Table of Contents
2. The modal system, grammar, and grammaticalization
3. Ought (to) and should in PDE
3.2 Types of modality
4. Historical development
4.1 The development of ought (to)
4.2 The development of should
5.1 Parameters of grammaticalization (according to Lehmann)
5.2 Principles of grammaticalization (according to Hopper)
5.3 The emergence of epistemic modality
illustration not visible in this excerpt
It is a disputed question if the lexicon and the grammar of a language can be seen as separate entities which constitute a dichotomy. As it is often the case, the answer depends on the particular point of view. It is well-known that structuralism starts out from a synchronic perspective, a particularity it has often been criticized for. As Lehmann (1995: 8) points out, “structural linguistics has tended to favour a static view of language and clear-cut binary distinctions”. Due to the striking influence of this school of linguistics since its emergence in the first half of the 20th century, it is not surprising that many introductory works that draw on its ideas convey the idea of absolute separateness of lexicon and grammar. Brinton (2000: 8), for example, defines ‘grammar’ in her book The Structure of Modern English: A linguistic introduction as “the rules or principles by which a language works, its system or structure”. She indicates that one part of morphology can be clearly attributed to grammar, since it is concerned with “grammatical morphemes [which] may be either free roots (function words) or bound affixes (inflectional suffixes)” (Brinton, 2000: 103). She further states that “in analytic languages, such as Modern English […], the grammatical categories are expressed primarily by word order […] and by function words, as well as by a few inflections” (Brinton, 2000: 103). This view mirrors the common perception that grammar includes morphology and syntax. The other part of morphology is then supposed to be taken up by “[l]exical morphemes [which] express lexical, or dictionary, meaning” (Brinton, 2000: 76). All this gives the impression that grammatical meaning can be thoroughly distinguished from lexical meaning. A theory that “challenges the discreteness of linguistic categories, even that between grammar and the lexicon” is grammaticalization, because it “relies […] on the gradualness of change that allows intermediate stages” (Wischer, 2006: 130). This statement already implies that grammaticalization manifests itself best in diachronic investigation and that it is incompatible with structuralism. The term ‘grammaticalization’ can be defined in the following way:
‘grammaticalization’ refers to that part of the study of language change that is concerned with such questions as how lexical items and constructions come in certain linguistic contexts to serve grammatical functions or how grammatical items develop new grammatical functions. (Hopper & Traugott, 2003: 1)
This paper aims at observing the processes and mechanisms of grammaticalization, taking as an example the development of ought (to) and should from earlier stages of the English language up to the present day. In the very beginning, terminological questions will be discussed. Therefore, the first step will be to concentrate on the term ‘grammar’, before the terms ‘modality’, ‘mood’, and ‘modal system’ are analyzed. What will follow is a look at attempts to categorize the constructions under examination, since the allocation of ought (to) is disputed and somewhat fuzzy. Although the next step will mean not to proceed completely chronologically, it is important to describe how should and ought (to) work in PDE to better understand their development. After that, the semantic and morphosyntactic development of both constructions throughout the history of the English language will be described. Taking the observed changes as a basis, the development of the forms will finally be related to grammaticalization, where the principles proposed by Lehmann (1985) and Hopper (1991) will play a major role.
2. The modal system, grammar, and grammaticalization
It is one common view that the verbal grammatical categories in English are number, person, tense, correlation, aspect, mood, and voice. However, Palmer (2003: 2) asserts that “what has traditionally been called ‘mood’ is only one type of grammatical sub-category within a wider grammatical category”. According to his view, “the name used for the grammatical category is simply ‘modality’ and […] there are two sub-categories of modality, mood and modal system” (Palmer, 2003: 2). So far, Palmer’s view does not seem to be too novel, but he goes as far as to affirm that “Modern English has a modal system, but no mood” (Palmer, 2003: 4). This denial of the existence of mood in English seems quite a daring claim, especially if we consider the number of publications on the subjunctive. Many scholars would therefore certainly not subscribe to this view. Nevertheless, the focus in this paper will be on what Palmer calls the modal system, and it will follow the terminology he proposes. In contrast to mood, which “involves a binary system” and indicates a claim made by the speaker, a “modal system” always comprises “a set of modal forms”, which “simply indicate judgements by the speaker” (Palmer, 2003: 2-3). In English, these forms are without a doubt the modal auxiliary verbs.
In Palmer’s work, it seems to be automatically assumed that modality is a grammatical category, what implies that the linguistic forms used to express this grammatical category belong to the realm of grammar. But one of the most serious dilemmas that emerge in discussions about grammaticalization cannot just be overlooked here: if we presuppose a gradual transition from lexicon to grammar, it is impossible to draw a line between them (Hopper, 1991: 19). Nevertheless, there seems to be a way out if we rely on what has been called ‘grammar’ based on cross-linguistic comparisons:
While a definition of grammar, and hence of grammaticalization, is problematical for an individual language taken in isolation, some working assumptions about grammar are possible from a cross-linguistic perspective. (Hopper, 1991: 19)
So what can be said about modality? Hopper (1991: 19) indicates that “[c]ategories which are morphologized might safely be said to be part of grammar”. Although modal verbs in English undoubtedly help to form periphrastic constructions, there are languages where modality can be morphologized, as for example in Turkish (Göksel and Kerslake, 2005: 326). Consequently, modal verbs can be considered as a part of grammar. The following observation confirms this, since it links modal verbs directly to grammaticalization:
Modality […] is a gradient (ant not binary) notion, and change within the modal system of any one language is endemic. Consequently there is a host of modal expressions in contemporary English […] which either already exhibit full-fledged modal features, or at least show signs of change towards modal behaviour, i.e. of grammaticalization. (Facchinetti et al., 2003: viii)
There are several conclusions that can be drawn from this. Firstly, it is obvious that modal verbs do not constitute a class quite as rigid as for example the class of personal pronouns, although the group of modal verbs could certainly be seen as a closed class in PDE. Secondly, not all the modal verbs are alike, but can be grammaticalized to a different degree. An increase in “modal behaviour” corresponds to a higher degree of grammaticalization. Thirdly, and most importantly, it is definitely clear that modal verbs are subject to grammaticalization and that it is therefore justified to examine them with regard to this phenomenon.
3. Ought (to) and should in PDE
It will be useful to have a brief look at the categorization of ought (to) and should in PDE, as well as at the functions they fulfill, before describing their diachronic development. Keeping in mind the observations just made about modal verbs in PDE, it can be stated here - in the same way as Harris (1986: 347) does - that there is no doubt whatsoever that should belongs to the class of PDE modal verbs (or modal auxiliaries) alongside shall, can, could, will,
would, may, might, and must. Biber at al. (2002: 174) describe them as the “central modal verbs”, Mindt (2000: 116) states the existence of “nine central modals” as well, and Facchinetti et al. (2003: vi) speak of the same group of verbs as “’core’ modals”. Harris (1986: 347) concisely summarizes the “well-known characteristics” all these verbs have in common:
the absence of flexional -s on third person singular present forms, the ability to form questions by simple inversion and negation by the addition of not/n ’ t, and […] the fact that they collocate with the bare infinitive, i.e. without the particle to. (Harris, 1986: 347)
Furthermore, they do not allow the formation of a present participle or past participle form. Consequently, all the words that are generally accepted as PDE central or core modals share the same morphosyntactic behavior.
Ought (to) deviates from this pattern insofar as there is one single formal feature of the central modals it does not comply with: it is normally used with the full infinitive (also called to - infinitive) and almost never with the bare infinitive. According to Harris (1986: 351) there are some speakers who use ought followed by a bare infinitive, but they are definitely a minority. Hence, the use of ought with bare infinitive in PDE will be disregarded in this paper, although it cannot be disregarded in other historical stages of English.1 All in all, it can be concluded that ought to does not belong to the class of central modals, at least according to the morphosyntactic criteria. From a semantic point of view, however, it “clearly belongs” to the already mentioned class of modal auxiliaries (Harris, 1986: 349). How can ought to then be unambiguously categorized if it is by definition not one of the central modals, although it is very similar to them? Harris (1986: 351) concludes that this construction “sits uncomfortably astride the major morphosyntactic category division within the English verbal system”. In an attempt to solve this dilemma, a variety of classifications have been proposed: Biber et al. (2002: 174) classify ought to under the group of “[s]emi-modals”, which they describe as “multi-word constructions that function like modal verbs”, whereas Mindt (2000: 116) categorizes ought (to) as one of two existing “modal catenative constructions” alongside used (to). Krug’s observation is that ought (to) can function in very different ways, but argues that the most reasonable classification - if any is made - would be as a lexical verb, basing his assumption on quantitative data:
OUGHT TO is somewhat equivocal regarding its syntactic behaviour. Overall, however, due to the fact that truly main verb behaviour (to infinitives in positive contexts) accounts for some 95% of all uses, while the remaining options (truly modal and hybrid structures) divide the remaining 5 percent among themselves, the affinity with lexical verbs is more striking. (Krug, 2000: 206)
This point of view has to be handled with great care. Krug interprets every instance of ought followed by a full infinitive automatically as a main verb, since he only takes into consideration morphosyntactic, and not semantic criteria. Seen from another angle, this seems quite unacceptable, since modality, e.g. “judgements by the speaker” (Palmer, 2003: 3) or an obligation, can definitely be expressed by ought to. The fact that ought to and should are interchangeable in so many contexts underlines this statement. Accordingly, in this paper the focus will be more on the meaning of ought (to) throughout history than its conformity with certain morphosyntactic rules.
3.2 Types of modality
There have been several propositions to categorize different types of modality. Bybee et al. (1994: 177) suggest a distinction between “agent-oriented, speaker-oriented, epistemic, and subordinating” modality. Nevertheless, Palmer (1991: 85) names comprehensible arguments why this analysis might be confusing and “difficult to understand”, and proposes an alternative on which the considerations in this paper will be based on. According to him, the following tripartite categorization is more adequate: “Epistemic modality”, as the first type, “is concerned solely with the speaker’s attitude to status of the proposition” (Palmer, 2003: 7). Should and ought to can both express this kind of modality:
(1) It should be safe here.
(2) It ought to be safe here.
In both cases, the speaker expresses the high probability that it is safe here. Together with “Evidential modality”, which will be disregarded here because Palmer does not include it into the three main types, “Epistemic modality” is also called “Propositional modality”. The second type, which is called “Deontic modality”, can either express “permission” or “obligation”, and “the event is controlled by circumstances external to the subject of the sentence” (Palmer, 2003: 7). In the case of should and ought to obligation that comes from outside can be expressed (although a weaker obligation than expressed by must or have to):
1 For this reason, ought to will be written without brackets when PDE phenomena are treated. 4