initially, though this meaning will gradually undergo semantic bleaching, leaving only a purely functional usage (Klein, 1998). This grammaticalisation process is how adverbs of degree typically originate (Klein, 1998). However, it should be noted that both the original meaning and the adverb of degree can coexist (Klein, 1998). As an example, Klein (1998) uses the Dutch word “heel,” which can be used as an adjective meaning

“whole” and as a functional adverb of high degree. It should be noted that not every adverb of degree has completely lost its lexical meaning. In certain cases, the old lexical meaning can even new supplant the functional meaning, which happened with the English word “pure,”

which was used as an adverb of high degree in the past, while today it is a purely lexical word (Bolinger, 1972).

It is also commonly argued that adverbs of high degree tend to lose their emotive power and intensity as they become more frequent (Bolinger, 1972; Klein, 1998; Lorenz, 2002;

Méndez-Naya, 2003), though there may be more to this observation than just frequency. Over time, adverbs of degree tend to undergo a grammaticalisation process known as “renewal”

(Hopper & Traugott, 2003, p. 122), which effectively means that, as a particular adverb

becomes more frequent, it will lose its intensity over time, and another word will take its place as a high-intensity adverb (Klein, 1998; Lorenz, 2002; Méndez-Naya, 2003; Hopper & Traugott, 2003). Hopper and Traugott (2003) argue that it is precisely because of their strong emotive value, and that adverbs of high degree are therefore particularly prone to undergo renewal.

However, ten Buuren, van de Groep, Collin, Klatter, and de Hoop (2018) found that a high perceived frequency of an adverb of (extremely) high degree actually resulted in a higher intensity score among 324 Dutch secondary school students, which contradicts most of the established literature. They also found that a higher perceived modernity were also rated higher in intensity. Adverbs of extremely high degree that are highly frequent and modern were rated the highest in intensity (ten Buuren et al., 2018). This connection between intensity and novelty is also noted by Lorenz (2002). It is therefore likely that it is not a high frequency that causes an adverb of degree to lose its intensity, but perceived old-fashionedness (ten Buuren et al., 2018). Ten Buuren et al. also argue that words associated with a particular generation tend to be perceived as more old-fashioned.

Not only the perceived intensity of an adverb of degree is prone to change, but the context in which it can be used tends to change as well (Klein, 1998). As is common in grammaticalisation patterns, the original context in which a word or phrase a word was restricted, but as it undergoes reanalysis its context is expanded (Hopper & Traugott, 2003), and this very much applies to adverbs of degree as well (Klein, 1998). However, a seemingly opposite process also applies: as adverbs of degree acquire a more specialised meaning, its context can become increasingly more restricted (Klein, 1998), or, as Bolinger (1972) poetically describes it, “[t]he old favorites do not vanish but retreat to islands bounded by restrictions”

(p. 18). For an overview of potential restrictions on context, see sections 2.1.2 and 2.1.3.

While it is true that adverbs of high degree are especially prone to undergo change and

renewal in Modern times, this may not have always been the case. For example, an interesting

observation is made by Hoeksema (2011), as he states that the number of adverbs of degree

in Dutch has dramatically increased since the Early Modern period, while the number in the

Middle Dutch period was fairly stable (see section 2.2.4 for a description of the Middle Dutch

situation). Similar observations have been made for English (Stoffel, 1901) and German (van

Os, 1988), which would suggest that the tendency for change is something which began only

began comparatively recently in these languages. This increase is remarkable, as it appears to

violate the principle that languages have a tendency towards one-on-one relationships

between form and meaning (Hoeksema, 2005). This concept is known as Humboldt’s Principle (Koefoed, 1978). While the adverbs of high degree may contradict this notion in the Modern West Germanic languages, whether or not this also held true for earlier stages of these languages is unclear. Having fewer distinct adverbs of degree could indicate that they were differentiated more clearly in their usage. In this context, the tendency for existing adverbs of degree to specialise in function could also be seen as a product of Humboldt’s Principle. The reason for the increase in the number of adverbs of degree remains unclear, though Hoeksema speculates that this may be due to an increase in literacy beginning in the Early Modern period. Regardless of how one accounts for this observation, the fact that a fairly stable system has existed for hundreds of years would make an explanation based on emotion seem unlikely, unless it is assumed that people began using more expressive language in the Early Modern period compared to medieval times.

Not everything has been said about the terminology, however. In various literature on the subject, the term “intensifier” is used instead of “adverb of degree.” For example, Bolinger (1972) uses “the term intensifier for any device that scales a quality, whether up or down or somewhere between the two” (p. 17). On the other hand, Méndez-Naya (2003) uses the term

“intensifier” in a more narrow sense, as she defines it “as an adjective and adverb modifier which scales upwards from the assumed norm” (p. 374) citing words like “very” and

“extremely” as examples, while Bolinger uses the term “booster” (p. 17) for this function, as

mentioned above. These terms are roughly equivalent to Klein’s (1998) adverb of high degree

and those of extremely high degree. Méndez-Naya specifically argues that intensifiers

constitute their own lexical category and that they cannot truly be considered adverbs on the

basis that the category includes both function words like “very,” which hold little lexical

meaning, and lexical words like “absolutely,” which does. This is something that is also noted

by Bolinger and Lorenz (2002). “Very” is arguably the most grammaticalised adverb of degree

in English, as it has completely lost its original meaning of “truly” or “truthfully,” which

indicates a complete delexicalization (Lorenz, 2002). Bolinger refers to these two different

types as “grammaticized and ungrammaticized intensifiers” (p. 22) respectively, though he

acknowledges that ungrammaticized intensifiers still constitute a partially open class, while

Lorenz considers them a closed class. Lorenz’s claim seems difficult to defend, however,

because adverbs of degree have a high propensity for renewal, as laid out above, which means

that they cannot truly be considered a closed class, as an existing lexical degree word could

grammaticalise further, lose its expressivity, and become a function word like “very.” This is something that Lorenz himself notes, as he predicts that “really” is likely to follow in the footsteps of “very” based on corpus data. Because of this distinction between functional adverbs of degree and lexical ones, Méndez-Naya describes intensifiers as occupying an

“intermediate position between grammar and lexis” (p. 374). However, Klein does not support this distinction and considers all adverbs of degree to be part of a functional category, though he does consider the category of adverbs of degree to be an open class, unlike other function words. Méndez-Naya also specifically makes a distinction between intensifiers as adjective and adverb modifiers and adverbs of degree as modifiers of lexical verbs. For clarity, this paper will continue to use the term “adverbs of degree” as the general name for these words, as Klein’s degrees, which are shown in Table 1, remains the most detailed way of classifying them.

However, it is notable that different adverbs can show different levels of grammaticalisation, which also shows that the distinction between lexical words and function words is not always clear-cut.

The distribution of adverbs of degree is governed by a variety of rules. These consist of both internal, which consist of both syntactic and semantic conditions, and external factors, which are predominantly sociolinguistic. All three will be discussed in sections 2.1.2-2.1.4.

2.1.2. Syntactic conditions.

Adverbs of degree most commonly modify adjectives and adverbs, though they are capable of modifying words of other categories as well (Bolinger, 1972; Lorenz, 2002).

Commonly, adverbs of degree are capable of modifying adjectives, nouns, verbs, determiner phrases (DPs), and prepositional phrases (PPs) (Klein, 1998). In certain instances, the distribution of adverbs of degree is governed by simple syntactic rules, and this is particularly true for more functional ones. As would be expected based on the grammaticalisation patterns outlined in section 2.1.1, the more functional adverbs of degree can be used to modify more syntactic categories than lexical ones, though some are still bound by syntactic restrictions (Klein, 1998). For example, the Dutch words “heel” and “erg” hold little lexical value when used as adverbs, and they are simply used to indicate a high degree (Klein, 1998).

However, “heel” almost exclusively modifies adjectives and adverbs, whereas other adverbs of high degree like “erg,” “heel erg,” and “zeer” can modify verbs as well (Hoeksema, 2005;

Klein, 1998). The one exception for “heel” is that it is also capable of modifying certain PPs

(Hoeksema & Korterink, 2011), though this usage is restricted to highly specific contexts.

Examples (1a) and (1b) illustrate the difference between “erg” and “heel” when it comes to modifying verbs, while (1c) and (1d) highlight the difference for PPs:

(1) a.

b.

c.

d.

This distribution is similar, though not identical, to English “very” and “much” (Klein, 1998).

Adjective degree can also be a relevant factor, as the Dutch adverb of high degree “veel” can exclusively modify comparatives, while “erg” and “heel” cannot be used in this context (Hoeksema, 2011), as is shown in the following example:

(2)

However, “veel” has another usage, as it can also be used in constructions with “te” (“too”) (Hoeksema, 2011), and “erg” and “heel” cannot be used in these constructions, as can be seen in example (3):

(3)

“Veel” cannot be used in other contexts. By contrast, adverbs of minimal degree like “een beetje” can modify both the positive and the comparative without any issues in Dutch (Hoeksema, 2011). All of these examples show that grammaticalised adverbs of degree can still be sensitive to syntactic context.

When it comes to less grammaticalised adverbs of degree, some additional syntactic restrictions are present when modifying verbs in English. In these instances, the lexical meaning of the adverb appears to play a role (Bolinger, 1972). According to Bolinger (1972),

Ik waardeer het erg

I appreciate it very much (Klein, 1998, p. 13)

*Ik waardeer het heel

*I appreciate it very (Klein, 1998, p. 13) Heel in de verte zie ik een bekende very much in the distance see I a familiar person

I see a familiar person very much in the distance (Hoeksema & Korterink, 2011, p. 30) Jan is erg/*heel in de war

John is very/*much confused (Hoeksema & Korterink, 2011, p. 30)

Het is veel/*heel/*erg beter It is much/*very/*very much better

Het is veel/*heel/*erg te groot

It is much/*very/*very much too large

the less literal the perceived meaning of the adverb is, the more likely it is to be accepted as a verb modifier. This holds particularly true for adverbs with a hyperbolic meaning like

“dazzlingly” or “whoppingly,” both adverbs of extremely high degree, as these are generally acceptable as verb modifiers, and they are used chiefly metaphorically (Bolinger, 1972). For other adverbs, it depends on the context. For an adverb like “badly” its acceptability varies based on usage (Bolinger, 1972; Greenbaum, 1970). Compare the following example sentences from Greenbaum (1970, p. 64):

(4) a. He badly needed the money b. They badly wounded the elephant c. They badly treated the servant

In these cases, example (4a) was unanimously accepted, (4b) was accepted by 77% of the participants, and (4c) by only 34% (Greenbaum, 1970). Bolinger explains this by arguing that the use of the word “badly” is increasingly more literal. In this case, (4a) only allows for a non- literal reading, both readings are available for (4b), and (4c) only allows for a literal reading.

Therefore, the acceptability of a lexical adverb of degree in these constructions should not be dependent on the adverb itself but on its meaning. Unfortunately, Bolinger provides no typology of verbs that indicates what kind of verbs can combine with which adverbs, which leaves it somewhat ambiguous when a certain interpretation is available, and it will mainly come down to the judgement of individual speakers. In some cases, certain lexical adverbs of degree can combine with verbs in fixed expressions like “hopelessly lost,” “devoutly believe,”

and “desperately need” (Bolinger, 1972, p. 245). It should be noted that all three sentences in example (4) were deemed acceptable when they the modified verb was changed to a past participle (Greenbaum, 1970). Regardless, all of this demonstrates that the more lexical meaning an adverb of degree has, the more syntactic restrictions apply to its usage.

2.1.3. Semantic conditions.

It has long been established that syntactic conditions alone cannot fully capture the

distribution of adverbs of degree in English at least (Bolinger, 1972; Kennedy & McNally, 2005),

and therefore a description of the various semantic conditions will be necessary. One such

condition is the scale type of adjectives (Kennedy & McNally, 2005). Kennedy and McNally

(2005) argue for a distinction between open scales and closed scales. Closed scales have a

distinct maximum, a minimum, or both, while open scales have neither (Kennedy & McNally,

2005). In this system, closed scales are further separated into three types: lower closed scales with a minimal element, upper closed scales with a maximal element, and totally closed scales with both (Kennedy & McNally, 2005). These can be distinguished by modifying adjectives with adverbs of absolute degree like “completely” or “fully,” as open scales cannot be modified by this type of adverb (Kennedy & McNally, 2005; Klein, 1998). For lower closed scales, they can only modify their antonyms, upper closed scales can be modified by these words but their antonyms cannot, and for totally closed scales both variants can be modified (Kennedy &

McNally, 2005). Not only adverbs of absolute are sensitive to adjective scales, but the more grammaticalised adverbs of high degree are as well (Hoeksema, 2011; Kennedy & McNally, 2005). For Dutch, the functional adverbs of high degree “heel,” “erg,” and “zeer” are generally preferred for adjectives with an open scale (Hoeksema, 2011). This can be seen in the following examples:

(5) a.

(open scale) b.

(closed scale)

Another factor relevant to the distribution of adverbs of degree is the standard value of adjectives, which refers to the distinction between relative and absolute adjectives (Kennedy

& McNally, 2005). Relative adjectives have a meaning relative to the noun that is modified (i.e.

a large spider is different from a large elephant), while for absolute adjectives the meaning is fixed (Kennedy & McNally, 2005). In English, the adverb of degree “very” is only capable of modifying relative adjectives, and it does not work with absolute ones (Kennedy & McNally, 2005). This is illustrated in the following examples:

(6) a. The spider was very large.

b. ??The spider was very alive.

Note that this is not completely fixed, as certain absolute adjectives can be relative depending on its context (Kennedy & McNally, 2005). For example, the spider in example (6b) could be more alive today than yesterday. These are the ways in which the type of adjective can influence the choice of the adverb of degree.

De deur is heel/erg/zeer groot

The door is very large

??De deur is heel/erg/zeer open

??The door is very open

A number of adverbs of degree are sensitive to polarity, which includes both inherent polarity and the polarity of the environment (Klein, 1998). Being sensitive to inherent polarity effectively means that certain adverbs of degree can only modify words that are either positive or negative (Klein, 1998). This distinction between positive and negative adjectives includes not only word pairs like “happy - unhappy” (Klein, 1998, p. 71), but also scalable adjective pairs like “short - long” (Klein, 1998, p. 71), for which the one closest to zero is considered the negative variant (Klein, 1998). This is because, in both cases, the positive variant is semantically unmarked, while the negative one is marked (Klein, 1998). In other cases, the negative variant indicates the absence of a property, while the positive one indicates its presence (Klein, 1998). This is known as “natural polarity” (Klein, 1998, p. 72).

Another type is “evaluative polarity” (Klein, 1998, p. 72), which often takes precedence, and it considers word of praise to be positive and words of criticism to be negative. Therefore, a word like “clean” would be considered positive, despite indicating an absence of dirt (Klein, 1998). An example of an adverb that is sensitive to inherent polarity is the Dutch adverb of moderate degree “aardig,” which can only be used with positive words, as can be seen in the following examples from Klein (1998, p. 77):

(7) a.

b.

Klein (1998) hypothesises that the sensitivity of adverbs of degree to inherent polarity is caused by the original lexical meaning of the adverb.

Not only are certain adverbs of degree sensitive to the inherent polarity of the negative words, some are also sensitive to the polarity of the environment. This means that certain ones can only be used in a negative-polarity context, while others can only be used in a positive-polarity context (Klein, 1998). The former ones are known as negative polarity items, while the latter ones are referred to as positive polarity items. Much can be said about the exact definition of negative polarity (see Klein, 1998), but it most commonly describes a phrase within the scope of negation. Other environments that permit negative polarity items are questions, comparative clauses, conditional clauses, and clauses introduced by “before,”

though this list is by no means exhaustive (Klein, 1998). As an example, adverbs of minimal

?Hij is aardig traag

he is pretty slow

Hij is aardig actief

he is pretty active

degree are typically limited to positive environments (Klein, 1998), as can be seen in the following example:

(8) a. John is somewhat happy.

b. *John is not somewhat happy.

Negative and positive polarity items are thought to be a result of a specific grammaticalisation process, in which expressions with a preference for negative or positive constructions become restricted to that specific context (Hoeksema & Korterink, 2011). This is supported by the fact that certain adverbs of degree show a preference for either context but not always a hard restriction (Hoeksema & Korterink, 2011). For example, the Dutch adverb “echt” has a clear preference for negative environments, though it can still occasionally be used in positive ones (Hoeksema & Korterink, 2011). When negated, “echt” typically has a downtoning meaning, but it can also have an understating meaning when combined with evaluative adjectives (Klein, 1998). This also further supports the idea that adverbs of degree are on different levels of grammaticalisation.

2.1.4. Sociolinguistic conditions.

Outside of the aforementioned language internal factors, the distribution of adverbs of degree is also affected by external sociolinguistic factors. One of these is gender. In the field of sociolinguistics, two main principles exist with regard to language and gender: in a stable situation, women tend to use more forms associated with the standard language, and women also tend to use more novel forms than men (Labov, 1990). This second principle tends to apply to the usage of adverbs of degree as well (Tagliamonte & Roberts, 2005). By examining data from the television series Friends (1994-2002), Tagliamonte and Roberts (2005) found that the female characters tend to use the innovative adverb of high degree “so” more than the male characters, though the gender difference decreased throughout the different seasons, which would be expected once the incoming form becomes more established. The rate of intensification in the Friends data matched that of data based on real language use, so they argue that it should be reflective of developments in natural language. The use of a high number of intensifiers in general has often been associated with women’s language, though this has never been quantitatively demonstrated (Tagliamonte & Roberts, 2005).

Age effects also play a role in the distribution of the adverb of degree. For example,

people over the age of 35 in the United States of America showed a strong preference for

using “very” over “really” as the main adverb of high degree, while, for people between the ages of 17 and 34, this was the other way around (Ito & Tagliamonte, 2003). “Really” is generally perceived to be the more innovative form, even though it has been used as an adverb of high degree since at least the eighteenth century (Ito & Tagliamonte, 2003). It is also relevant to note that the choice of adverb of degree can serve as a strong in-group marker for specific generations (ten Buuren et al., 2018; Lorenz, 2002; Peters, 1994). This is especially apparent in the language of teenagers, as every generation adopts a new way of speaking, and this is clearly visible in the adverbs of degree (Lorenz, 2002). Another relevant factor is education, as Hoeksema and Korterink (2011) found that Dutch university students use more forms associated with the formal register, while VMBO students tended not to use such forms.

It is important to note that there is both an education and an age effect here, as the VMBO students were approximately 13 years old, while the university students were around 21 years old (Hoeksema & Korterink, 2011), and it is not yet clear how these two variables interact.

2.2. Old and Middle Dutch 2.2.1. Old Dutch.

The oldest stage of the Dutch language is referred to as Old Dutch or Old Low

Franconian, and it is considered part of the Continental West Germanic languages (van der

Wal & Quak, 1994). It encompasses all texts from before 1170 (Gysseling, 1970). The language

is distinguished from Old High German based on the lack of the Old High German Sound Shift,

and it differs from Old Saxon based on other phonemic and morphological criteria (van der

Wal & Quak, 1994). One of the most salient distinctions is that Old Saxon shows deletion of

nasals before fricatives, like Anglo-Frisian, while Old Dutch generally preserves the nasals in

these positions (Robinson, 1992). The coastal varieties of Old Dutch are also characterised by

features from North Sea Germanic (also referred to as “Ingvaeonisms”), which are usually

explained by language contact with North Sea Germanic or a language shift to Old Dutch in

these regions (see de Vaan, 2017). There are three longer texts that could be described as Old

Dutch: The Wachtendonck Psalter (10th century), the Leiden Willeram (ca. 1100), and the

Middle Franconian Rhyming Bible (1151–1200) (de Vaan, 2017). The remaining attestations of

the language are fragmentary, and they predominantly consist of short phrases, glosses in

Latin texts, toponyms, and proper names (de Vaan, 2017). Arguably, the most famous Old

Dutch phrase is the Rochester Poem (1083–1110) (de Vaan, 2017), also known as Hebban olla

Vogala.

The Wachtendonck Psalter contained interlinear glosses of the Book of Psalms based on the Latin Vulgate (Robinson, 1992; de Vaan, 2017). Because of this, its syntax is entirely based on the Latin exemplar. The original manuscript is unfortunately lost, and our knowledge of it primarily comes from two sixteenth and seventeenth-century letters and three partial copies (Robinson, 1992). The first letter is a 1602 printing of a 1598 letter by Justus Lipsius identifying the manuscript, and it contains a word list of 670 words along with their Latin translations (Robinson, 1992). The second letter contains a longer word list with 822 words along with references from which psalm the word in question was taken (Robinson, 1992), including those from psalms that did not survive. The three partial copies are of Psalm 18 produced in 1612, a manuscript containing Psalms 53:7-73:9, and a manuscript copy of Psalms 1:1-3:5 (Robinson, 1992). Additionally, a fragment containing seven words from Psalm 55 also exists (Quak, 2010). The language is by no means uniform, as the early psalms especially contain High German forms, and it is not completely clear how these should be interpreted (see Robinson, 1992). It is most commonly assumed that the text was copied from a Middle Franconian exemplar into an eastern dialect of Old Dutch (de Vaan, 2017). More specifically, Quak (2010) argues that the original manuscript was likely produced in Munsterbilzen (present-day Belgian Limburg) based on a letter by Lipsius. Because of this, the dialect could be considered an early variety of Limburgian, though it lacks certain archaisms found in Limburgian Middle Dutch such as a distinction between the accusative and the dative case in first and second person singular pronouns (Kerkhof, 2017). This pronoun system is therefore arguably closer to a Western variety of Middle Dutch. The text also shows occasional deletion of nasals before fricatives (de Vaan, 2017), which is a feature associated with the coastal dialects of Dutch, as this is considered to be an Ingvaeonism.

The second major Old Dutch text is the Leiden Willeram. This text is an Old Dutch copy of an Old High German commentary on the Song of Songs likely produced in the Egmond monastery (de Vaan, 2017). The language is heavily mixed with Old High German, though the scribe introduced Old Dutch features (de Vaan, 2017). In particular, the stop system is largely taken over from Old High German (Kerkhof, 2017). De Vaan (2017) states that it is now commonly assumed that the scribe spoke a Hollandic dialect of Dutch based on the presence of Ingvaeonisms in the text, though the scribe may have also been Flemish (Kerkhof, 2017).

Regardless, these features point to a coastal dialect of Dutch. The third text is the Middle

Franconian Rhyming Bible, which was likely produced in Westphalia, though de Vaan (2017)

argues that its language could not have been far removed from what was spoken in the Low Countries at that time, though no justification is provided for this statement.

2.2.2. Middle Dutch.

The Middle Dutch period is usually set to begin around 1200, and lasts until approximately 1550 (de Vaan, 2017). It can also be useful to distinguish between Early Middle Dutch (1200-1300) and Late Middle Dutch (1300-1550), based on the existence of the Corpus Gysseling (2013) which covers the texts from the Early period (see section 3.1 for details). The present study will concern itself with the Early Middle Dutch period. Unlike the Old Dutch period, there is no shortage of texts, as the thirteenth century marks the beginning of vernacular literature (de Vaan, 2017). The Middle Dutch period is also characterised by dialectal variation. Traditionally, the following dialects are distinguished based on Bloemhoff and Streekstra (2015):

• Flemish

• Hollandic

• Brabantian

• Limburgian

• Rhinelandic

These dialects are claimed not to show a great degree of lexical variation, and the differences between them are said to be chiefly phonological and morphological (Bloemhoff & Streekstra, 2015). Flemish and Hollandic are Western dialects, and they continue the coastal varieties of Old Dutch (Bloemhoff & Streekstra, 2015). These dialects may therefore show influence from North Sea Germanic (Bloemhoff & Streekstra, 2015), which may indicate a greater degree of similarity to Frisian and English. The other dialects continue mainland varieties of Old Dutch.

Limburgian, an Eastern dialect, shows a partial High German consonant shift (Bloemhoff &

Streekstra, 2015), which means that this dialect falls between the other varieties of Middle Dutch and Middle High German.

2.2.3. Available sociolinguistic factors.

While it is true that the choice of adverbs of degree is affected by a variety

sociolinguistic factors, as mentioned in section 2.1.4, not all of these are known to us for Old

and Middle Dutch. One major issue is that texts by women are lacking in the early period,

which is unfortunate as gender plays a role in the choice of adverb (Tagliamonte & Roberts,

2005). The only named writers in the Early Middle Dutch period, as present in the Corpus

Gysseling (2013), are Jacob van Maerlant, Willem van Affligem, Heinric van Veldeke, and Broeder Geraert, though most texts from this period are anonymous. It is also impossible to examine differences in education level, because the medieval lower classes were typically illiterate, meaning that the available texts are from higher educated people. The main available sociolinguistic factor that is available is dialect, which will be the main focus of the present study when it comes to sociolinguistic factors.

2.2.4. Adverbs of degree in Old and Middle Dutch.

According to Hoeksema (2011), the number of adverbs of degree in Early Middle Dutch was limited. The main functional adverbs of high degree were “hard” and “zeer,” while “een deel” functioned as an adverb of minimal degree, and “bore” existed as a negative polarity item (Hoeksema, 2011). However, VMNW (2015) notes that “bore” was also sometimes used in a context without negation, and, in these cases, it effectively functioned as an adverb of quasinegative degree (VMNW, 2015). The Old Dutch equivalents of “hard” and “zeer” were

“hardo” and “sēro,” though the latter is not attested as an adverb of degree (ONW, 2015).

When “hard” modified “naar” in Middle Dutch, it functioned more like an adverb of approximate degree (VMNW, 2015). Aside from these adverbs, VMNW lists “veel” as an adverb of high degree, the former of which often modified the comparative, or it modified the positive which would give it a comparative meaning. The Old Dutch counterparts to this adverb was “filo,” and Old Dutch also had “swītho” as an adverb of high degree (ONW, 2015).

“Mahtigan” was used as an adverb of extremely high degree in Old Dutch (ONW, 2015), while Middle Dutch had “utermaten” and “grotelijk” for this function, and these became more prominent in the later Middle Ages (Hoeksema, 2011). For minimal degree, Middle Dutch also had “iewet” and “weinig” (VNMW, 2015), and the Old Dutch equivalent to the former was

“iewiht,” while the latter was not yet used (ONW, 2015) Outside of what has been listed here, the usage of these adverbs has not yet received a proper documentation, which is what this paper will seek to accomplish.

For comparison, the usage of the same adverbs of degree listed above in Modern Dutch

will be discussed. First, the Middle Dutch adverbs “bore” and “een deel,” no longer in use and

neither is the Old Dutch adverb “swītho” (though see section 2.3.1). “Iewet” is no longer used

in this form, though it still serves as an adverb of minimal degree in the form “iets,” which

derived from “iewet” + “wes,” the genitive form of “wat” [what], and it is restricted to

comparatives (WNT, 2015). Another formation that is still used is “ietwat,” from “iewet” +

“wat,” which can be used with both the positive and the comparative (WNT, 2015). Klein (1998) states that “hard” is no longer used in Modern Dutch, though this is not completely true, as it can still be used as an adverb of degree in restricted contexts like “hard nodig” [very necessary] (Hoeksema, 2005, p. 4), and “ergens hard aan toe zijn” [to be very ready for something] (Hoeksema, 2005, p. 4). These examples reflect the Middle Dutch usage of this adverb (Hoeksema, 2005). WNT (2015) also lists examples in which it modifies the verbs

“twijfelen” [to doubt] “verlangen” [to long], “vrezen” [to fear], and “verwonderen” [to wonder, be surprised], though these examples are from the eighteenth and nineteenth centuries, and a corpus study would be necessary to examine its genuine modern usage. WNT also notes that

“hard” is used more often in the northern Netherlands, though, once again, it is unclear how current this information is. On the other hand “zeer” is still used as an adverb of high degree, as discussed earlier in section 2.1.3 for example, but it is generally less frequent than “erg”

and “heel” (Hoeksema, 2011), and it is considered to be a part of a more formal register (Hoeksema & Korterink, 2011). “Veel” is also still in use, though its usage is restricted to the comparative and constructions with “te” (Hoeksema, 2011) (see section 2.1.2). According to WNT, “machtig” still functions as an adverb of extremely high degree and is capable of modifying adjectives as well as verbs, though the examples are once again older. Finally,

“utermaten” is still used as a general purpose adverb of extremely high degree in Modern Dutch in the form “uitermate,” and it can be used to modify adjectives, adverbs, and verbs (WNT, 2015), while “grotelijk” is no longer used. “Uitermate” is generally perceived to be stronger that “zeer” (WNT, 2015). In Modern Dutch, all adverbs of degree tend to directly precede the modified word.

2.3. Adverbs of degree in other Old Germanic languages

It will also be relevant to examine the usage of adverbs of degree in languages closely

related to Dutch. This will mainly be done on the basis of three languages: Old English, Old

Saxon, and Gothic. Old English was chosen because it is generally the best documented of

these languages, Old Saxon because of its closeness to Dutch, as there was no hard linguistic

border between the two (de Vaan, 2017), and Gothic because it is the oldest of the Germanic

languages and should therefore most closely resemble the original Germanic system. The

adverbs in the subheadings are given in their Old English forms.

2.3.1. Swīþe and ful.

In Old English, the most common adverbs of high degree are “swīþe,” which is a cognate with Old Dutch “swītho,” and “ful,” with the former being the most commonly used one (Méndez-Naya, 2003). “Swīþe” continued to be used in the early Middle English period, and its usage is usually said to decline after 1250 (i.e. Ito & Tagliamonte, 2003), though corpus data places the start of the decline slightly later, and its last attestation was in 1525 (Méndez- Naya, 2003). During the Middle English period, its function was gradually taken over by “ful,”

until the usage of that adverb began to decline as well in Late Middle English and the Early Modern period (Méndez-Naya, 2003). Today “full” is still used in fixed expressions like “full well,” but it can no longer be used in general contexts (Bolinger, 1972). “Swīþe” is a grammaticalised adverb of degree derived from the adjective “swīþ,” which meant “strong”

(Ito & Tagliamonte, 2003, p. 259) (c.f. Gothic “swinþs” [strong, sound, healthy] (Wright, 1910, p. 346), Old Saxon “suīð(i)” [powerful] (Cathey, 2002) and Middle Dutch “swinde” or “swide”

[powerful] (MNW, 2015)), using the derivational suffix that is reflected in Old English as -e or -a, and in Gothic and Old Saxon as -o (Wright, 1910). The Gothic form, as well as the Middle Dutch form with -n- show that this form illustrate the deletion of nasals before fricatives mentioned in section 2.2.1. Its usage in Old English has been excessively studied by Méndez- Naya (2003), and she found that it most often modifies adjectives and adverbs, but it was also capable of modifying verbs and participles, though no semantic analysis on the modified words is provided. “Swīþe” clearly differs from “ful” in this regard, as the latter is only rarely attested modifying verbs (Méndez-Naya, 2003). The following examples illustrate the different usages of “swīþe” and are adopted from Méndez-Naya (p. 380) ¹ :

(9) a.

1

The normalisation of the Old English spelling has been adjusted to the format used in Mitchell and Robinson (2012), as Méndez-Naya (2003) applied length marks inconsistently. The translations and glosses have been slightly altered compared to Méndez-Naya (p. 380).

forðon ðe iċ eam swīðe mildheort

because I am very merciful

(adj.)

b.

c.

(verb) d.

One of the Saviour’s retainers was then very enraged (ptc.)

No detailed description exists of the adverbs of degree in Old Saxon, but “suīðo,” a cognate with Old English “swīþe” is attested 56 times in the Old Saxon epic poem Heliand, and its usage appears similar, as can be seen in the following example sentences based on the edition from Cathey (2002) ² :

(10) a.

Joseph, the wise man, the very good man, began to think of those things to himself in his mind (Heliand, 312-314) (adj.).

b

I should very truly and eagerly tell you something more pleasant (Heliand, 397- 398) (adv.).

2

The normalisation used for Old Saxon differs slightly from Cathey’s (2002), in that the present study uses macrons to indicate long vowels in place of circumflexes in order to bring it in line with the normalisation used for Old English, and Gothic.

ġeseah Crist standan and þone clǣnan flocc mid

saw Christ stand and the pure flock with

him […] swīðe hlūde singende þone heofonlīċan sang

him very loudly singing the heavenly song

(adv.)

iċ swīðe wundrade þā ġesǣliġnesse þǣre eorðan I very much marvelled at the happiness of the earth

Hǣlendes þeġna sum þā wæs swȳþe ġebolgen

of the Saviour of the retainers a certain then was very enraged

Bigan im the uuīso mann, suīðo gōd gumo,

began to himself the wise man very good man

Ioseph an is mōda thenkean thero thingo

Joseph in his mind to think of those things

ic scal eu […] lioƀora thing suīðo uuārlico uuilleon seggean

I should you.PL dearer thing very truly eagerly say

c.

The women very much worried in their spirits (Heliand, 5789-5790) (verb).

d.

See now how you stand fastened and very injured (Heliand, 5578-5579) (ptc.).

A peculiar usage attested in Old Saxon that is not mentioned by Méndez-Naya (2003) for Old English is its ability to modify prepositional phrases, though, in the Heliand at least, this appears limited to the fixed expression “suīðo an sorgun,” as seen in this example:

(11)

Then Judas again proceeded to go from there to the sanctuary of God very worried (Heliand, 5159-5161).

A cognate with “swīþe” is not attested as an adverb of degree in Gothic. Usage of “swīþe”

persists in Modern West Frisian (WFT, 2015) as “swiid” and certain Dutch Low Saxon dialects as “zwied” (WNT, 2015), though its usage in the former is at least chiefly literary. The following example from 1911 found in WFT (2015) illustrates its usage in West Frisian:

(12)

The usage of the adjective form is also attested in Hollandic texts from the seventeenth century in the form “zwijdig” (de Vaan, 2017).

2.3.2. Sāre.

As mentioned in section 2.2.4, “sēro” is used as an adverb of high degree in Old Dutch.

Notably, this word is also used as such in the Heliand, though it is only attested twice. The following line illustrates its usage:

Thiu uuīf soragodun an iro seƀon suīðo the women worried in their spirit very much

Sīnu huō thu nu gifastnod stēs, suīðo gisērid see now how you.SG now fastened stand, very injured

Thō giuuēt imu eft thanan Iudas gangan

then proceeded himself again from there Judas to go

te themu godes uuīhe suīðo an sorgun

to the of God sanctuary very in worries

De Friezen wierne fen âlds swiid forneamd

The Frisians were of old very famous

The Frisians have been very famous since old times.

(13)

Interestingly enough, the Old English cognate “sāre” is not used as such, and it is only used as a lexical adverb (Hall, 1916), while the word is only attested as a noun in Gothic in the form

“sáir.” This, along with the fact that Modern German also uses “sehr” exclusively as a functional adverb of high degree (van Os, 1988), may indicate that this is a specific development in the Continental West Germanic languages.

2.3.3. Fela.

In Gothic, adverbs of degree are scarcely attested, but “filu” in Gothic could serve as an adverb of high degree. This adverb is generally found as an adverb of frequency with the meaning of “much” or “many,” but it is used as a translation of Greek “σφόδρα” [very] in Luke 18:23 (Carlson, 2012), which indicates that “filu” could be used as an adverb of high degree in Gothic. Its exact usage has not yet been studied, but it is attested modifying adjectives, verbs, and present participles, as can be seen in the following examples based on Wright (1910) in which it must be translated as an adverb of high degree:

(14) a.

They had met him two people possessed with devils, very dangerous, running out of the tombs (Matthew 8:28) (adj.).

b.

And he very much begged Him, saying “my daughter is at the point of death”

(Mark 5:23) (verb).

uuarð imu Satanas sēro bitengi became to him Satan very allied Satan became very allied to him (Heliand, 4624).

gamōtidēdun imma twái daimōnarjōs us

they had met him two possessed with a devil out of hláiwasnōm rinnandans sleidjái filu

tombs running dangerous very

jah baþ ina filu qiþands þatei daúhtar

and he begged Him very much saying that daughter meina aftumist habáiþ

mine is at the point of death

c.

And He had seen the tumult and those weeping and those very much lamenting loudly (Mark 5:38) (ptc.).

Constructions in which “filu” modifies an adverb are not attested. Another adverb of high degree that is used in Gothic is “abraba” (likely from “abrs” [strong, violent], c.f. Old Icelandic

“afar” (Wright, 1910, p. 302), likely with the adverbial suffix -ba (see Joseph, 1982)), which is also used as a translation of “σφόδρα,” though Carlson (2012) argues that this is a lexical adverb of degree meaning “terribly” based on the contexts in which it is attested. “Filu” also appears to be used as an adverb of degree in Old Saxon (Cathey, 2002), though this usage is not always clearly distinguishable from its usage as an adverb of frequency. In Old English,

“fela” could also be used as an adverb of high degree, and it often occurred in fixed expressions like “felamōdiġ” [very bold], “felalēof” [very dear], and “felameahtiġ” [most mighty] (Hall, 1916). The fact that cognates with “fela” is used as an adverb of degree in all of these older Germanic languages, including Old and Middle Dutch, suggests that this may be the oldest one.

2.4. Research question and hypothesis

As mentioned in section 2.2.4, no proper description of the adverbs of degree in Old and Middle Dutch currently exists. The present study will therefore seek to document and describe them based on the syntactic, semantic, and sociolinguistic factors as laid out in sections 2.1 and 2.4 using data from historic corpora. The purpose of the present study is twofold: it will first seek to document how the individual adverbs of degree in both Old and Middle Dutch, and then it will attempt to investigate how well adverbs of the same degree can be distinguished based on linguistic and sociolinguistic factors. For the first part, this study will investigate which particular words are most strongly associated with the specific adverbs, which lexical categories the adverb in question is attested modifying, and the position of the adverb relative to the modified phrase. It will also examine in which syntactic or semantic contexts the adverbs occur; for example, whether certain adverbs have a preference for positive or negative words, specific adjective scales, or positive or negative environments. This

jah gasaƕ aúhjōdu jah grētandans

and He had seen tumult and those weeping

jah wáifaírƕjandans filu

and those lamenting loudly very much

study will also attempt to investigate whether there is dialectal variation in their usage, as this is the main sociolinguistic factor available, and whether there is variation between writers if this information is known. The Old and Middle Dutch material will be analysed separately.

Based on the grammaticalisation patterns outlined by Bolinger (1972) and Klein (1998), it is also expected that functional adverbs of degree that occur in a restricted context in Modern Dutch will show fewer restrictions in the earlier stages of the language, and they may show more traces of their lexical meanings, while those that have fallen out of usage will be more specialised. For the second part, this study will seek to answer the question to what extent dialect, syntactic, and semantic factors influence the choice of the adverb of degree in Old and Middle Dutch. If Humboldt’s Principle held true in Old and Middle Dutch, it is expected that the different adverbs of degree were clearly distinguishable based on at least some of these factors.

3. Method 3.1. Materials

The main corpora used for the analysis are the Corpus Oudnederlands (INL, 2012) for the Old Dutch material and the Corpus Gysseling (2013) for Early Middle Dutch. The former includes all texts between 475 and 1200, which includes the material described in section 2.2.1, as well as shorter text fragments and runic inscriptions considered part of Old Dutch, and it forms the basis for the ONW (2015). The corpus comprises roughly 26,000 words in total, excluding toponyms. The Corpus Gysseling contains all Middle Dutch texts from before 1300, and it forms the basis for the VMNW (2015). The corpus includes both literary texts as well as charters, and it contains approximately 1.5 million words in total. Statistical analyses were performed using R.

3.2. Procedure

The corpora mentioned in section 3.1 were searched for the various adverbs of degree

as laid out in section 2.2.4, and the information was recorded in a database. The adverbs used,

along with their respective degree, are listed again in Table 2. For each instance of an adverb,

both the word itself and the phrase it modified were recorded in their original attested forms,

as well their Modern Dutch lemma forms in order to properly group them, as the orthography

in Old and Middle Dutch was not standardised and spelling therefore varied widely. For

context, the full phrase in which the adverb was found was also included in the database. The

adverb’s degree was also recorded following the conventions shown in Table 1 based on VMNW (2015) and ONW (2015), as well as the category of the modified word or phrase. The included categories were as follows:

• Positive adjective

• Comparative adjective

• Superlative adjective

• “Too” + adjective

• Adverb (general)

• Comparative adverb

• Adverb of degree

• Verb

• Participle

• PP

• Numeral

It was also recorded whether or not the sentence is negated and whether the modified phrase is inherently positive or negative. The text in which the sentence or clause is attested, its dialect, and the year in which it was written were also included in the database, as well as the writer of the text, if this information was available. Finally, the corpus in which it was found was also included in order to compare the Old and Middle Dutch material. In total, 1743 entries were recorded in the database.

Table 2.

A table listing the different adverbs of degree in Old and Middle Dutch that are included in the analysis grouped by their respective degree.

Degree Old Dutch Middle Dutch

High hardo, sēro, filo, swītho hard, zeer, veel Extremely high mahtigan grotelijk, utermaten Minimal and

quasinegative

iewiht een deel, bore, iewet, weinig

(Church & Hanks, 1990, p. 23) 3.3. Analysis

The database was then loaded into R in order to perform various statistical analyses.

In this analysis, the choice of adverb was the dependent variable, and the modified phrase, category of the modified phrase, whether or not the clause is negated, whether or not the modified phrase was positive or negative, dialect, writer, and year were the independent variables. For each adverb of degree, a list of modified phrases was obtained along with the frequency of each pair, along with its pointwise mutual information (PMI) in order to evaluate the strength of the association. The PMI 𝐼(𝑥, 𝑦) can be calculated using the following formula:

𝐼(𝑥, 𝑦) = log ₂ 𝑃(𝑥, 𝑦) 𝑃(𝑥)𝑃(𝑦)

Here, 𝑃(𝑥, 𝑦) represents the joint probability of words 𝑥 and 𝑦 , while 𝑃(𝑥) and 𝑃(𝑦) represent their individual independent probabilities (Church & Hanks, 1990). A PMI value greater than zero means that their co-occurrence is greater than chance, while a value below zero means that the probability of them occurring together is below chance (Church & Hanks, 1990). The individual probabilities are the observed frequencies of each of the two words normalised by the corpus size (Church & Hanks, 1990). For this study, the joint probability was calculated based on the frequency of the adverb 𝑥 modifying phrase 𝑦 normalised by corpus size, rather than 𝑦 simply following 𝑥 as used by Church and Hanks (1990), because the two did not always occur adjacent. For individual words, the frequencies were taken from VMNW (2015) for Middle Dutch and from ONW (2015) for Old Dutch. For PPs, the phrase was treated as one chunk, and therefore the frequency of the whole phrase was used. The mean separation along with the standard deviation (SD) was also provided, as suggested by Church and Hanks (if the mean separation is negative, 𝑦 most often preceded 𝑥). This analysis was based on Church and Hank’s “association ratio” (p. 23). Due to the fact that the corpora contained a large number of words with only a small number of attestations, only word pairs with at least five attestations were included in the PMI tables. Otherwise, the words with only a few attestations will have the highest PMI value, which would not be as meaningful.

Whenever this was not possible due to a lack of pairs with at least five attestations, the ten

words with the highest PMI values were included regardless of pair frequency. This was

especially applicable to the Old Dutch material due to the smaller corpus size. When counting

the separation, enclitic elements were treated as separate words. This was followed by a

qualitative analysis of the modified phrases to see if patterns can be detected in the kind of

words that were modified. Similarly, a list of the most commonly modified categories was obtained along with their frequencies. Furthermore, they were then analysed to see if certain adverbs have a preference for negative or positive contexts, both inherent and based on environment, based on frequencies. The usage of the adverb in question was then analysed for the different dialects, texts, and writers to see if certain constructions were restricted to or more frequent in particular dialects. Finally, its usage in Old Dutch was compared and contrasted with its Middle Dutch counterpart.

After having analysed the different adverbs of degree separately, the ones of the same degree were then compared and contrasted to find differences in their usage. When there were more than two adverbs to compare, a multinomial logistic regression model was fit using the “one vs. rest” (Levshina, 2015, p. 283) approach to investigate which of the language internal and external factors mentioned above significantly predict the choice of adverb of degree in Middle Dutch. Whenever there were only two adverbs to compare and contrast, a binomial logistic regression model was employed instead.

4. Results 4.1. Adverbs of high degree

4.1.1. Hard.

“Hard” is attested as an adverb of degree 424 times in the Corpus Gysseling (2013). It

most often modified adjectives and other adverbs in Middle Dutch, though it could also modify

verbs and participles. The distribution of the modified categories is shown in Figure 1:

Figure 1. The distribution of lexical categories for the adverb “hard” in Middle Dutch.

However when broken down by text, it is revealed that all seven attestations of “hard”

modifying participles are from a single text: Sente Lutgart, handschrift K, a Brabantian text dated to 1265 and written by Willem van Affligem. This text also contained ten out of thirteen attestations of it modifying a verb. Examples in which “hard” appeared to function as an adverb of manner are not included in this figure. Using Fisher’s exact test, it was found that the distribution of categories for this text differed significantly from the others (p < .001).

Notably, “hard” also often modified other adverbs of degree, such as “zeer” (45 times), “zwaar”

(9 times), “naar” (6 times), which functioned as an adverb of approximate degree in this context (VMNW, 2015), “utermaten” (2 times), and “groot” (1 time). “Hard” had also a preference for modifying positive words, which it did 324 times (76.42%), while it modified negative words 100 times (23.58%). A list of words with the strongest association with “hard”

ordered by PMI is shown in Table 3. Typically, “hard” directly preceded the word it modified,

though some variation is observed. This is especially true for verbs and participles, as can be

seen in Table 4. While was is not strictly a positive polarity item, “hard” had a strong

preference for positive sentences, as it occurred there 417 times (98.35%) and only 7 times in

negative ones (1.65%).