The Development of Periphrastic do in English

Contents

1. Introduction

2. Language change
2.1. Societal background
2.2. Standardization
2.3. Inflectional loss

3. The evolution of do
3.1. Periphrastic do
3.2. Causative do
3.3. Do in questions
3.3.1. Syntactic approach
3.3.2. Semantic approach
3.3.2.1. Rhetorical questions
3.3.2.2. Undesired endings

4. Do as a tense-carrier in affirmative sentences

5. Negative do

6. Emphatic do

7. Final remarks References

1. Introduction

The history of the English language can be divided into at least four stages, Old English, Middle English, Early Modern English, and Modern English. The language development has its characteristic features in each stage. Though during the change from Middle English to Early Modern English in the fifteenth and sixteenth century the language underwent basic changes in its syntax and vocabulary. These changes affect the English language of today insofar that the syntactic constraints still determine Modern English syntax. In this paper I will try to figure out the influences at work in this process. People, language, society - these three can not be separated since they can only work as one. The constraints of one on the other are a highly complex matter that goes beyond the border of linguistic science alone. For this reason chapter two presents a general background on history and society in relation to the English language in the time of interest.

One main aspect of these changes is the strong rise of do in all contexts in the sixteenth century. The adoption and grammaticalization of do was a relatively rapid process of probably less than 150 years. Chapter three will focus on the development and spread of periphrastic do, particularly on its use in questions because of the high increase in this context, and in this respect contrasts syntactic and semantic points of view. Chapters four, five, and six highlight further uses of do and work out their relation to its use in questions. In chapter seven I will sum up the basic results and provide a personal evaluation.

2. Language change

2.1. Societal background

Usually, we think of languages as a means to communicate with each other. The spoken language of the period of time this paper is concerned with can only partly be reconstructed by written texts that survived up to now. These written texts reflect the attitude of their authors towards correctness of spelling and content, for example.

The language of a medieval manuscript is the product of many people. Hence each text "speaks" with several voices - at least with those of the author and the copyists, and sometimes readers who add comments. Any historical account of the English language has to consider that not only the language in general but also individual texts represent the work of many people. One of the problems we face today concerning the history of the English language is that the West Saxon perspective is mainly the basis for the story of English. Especially scholars, philologists, and linguists of the 19th century adapted to this point of view. The reasons for this may be that the surviving Old English texts are mainly written in the West Saxon dialect and thus hand down a certain attitude towards the English language, and "because evidence has been selected and interpretations shaped to suit particular stories" (Graddol et al. 1996: 131). In the introduction to that chapter Graddol et al. state that "the most familiar story of the English language [...] was mainly constructed in the nineteenth century, and [...] draws particularly on nineteenth-century ideas of national identity." (1996: 95). It seems likely to me that in the period in which periphrastic do generated and spread there was no such concept of nationalism as in the nineteenth century, rather there may have been some kind of patriotism resulting from the Hundred Years War against France.

For this reason we have to bear in mind the sociological, political, and economic context in which the texts were produced to understand their historical relevance for the development of the English language. The development of periphrastic do as a marker of courtly speech, for example, is a clearly socio-cultural effect. The distinct use of do in direct speech in courtly romances in the sixteenth century is a definite indicator for high style (Stein 1990: 115). Thus the association of do with courtly style of speech leads to the cultural manners which are very likely related to speech. The use of do related to a certain stereotype of life and style went down with the disfavor of that style in the public. Courtly speech in association with do is clearly ridiculed in one of Shakespeare's plays (Stein 1990: 126). If it had not fallen into disfavor it could not have been ridiculed. In Figure 1 this effect can be traced shortly before the year 1600. In almost all cases the use of do declined for a certain time.

illustration not visible in this excerpt

Frequency of do -forms in different sentence types

(adapted from Stein 1990: 15)

A = affirmative statements

B = negative statements

C = negative imperatives

D = direct affirmative questions

E = direct negative questions

In that time several social, economic, and political factors led to this loss of courtly prestige. During the Early Modern English period there developed a radically new political and economic form. From the social and economic changes in society (early capitalist system based on manufactures) a new class system developed. Together with the new economic form came the rise of the middle class. The aristocracy was heavily in debts and had to sell of parts of their property.

The Renaissance brought intellectual liberation with it which resulted in a great number of new discoveries in scientific research. The gain in sciences and the radical societal changes brought with them the necessity for new English words. The concept of nation states and a national language became another leading factor. Graddol et al. (1996: 118) indicate that because of the multiple factors that influence language development it is difficult to determine the cause of language change:

[...] it is only relatively recently that the question of why languages change has been investigated. This process remains mysterious, however, not least because of the multiplicity of factors which appear to influence it. For instance, we know that a variety of non-linguistic (external) factors can lead to language change; the effect of repeated invasions on the vocabulary of English is one clear example. But many linguists have long argued that there also exist internal reasons for linguistic change. That is, the grammar of English may have had some kind of built-in instability which made certain kinds of change likely.

So the question is: How does language change come about? If one speaker of a language community starts using the language in a new way respectively introduces a new element why should other members of that community follow? An answer to this must take into consideration the social relations between speakers, and this is clearly a step towards semantic explanation since it is apparently a part of the language-external world.

Stein asks the same question: "[...] what is it that changes individual choices to social ones, to group choices?" He assumes the process at work to be a natural one, that is, language-internal (c.f. chapter 3.3.2.2.). He excludes some "external masterminding or monitoring" but speculates that there would be a kind of "unconscious, permanent self-monitoring of language evolution" (1990: 287).

Graddol et al. state that "in recent years [...] sociolinguists have shown that linguistic changes are often associated with particular groups in society, and that people tend to adopt changes introduced by more powerful or prestigious groups" (1996: 119). This is evidenced by the findings they present on the process of standardization ( Chapter 2.2 of this paper), the process of selection would account for this proposition. Traugott (1972: 13) also observes that "over time gradual quantitative change is correlatable with such extra-linguistic factors as sex, age, class, and so on".

illustration not visible in this excerpt

As I mentione d before, on the backgrou nd of the 100- years- war against France develope d a certain kind of patriotism in the consequence of which the English identified themselves with their own vernacular after French has been the official language of the court for about 300 years. In 1362 English was introduced as the official language of legal affairs. The English language gained so much prestige during the Early Modern English time that it was even used for scientific writing although Latin remained the main language in science for several centuries as is shown in Figure 2:

However, scientists began to publish their works in both languages. Since Latin was a highly developed language with a fixed grammar that had been in use ever since the Romans had left it was ideal for the explanation of scientific ideas. To the contrary was English with its variety of dialects, there was no such thing as a standard English language not to speak of fixed grammatical rules respectively any conventionalized spelling which caused difficulties for publishers as well as scientists (Graddol et al. 1996: 138).

2.2. Standardization

In spite of Graddol et al.'s critique (1996: 123) that in the nineteenth century a codified "standard" language was seen as "the triumphant happy ending to the story" and this "embodies value judgment of a different kind - that diversity and variations are less desirable than a unified standard form", the necessity for a certain standard eventually was there with the invention of printing. When William Caxton introduced the first printing press in England around 1476 he was confronted with such problems as diversity and dialectal variation (as well as spelling), "[...] that comyn englysshe that is spoken in one shyre varyeth from another. [...] certaynly it is harde to playse euery man / by cause of dyuersite & chaunge of langage" (William Caxton 1490, from the prologue to Eneydos, as quoted in Görlach 1993: 217). For his printings Caxton eventually decided to adopt the dialectal variety of the area which he lived and worked in, London and the Southeast which was also socially and economically the most powerful area. With this decision he has built the foundation for establishing this dialect as "the English language". In the mid sixteenth century the publication of an English translation of the Bible was another crucial moment in the process of standardization.

Apparently this process is strongly connected to the societal structure and other language-external circumstances. Graddol et al. (1996: 139) differentiate four main processes of standardization:

1. Selection: of an existing language variety as the basis. The variety selected is usually that of the most powerful or socially influential social or ethnic group.
1. Codification: reduction of internal variability in the selected variety, and the
establishment of norms of grammatical usage and vocabulary. Since standard languages are rooted in written forms, standardization often also involves the establishment of a standard spelling for words.
1. Elaboration: ensuring that the new language can be used for a wide range of functions. This may involve the extension of linguistic resources; for example, new specialized vocabulary or even new grammatical structures.
1. Implementation: the standard language must be given currency by making texts available in it, by discouraging the use of alternative language varieties within official domains, and by encouraging users to develop a loyalty and pride in it.

In short: minimal variation in form and maximal variation in function.

2.3. Inflectional loss

The necessity for new words especially in certain fields of science, like medicine and anatomy, biology, theology, crafts and technology to name only a few, rapidly increased the word stock and also the polysemy of words around 1600. It has been estimated that in the two centuries between 1500 and 1700 about 30.000 new words were introduced to the English language with a peak around the year 1600 (Graddol et al. 1996: 142).

illustration not visible in this excerpt

Before this extensive increase in the word stock of English the syntax had undergone major changes in the ME period. The general process that occurred in ME was a simplification of forms or segmentalization, that is, a change by which an inflection is replaced by a word. The inflectional system that was fully in operation in OE was reduced in ME and inflections were replaced by prepositions. In Germanic languages the root syllable carries the main stress, pre- and suffixes are not stressed; if they were stressed they became meaning carrying such that they are no affixes any more. Especially the loss of unstressed vowel endings was made possible because they were already redundant and exchanged for other grammatical elements, that is, prepositions and word order. The inflectional loss in this case is apparently exclusively due to phonological and phonetic changes. The vowels a, o, u in unstressed inflectional endings were reduced to [¶ ] and eventually the unstressed inflections were lost, as shown in the example for sell:

illustration not visible in this excerpt

Although the major syntactical changes took place in ME Traugott (1972: 110) states that "both segmentalised aspect and prepositions are to be found in earliest OE; what we find in ME and later English is extensive generalisation of an already operative system". In general it is possible to say that non-informative elements are simplified in a language. If an element is functional and a carrier of meaning that element will usually be retained at least until a new one with the same function replaces it. Elements which are not functional and do not carry meaning are likely to be lost.

Since the inflectional endings that indicated number, gender, and case in OE were lost during the ME period word order became particularly important. Gender, for example, became obscure and could not be distinguished grammatically any more but only biologically. In the fifteenth and sixteenth century English changed from a verb first to a verb second language, that is, in main clauses the subject always precedes the verb (subject-verb inversion). The ME word order already shared the general characteristics with PDE (Traugott 1972: 160):

Sentence type 1: Subject (Auxiliary) Verb (Object) for coordinate and subordinate clauses like John came and Mary left and John came after Mary had left.

Sentence type 2: (X) Verb + Subject (Object) and (X) Auxiliary + Subject + Verb (Object) were generally used for interrogative sentences/clauses like Why is he in the house? (is appears as a main verb here) and Why does he stay in the house?

illustration not visible in this excerpt

As far as word order is concerned pronoun objects were treated differently from nouns. Until the do -periphrasis developed for negatives there were certain rules which were still at work to prevent a pronoun to appear in final position in several structures. An object pronoun usually preceded the negative not whereas a noun followed as in I know him not and I know not your brother (1972: 160)

In the following chapters I try to answer the questions: What is do -periphrasis? What factors influenced the spread of periphrastic do and its rapid increase in the sixteenth century? The OE verb don was a main verb. Which process was at work that made it become a "dummy" or "empty" operator?

3. The evolution of do

3.1. Periphrastic do

Do can function as an auxiliary and as a main verb. As an auxiliary it has only

present and past tense forms but no non-finite form. Finite forms are forms that can occur on their own in a main clause. Non-finite forms, however, occur on their own only in dependent clauses. The base form of do occurs in finite as well as in non- finite form.

illustration not visible in this excerpt

Table 1: Forms of do (adapted from Quirk et al. 1985: pp. 96, 133)

An auxiliary or operator is necessary if a positive declarative sentence is turned into another major structure in the language, for example a question or negation. The element is introduced as an operator because it fulfills an "operational" function in this case. The auxiliary form of do is inserted if such a sentence has no auxiliary verb and hence no operator. It is used as an "empty" or "dummy" operator since it does not carry meaning but defines number, person, and tense. It is this function of do that the term do -periphrasis refers to.

Like in PDE, OE had a verb don meaning 'to perform'. In this sense do may function as a main verb as well. In this case it has the full range of possible forms including the - ed participle done and the - ing participle doing. In example 2 below both functions co-occur, the auxiliary and the lexical use of do meaning 'to perform' (c.f. Table 1):

1. In questions where the verb is in the simple present or past tense form including inversion of subject and operator,

Did /Does she repair the bike?,

including tag questions where the operator is not accompanied by a main verb.

She repairs the bike, doesn't she?

2. In indicative clauses negated by not with the verb in simple present or past tense.

She didn't/doesn't do it. (+ use of do as main verb)

3. In emphatic construction in the simple present or past.

Please, `do come again.

4. In reduced clauses, where do replaces a verb phrase.

She knows much more about bikes than I do. [do = know about bikes] (adapted from Quirk et al. 1985: 134)

Like other auxiliaries do has various functions. It can be used as an optional tense carrier in any sentence construction. It can be used as an emphasizer in propositional phrases and can also be applied as a causative.

With the loss of the subject-verb inversion during EME (c.f. 3.3.1.) also the syntax of questions changed. The inversion was replaced by the do -form. There was a dramatic increase in the use of non-causative do in all syntactic contexts in the early sixteenth century. Two major uses of do can be identified: the semantic, contrastive meaning type and the formal, meaningless type which is used only for aesthetic purposes (for example, to achieve a prose rhythm). Stein (1990: 61) has made an approach towards the semantics of this syntactic change. He argues that the increase of periphrastic do in that time can not be deduced from language-internal causes because of its rapidity since natural change is usually gradual. He puts forward a different suggestion, that it is a culturally determined process that started with the use of do for stylistic reasons in poetry and literature which then had syntactic consequences. He further argues that it "was part of the quest for a first English vernacular prose style, of the taking over of Latinate stylistic ideals, and of the transition from orality to more literacy" (1990: 143). The general acceptance of the English vernacular in that time, as I have mentioned before, may have supported that development.

In his study Stein makes an attempt to show that periphrastic "meaningless"do in fact had meaning and how this meaning correlates with syntactic features like inversion, negation, emphasis, and interrogatives. He argues that periphrastic do derives from the development of an agentive and perfective semantics of do (see chapter 3.2.) which has a natural closeness to narrative texts (c.f. examples from East Somerset in chapter 4 on affirmative do): "Somebody does something where that something is evaluated as foreground on the discourse-structural level" (1990: 103). He found out that periphrastic do in early sixteenth century texts functions as a marker of discourse-semantic foreground, that is, it occurs at the high point or peak of an utterance and mainly in narrative contexts (1990: 35).The high point contains the reason why a speaker reports something at all. The motivation for the telling derives from the assumption of the speaker that there is a contrast, that is, something which seems unusual or unexpected according to the speaker's personal experience. Thus do involves the idea of contrastivity in the way that "a special circumstance or event is contrasted with a normality or otherwise 'unmarked' states, actions or circumstances" (1990: 38). Foregrounding is then the linguistic device used to highlight what is seen as the characteristic feature of the story. The prototypical semantics of do would be best characterized as contrastiveness.

Stein criticizes that in contrast there is no unifying principle for the syntactic approaches towards an explanation of periphrastic do. None of the numerous examples with periphrastic do in early sixteenth century texts contains subject-verb inversion or an adverb. Word order hypothesis would explain the history of do with the directly opposite finding (1990: 104), namely that the syntactical surroundings influenced the development of do and not vice versa (c.f. 3.3. on do in questions). The positional arguments including inversion, respectively the tendency to avoid inversion by the use of do (also in chapter 3.3.), highlights the opposition between semantic and syntactic approach: Is do present because its function is to avoid inversion or does it occur in these structures because its basic meaning is that of foregrounding? Stein pleads for the latter one (1990: 107), supported by a research result that in the Book of Common Prayer (1549) only 11% of all cases have an adverb between do and the verb; syntactic explanations argue that subject-verb inversion especially occurs in such sentences which contain an adverb (c.f. 3.3.1.). He states that "the synchronic connection of inversion with do is the result of a diachronic, semantically given overlap between inversion and intensity [more general term for foregrounding]" (1990: 280). In EME the discourse function of inversion + do was related to intensity. Do with subject-operator inversion occurs in such contexts. In this case do is expected because adverbs express the speaker's attitude towards the truth of a proposition and behave in that like negatives where do is also expected for that reason (c.f. chapter 3.3.2.1. and 5).

According to Stein (1990: 278) the distribution of subject-operator inversion can not be explained syntactically since in a merely syntactic context with space adjuncts as pre-posed constituents like in "In the doorway stood my brother" the statement is neither semantically connected to the truth of the proposition nor does it contain any personal attitude of the speaker towards it.

The notion of contrastivity is a basic hypothesis in the study of Stein which in my opinion adds one plausible semantic explanation for the spread of periphrastic do as will be shown in the following chapters.

3.2. Causative do

Do was used extensively as a causative in OE and ME, as in "He did him die", but was recessive by the end of the ME period. Many linguists support the hypothesis that periphrastic do developed from causative do. In a sentence like "He dude writes sende" (as quoted in Kroch et al. 1985: 284) the subject of the sentence is open to an agentive interpretation. The causative meaning would read as "He caused somebody to send writes". From the semantic point of view the subject of the lower clause somebody could be dropped in the original sentence. Thus it is possible to reinterpret the subject of the higher clause he as the agent, as doing the action himself. Finally there is no difference anymore between the "causing" and the "doing" of the action. There exist a lot of ambiguous cases like this one, since it can be interpreted both as causative or as periphrastic.

Stein suggests on the grounds of his findings in the earliest colloquial texts analyzed in his study, the Paston letters (1422-1509), that it could be a Latin-induced structure with perfective meaning (1990: 17) and the interpretation as causative or non- causative depends on the pragmatic context in which the proposition occurs. The perfective semantics of the construction was not focused on who did the action but on what happened. He points out that work has naturally been done only on written texts which at the time had strong Latin influence. Denison (1985: 52) also assumes causative do to have entered late OE as a Latinization that did neither clash with the syntax nor with the semantics of OE don. A problem arises if one assumes that periphrastic do developed from causative do because naturally the two should be successive in time then. What is striking here is that there is no time lag in the appearance of the two cases. Denison asserts that causative and periphrastic uses "appear more or less simultaneously" (1985: 47).

Stein highlights in his study the pragmatic background of the causative use of do. In most examples from the Paston letters the pragmatic context is directive. Do in this context then has a causative meaning. The agent is dropped, as in the example above, and focus is not on the subject or agent but on the directive being fulfilled either by the addressee or in his or her responsibility. Stein assumes that this second person, directive context has been the most frequent use of the original causative do.

Further he suggests that the most likely context was that of official writing which implies giving orders to individual persons. People usually received orders orally from a person in a socially higher rank, inclusively official and legal use. Probably people adopted this use but ignored the directive meaning and transferred it to the first and third person. There it developed its periphrastic form.

3.3. Do in questions

3.3.1. Syntactic approach

The rule for word order in Germanic languages is that the finite verb in main clauses should be in second position. This used to be the case in English until the fifteenth century. In German this rule is still at work (examples 1 and 2 below). By the fifteenth century the word order in English sentences had changed in that the subject in main clauses preceded the main verb (example 3.). Additionally, the object could no longer be placed before the main verb in declarative sentences because without inflection it was not distinguishable from the subject phrase any more (Traugott 1972: 160). With this restriction to Subject-Verb-Object word order the syntax of ME changed into the modern type that is still valid for PDE. This subject-verb inversion occurs especially in sentences with adverbs and NPs which function as a topic in front position:

1. G: In diesem Jahr ging König Edward nach Maldon.
2. OE and EME: This year [...] went King Edward to Maldon.
3. ME: That oon of hem gan callen to his knave (Chaucer: 666).

Kroch et al. (1982) argue that the change in the syntax of declarative sentences obviously had an effect on the syntax of questions. Here one may ask, then, why do - periphrasis is twice as high in questions compared declarative sentences (c.f. chapter 3.3.2.2.) and shows a much higher frequency in Figure 1.

Until the end of the fifteenth century wh-questions as well as yes/no-questions show subject-verb inversion as in "Why cridestow?" (Chaucer:1083) which is apparently due to the Germanic verb second constraint, at least for wh-questions. This is particularly obvious when the wh-element is the subject of the sentence as in "Who hath thee doon offence?" (Chaucer:1083). By then inversion in sentences with constituents in front position as in examples 1.-3. was replaced by the Subject-Verb- Object (SVO) word order more easily in transitive than in intransitive sentences which greatly influenced the occurrence of do in questions as will be explained later in this paper.

In interrogatives with auxiliary verbs the surface result of subject-verb inversion in ME is identical to that of the Modern English subject-auxiliary inversion as in "What kan now faire venus doon above?" (Chaucer: 2663) . Similar to the change in transitive sentences the inversion would appear more readily in sentences where the subject inverts with the main verb than in those where this would happen with an auxiliary verb. Thus, by the end of the sixteenth century main verb inversion is almost lost in questions and the do -form took its place instead. From there it spread to or was adopted in other environments where inversion followed a pre-posed constituent.

Furthermore, with the replacement by the do -form a periphrastic tense carrier has been contributed, that is, a meaningless element that inverts with the subject, serving only grammatical purpose, (c.f. chapter 3.1. on periphrastic do). This connects both cases questions with main verbs and those with auxiliaries. (Kroch et al 1982 .: 283)

Periphrastic do had thus been in use for about a century before it was used in questions. Kroch et al. tried to find out why do was adopted as a semantically empty element in questions. They found out that the frequency of do depends strongly on its syntactical surroundings, namely that its use is favored in transitive sentences as opposed to intransitive ones (1982: 285). Differentiation between the subject and the object noun phrase is the problem with the main verb inversion in transitive sentences because the subject noun phrase is placed after the verb and in that way behaves like the object noun phrase. In intransitive sentences this problem does not occur since the object is clearly separated. Furthermore, do as a tense carrier is readily adopted when the subject is a full noun phrase instead of a pronoun. They argue that through the establishment of the SVO pattern during the late ME period periphrastic do was accepted for the syntax of questions because its use allowed to keep up the SVO order in questions (1982: 286).

If the subject of a sentence is a pronoun such a sentence shows a relatively low tendency to adopt the do -form. Sentences like "What seith she now?" (Chaucer: 2664) and "Wheither seistow this in ernest or in pley?" (Chaucer: 1125) could easily be found until the seventeenth century. Kroch et al. found out that the reason for this maintaining of the verb second pattern in such questions results "from the fact that inverted subject pronouns are enclitics; that is they attach to the verb which precedes them to form a single phonological word with it." (1982: 287). Apparently, enclitic pronoun forms as in seistow are rather seen as inflections on the verb than as independent syntactic elements. Phonologically they are already part of the preceding verb and thus if they are analyzed like this on the syntactic surface there is practically no subject-verb inversion. Since the pronoun subject does not appear as an independent constituent it does not occupy the position of the object noun phrase and in that way transitivity does not matter in such cases. In contrast to Modern German English has eventually lost the possibility to attach the pronoun to the verb. In informal spoken German the pronoun subject is phonologically part of the preceding verb as in hast du > haste; ist er > isser; geht sie > gehtse [my examples, R.Z.].

However, although questions with subject pronouns generally did not use the do -form there is one exception. If there is a construction with a pronoun subject and a pronoun object it is more likely to use periphrastic do than to invert the subject with the verb. Subject-Verb inversion in this case would produce two unstressed syllables as in:

a) Toke ye hym in the quenys chamber? (Kroch et al. 1982: 288)
b) Kan he hem thank? (Chaucer: 3064)

This phenomenon is even more obvious in negative questions with personal pronouns for subject and object. Together with the negative particle not they form a sequence of three unstressed syllables as in "Why lowyste thou not me?" (Kroch et al. 1982: 288). Kroch et al. argue that in such cases speakers willingly adopted the do -form to avoid the production of unhandy words by such phonologically difficult sequences. Later the subject pronouns lost their enclitic statuses to become a free form again and gave the do -form the chance to take over such constructions, as well.

3.3.2. Semantic approach

3.3.2.1. Rhetorical questions

The syntactic approach clearly states that the syntactic and phonological environment influenced the adoption of periphrastic do in questions. Stein (1990: 83), however, argues that its rapid rise in questions is not syntactically but socio- stylistically induced. As was shown above his approach is based on the concept of contrastivity. In the case of questions he separates normal questions from rhetorical questions. Not differentiating between the two would mean to ignore central differences in their pragmatic character. Rhetorical questions are characterized as indirekte Behauptungen which is exactly the opposite of normal questions. Rhetorical questions involve the contrasting of propositions. If they are labeled as (indirect) assertions containing do then the highlighting function of do as explained above includes the contrasting of propositions. For this reason they can be seen as a part of the contrastive meaning of do. Stein suggests that the use of do in direct affirmative questions and direct negative questions (c.f. Figure 1) has, compared to the other uses of do, such a high frequency because they had a "status as rhetorical stock-in- trade of the time" (1990: 83).

3.3.2.2. Undesired endings

However, the language-external factors that triggered the use of do in the sixteenth century can not explain the development of periphrastic do in the following century. Since causative do mainly occurs in the second person context (directive use) the evolution of do in questions in the seventeenth century starts from this context. It is a process with the aim to avoid undesired endings like -st or -edst. A high frequency of do can be detected in the case of the person marked second singular indicative as in thou received'st, thou sing'st. Here the function of do would apparently be to avoid the phonotactically difficult structure. Stein found out that verbs with the second person singular plus thou (in the following 2S) whose stem ends in vowels are three times more likely to be used with the -st -ending than verbs whose stem ends in /t/ or /d/. Thus, he concludes that there must be "a lexical-referential deficit" for the latter ones (1990: 146). According to its frequency the obvious item to compensate this deficit is periphrastic do. In this case periphrasis frequency then can be seen as a functional aspect of phonotactics. Stein supports this with the finding that a verb that is ending in /t, d/ is three times more likely to occur with do than a verb ending in a vowel (1990: 153). Additionally, as far as periphrasis-frequency is concerned, compared to declarative sentences it is on the whole twice as high and the difference between the two large categories - vowel and /t, d/ ending is also bigger in questions. He explains this phenomenon with the greater sensitivity to such undesired difficult phonotactic structures in questions than in declarative sentences. Verbs which end in /t/ or /d/ and in the 2S situation have -st added to them obviously show a word final consonant cluster as in received'st. Since in such environments do -periphrasis is three times more likely than with vowel endings it can be concluded that it is a means to avoid those undesired consonant clusters. Questions are the constructions that react syntactically most sensitive to such consonant clusters. They show an extraordinary differentiation between phonetic contexts from stops over fricatives to vowels (1990: 156).

Generalization of do took place from the 2S context to semantically similar contexts. The context most similar to 2S was the case of second person plural + you. This context was not phonotactically motivated since it did not have the typical 2S endings and thus no consonant clusters to be avoided. The motivation for a spread to this context was the semantic similarity to the originally motivated environment (Stein 1990: 171). This similarity determines the frequency of periphrasis. A cline of directionality would then be:

2S > you > non-you > third person singular present

The sensitivity to the phonotactic factor is highest in yes-no questions. Declarative sentences are least sensitive and wh- questions are in between. Yes-no questions are more advanced in the use of do at any time and in any sub-category because they are the most sensitive towards phonotactic constraints (1990: 192). Together with the lowest degree of sensitivity goes the lowest degree of do -periphrasis. To the contrary, by the second half of the sixteenth century yes-no questions have of all occurrences 60% which are in periphrastic form (c.f. Figure 1). If periphrastic do generalized from 2S to the semantically similar environment of the second person plural and from there to the next similar environment and so on we can assume that semantic similarity is an important feature of directionality, that is, it apparently determines the direction of change. Stein (1990: 200) explains that change in a certain direction occurs according to the markedness or unmarkedness (more weighted or less weighted) of an innovation. In the case of yes-no questions it is a phonologically triggered change (avoidance of undesired endings). Natural change - that is, language internal change - proceeds from more weighted to less weighted. The weighting defines the origin of a development in natural phonological change. For the seventeenth century it can be assumed, then, that the first most weighted context was defined by the use of do as a strategy to avoid the clustering of consonants.

How other uses are connected to the development of do in questions will be shown in the following chapters.

4. Do as a tense-carrier in affirmative sentences

From a syntactic point of view the use of do in affirmative sentences was simply periphrastic (for the use of emphatic do see chapter 6) and had the same meaning as simple present or past tense constructions without it. Examples are found in OE, "Æ ftre d æ m hie dydon æ yþ er ye cyninga ricu settan ye niwu ceastra timbredon" (K. Æ lfred ca.893 , as quoted in the OED 1985: 905, 25.a.). Although there are examples of it in OE manuscripts the use of do as a tense-carrier without meaning was only frequently used in early ME dialects of southern England. In his discussion of Margaret Paston's uses of do Norman Davis (1972) provides some examples of the auxiliary do from late ME, "I hope ye shull be at hom so sone that I woll do wryte nomore tydyngys to yow" (prob. 1451) and "I rood in to my mastres your moder and she dede aske me after my master Berney [...]" (1448 or 1451) It was used frequently from around the year 1500 to ca.1700. However, causative do was still in use at that time (c.f. chapter 3.2 on the causative use of do). The conclusion that it was a meaningless element was derived from the observation that in different manuscripts of the same text certain features were interchangeable as do + tense + verb and verb + tense.

In Early Modern English do as a meaningless element was used already in all types of writing. That it may have been meaningless can be concluded from contemporary comments such as that of Palsgrave (1530) as quoted in Traugott (1972: 138): "I do is a verbe moche comenly used in our tonge to be put byfore other verbes: as it is all one to say 'I do speke...' and 'I speke...'." Another evidence for this conclusion can be traced in Shakespeare's As You Like It (also quoted in Traugott 1972: 139):

Ros.: I'll tell you who Time ambles withal, who Time trots withal, who Time gallops withal and who he stands still withal.

Orl.: I prithee, who doth he trot withal?

Ros.: Marry, he trots hard with [...]

Orl.: Who ambles Time withal?

Ros.: With a priest that lacks Latin.

Orl.: Who doth he gallop withal?

Ros.: With a thief to the gallows...

Here the verbs gallop and trot are used in both variants, with and without do and do not change their meaning. Interestingly, Palsgrave's statement is still valid for many German dialects of today. In these dialects the German equivalent of do, tun, is used to paraphrase a verb without any modification in meaning as shown in the example from the area of Mainz (Erben 1969: 50). The verb is in infinitive form and tun as an operator defines number person and tense:

illustration not visible in this excerpt

By the end of the sixteenth century the use of periphrastic do in affirmative declarative sentences was decreasing. It was seldom used in everyday speech, especially that of the bourgeoisie (c.f. chapter 2.1. about courtly speech).

Nevertheless, it was common until the end of the eighteenth century (Traugott 1972: 139). In 1788 Charles Coote, an eighteenth-century grammarian, thought that this use of do was old-fashioned by that time: "The old English writers frequently used this verb (that is do) as an auxiliary in affirmative sentences, whether an emphasis was required or not...At present we do not use this auxiliary in mere affirmations, unless we wish to lay some stress on what we affirm" (as quoted in Ihalainen 1976: 609).

However, the use of periphrastic do as an unstressed tense marker is still very common in affirmative sentences in the dialect of East Somerset. Apparently it occurs typically in contexts of habitual or characteristic activities as shown in the following example by Ossi Ihalainen (1976: 615).

We did come back then and we did have a glass or two of cider, and then we did go and have a bit of breakfast, come out again and then we did have another drink before we did start off.

5. Negative do

As can be clearly seen in Figure 1 the use of do in negative declarative sentences arises about 50 years after it is being used in questions. The explanation seems to be that negation involves both semantic factors that are connected with the use of do, that is, foregrounding and contrast. These factors initially occurred in yes-no questions and from there spread to similar contexts. Stein (1990: 270) suggests that negatives are also similar in meaning to questions because beliefs can be expressed as sets of propositions. The similarity to yes-no questions consists in the range on the whole of the proposition and the focus on the truth of the proposition. The hypothesis is therefore that the occurrence of do in negatives is due to the partially identical semantics of yes-no questions.

6. Emphatic do

Emphatic do and negation with do are semantically the same thing. Emphatic do has also a contrastive nature. This use is reducible to the same basic components as negation. According to Stein (1990: 272) the first unambiguous uses date from around 1600. The following is a clear example of emphatic do with contrastive semantics: V: Thou art a merry fellow and car'st for nothing. C: Not so, sir, I do care for something, but...I do not care for you (Shakespeare 1599, OED 1985: 906, 25c). It may be characterized as both contrastive and focused on the truth of the proposition. It can be paraphrased as "The proposition is true in spite of what could be expected or what has been asserted to the contrary" (Stein 1990: 273). There is contrastiveness involved in these uses of do. They are based on a pre-suppositional set (1990: 274) , that is; our image of two entities in the shared knowledge of the world or reality who contrast each other. These entities do not necessarily imply a logical contrast but are based on the speaker's assumption of the truth of a proposition. Thus emphatic do in this case is brought about emotionally. Apparently, then, emphatic and negative do are semantically based on the same ground (truth of proposition).

7. Final remarks

The study of variation and its social correlates is only in its beginnings and is almost exclusively limited as yet to studies of present-day linguistic communities. While there is ample evidence that such variation has always existed and has been a major factor in linguistic change, it is almost impossible at the present stage of our knowledge to reconstruct exactly why and under what circumstances a particular change took place in the past. As it is clearly the less well educated who are the chief instrument of language change, and few records are available of their speech, we may in fact never be able to reconstruct the full picture of the conditions under which any one change or set of changes occurred. (Traugott 1972: 13)

The evolution of do can be subsumed under this statement as well. Although there are many attempts to explain the rise and spread of do none of them is able to present the initial trigger - if there are not more than one. Indeed, the development of do seems to be a highly complex topic since there are numerous internal and external factors that influence its direction and existence in different contexts. The arguments that were presented in this paper add plausible explanations. Though none of them can claim to be the primary one.

Syntactic approaches try to explain the rise of periphrastic do with its rise in causative constructions. This is problematic since the two should be successive, then, which is clearly not the case. They refer to word order constraints, that is, subject-verb inversion, in declarative sentences which derived from the inflectional loss during ME and argue that this restriction to SVO word order spread from declarative sentences to questions to keep up the SVO there as well. It is claimed that periphrastic do in questions with pronoun object and pronoun subject is willingly adopted to avoid phonologically difficult sequences. Syntactic approaches could not explain, however, why there was a decrease of the do -form at the end of the sixteenth century for a short time. Since it was clearly a means of linguistic economy why should its use have been reduced? Here, then, a semantic approach is able to explain this phenomenon as an external factor of change, namely the relation of do and courtly speech, the use of which as prestigious speech was disfavored by the public at that time.

Semantic approach would argue that the rapid rise of do in questions was socio- stylistically induced because they had a status as a rhetorical stock-in-trade of the time. Rhetorical questions are characterized as indirect assertions. This relates them to the notion of contrastivity since in assertions do has a highlighting function and includes the contrasting of propositions in which do is used as a marker.

The spread of do in the seventeenth century is due to a fact which syntactic explanation can not account for either- semantic similarity. The high frequency of do that is detected in the person marked second singular context with thou is due to the avoidance of phonologically difficult consonant clusters. This is clearly a language internal reason. The spread to other contexts, however, is due to their semantic similarity to the original context of the do -form. Semantic similarity seems to be an important feature for the direction of change, then. What has not been explained by any approach is the trigger for a language change into a certain direction. Stein (1990: 287) assumes that there may be "a kind of an unconscious, permanent selfmonitoring of language evolution", that is, each individual speaker of a language may have a "knowledge" of the direction of change.

On the whole the presented results, as well as those that have not been mentioned in this paper, seem to me like a small number of fitting pieces of a complex jigsaw which show a finished part but beyond them everything is open to speculation. So I have to admit that the evolution of do is still a confusing issue to me and I am afraid that I am not able to present a general view on this development by now. The approaches from different fields of linguistics seem to highlight one or the other aspect of language change but I could not find a general view that would explain it. Actually, I also have to admit that I felt quite relieved when I found that even Elizabeth Traugott seems to have some problems (c.f. quotation in chapter 2.1., page 4). However, I would prefer a semantic approach for the explanation of change since it takes important extra-linguistic factors into consideration which are very likely to have influenced people and therefore the language of that time. Syntactic change normally is a quite gradual and extended process. Thus it seems very likely that language-external factors were responsible for the rapid rise of periphrastic do. Though Denison (1985: 48) refuses that an arbitrary fact of cultural history like rhyming verse in poetry should have been the motivation for semantic change that produced periphrastic do. He would be happier with linguistic, structural factors. Stein (1990: 141) also admits that "in semantics there is no such thing as 'proving' that a form has a certain meaning."

Since all the evidence presented is by nature based on written texts it might be difficult to elaborate on the spoken language of that particular time. Because there are only rare hints of contemporary scholars as that of Palsgrave (1530), "I do is a verbe moche comenly used in our tonge to be put byfore other verbes: as it is all one to say 'I do speke...' and 'I speke...'." (Traugott 1972: 138), the language change in EME will probably continue to be open to speculation.

References

Davis, Norman. 1972. "Margaret Paston's uses of do."Neuphilologische Mitteilungen 73, 55-62

Denison, David. 1985. "The Origins of Periphrastic Do: Ellegard and Visser reconsidered." In: Roger Eaton et al., eds., Papers from the 4 th International Conference on English Historical Linguistics, Amsterdam, 10-13 th April. Amsterdam: Benjamins, 45-60

Erben, Johannes. 1969. "'Tun' als Hilfsverb im heutigen Deutsch." In: U. Engel et al., eds., Festschrift für Hugo Moser. Düsseldorf: Schwann, 46-52

Graddol, David et al. 1996. English history, diversity and change. London: Routledge

Görlach, Manfred. 1993. Introduction to Early Modern English. Cambridge: CUP

Görlach, Manfred. 1995. New Studies in the History of English. Heidelberg: Universitätsverlag C. Winter

Ihalainen, Ossi. 1976. "Periphrastic do in affirmative sentences in the dialect of East Somerset."Neuphilologische Mitteilungen 77, 608-622

Quirk, Randolph et al. 1985. A Comprehensive Grammar of the English Language. London: Longman

Stein, Dieter. 1990. The Semantics of Syntactic Change. Aspects of the Evolution of 'do' in English. Berlin: Mouton de Gruyter

Traugott, Elizabeth C. 1972. The History of English Syntax. New York: Holt, Rinehart and Winston, Inc.