Periphrastic do in English Witness Depositions 1560 – 1760

This paper exploits the genre of witness depositions for an examination of the development of periphrastic do in the Early Modern English period. Depositions are speech-related texts, reports of authentic speech events related to a court case. Periphrastic do (in contrast to simple V, i.e., without an auxiliary) is investigated in terms of the parameters of time, and sentence type (affirmative, negative, interrogative, and imperative) against the background of previous research. Further potential linguistic and extra-linguistics factors influencing usage that are taken into account include type of verb, and region. The development in the use of the do-construction across time, and as regards the parameter of region, was found to be generally in line with previous research; however, monosyllabic verbs were found to encourage do-periphrasis, which is in contrast to previous findings.


Introduction 1.1 Background
There has been much research on the origin and development of doperiphrasis in the field of historical linguistics (e.g.Ellegård 1953, Nurmi 1999).Previous research has shown that in early modern English, the non-periphrastic V construction, e.g. he knows not, was used alongside the do-construction, e.g. he does not know (Algeo 2010: 178).Much attention has been paid to the questions of when and why auxiliary do became a regular feature in the English language, and earlier studies have used both 'constructed' speech genres such as fiction and 'authentic' speech genres such as trials as sources to find the answers (see Section 3).However, the development of do-periphrasis in witness depositions (henceforth depositions), the focus of this study, has received comparatively little attention in earlier research.Depositions are oral testimonies taken down by a scribe in writing in connection to a legal case and organized according to a specific format (see Section 2.1).It is of interest to investigate this genre further for several reasons.To begin with, the use of do-periphrasis tends to vary according to genre (Nurmi 2000).Furthermore, purportedly recording witnesses' oral testimonies, depositions are 'speech-related' and thus contain certain linguistic and structural characteristics that set them apart from literary genres, as well as from other 'authentic' speech-related genres such as trials (Kytö and Walker 2003).There are two types of court represented in depositions, and I will look at whether this extra-linguistic factor plays a role.In addition, depositions represent different regions, and this has shown to be a relevant factor to examine (cf.Nurmi 2000).Depositions also contain sociolinguistic information such as the deponent's sex, age and occupation, but the inclusion of these variables as possible influencing factors on do usage is beyond the scope of this study.Instead, using An Electronic Text Edition of Depositions 1560-1760 (ETED), this quantitative study investigates the development of periphrastic do in English depositions, focusing on the variables described in Section 1.2.

Aim
The present study aims to explore the use of do-periphrasis in depositions in the early modern English period 1560-1760.In addition to comparing the relative frequency of periphrastic do to non-auxiliary V in different sentence types across time, this study looks at the degree to which linguistic and non-linguistic factors influence the use of dosupport.There are five main questions to be answered in the present study, all of which are inspired by previous research: • How does the frequency of do-periphrasis versus V in depositions change across time in the early modern period?• What is the distribution of do-support in affirmative declaratives, interrogatives, negative declaratives and imperatives?• What possible linguistic factors, e.g.type of verb or adverb, influence the use of do-support?• To what degree does the use of do-support depend on extralinguistic factors, i.e. type of court and region?• To what extent do the results of this study compare with previous research on periphrastic do?Based on the findings of previous research (see Section 3), the hypotheses of the present study are the following: do-support usage in negatives, interrogatives and imperatives is predicted to increase across the period 1560-1760.As far as negative do is concerned, the development of the know group should follow that of the main group, albeit with a lower frequency.In affirmatives, however, the use of the construction is likely to decline by the end of the seventeenth century.Moreover, the level of frequency of negative do in depositions should approximately be the same as in other 'authentic' speech-related genres such as trials.Furthermore, two linguistic factors are likely to influence do usage: firstly, do-support should operate long verbs to a greater extent than short verbs.Secondly, the use of do with an adverbial phrase in affirmatives (do+adv+V) should increase throughout the early modern period.As regards extra-linguistic factors, type of court has shown to play a role in language usage, and this might also be relevant to dosupport usage.Finally, region is expected to be a particularly influential extra-linguistic factor; the do-construction should be favored in the South and less preferred in the North.

Material and methodology 2.1 Material
The primary material used in this investigation is An Electronic Text Edition of Depositions 1560-1760 (ETED).The texts in ETED are divided into 4 different sub-periods: 1560-1599, 1600-1649, 1650-1699 and 1700-1760 (in this study also referred to as periods 1-4).Altogether, there are 905 text files totaling about 267,000 words.It is thus a relatively small corpus, yet large enough to yield sufficient data for my investigation.The information given for each deposition indicates the name of collection, period, deposition date and region, as well as which type of court (ecclesiastical or criminal) the deposition is connected to.In most cases, there is also information about the deponent's sex, age and occupation.However, as mentioned in the introduction, the inclusion of these parameters is beyond the scope of the present study.Therefore, as regards extra-linguistic factors that possibly influence the use of dosupport, this study will focus on region and type of court.
As explained by Grund and Walker (2011: 15), depositions are "oral testimonies taken down in writing by the scribe in connection with a legal case.They detail a person's experience or actions in a particular context pertaining to the case".In addition, depositions adhere to a certain format that makes them different from other genres.Example ( 1) illustrates what a typical criminal court deposition may look like.
(1) <no fol., recto (2)> <Hand 1> As illustrated by (1), a criminal court deposition typically begins with a statement which, among other things, indicates the name of the deponent and the date of the deposition.In ecclesiastical court depositions the initial statement is in Latin, but the information given is similar to that of criminal court depositions (Grund and Walker 2011: 37).As seen in ( 1), the initial statement is followed by the testimony proper in English (also in ecclesiastical court depositions) which contains the scribe's written rendering of the witness's narrative of his or her experience of an event connected to the legal case.However, there are differences between the two types of court that might be relevant to the use of do-periphrasis.It has been found that some "aspects of linguistic and structural variation in the ETED depositions are only readily explainable by reference to how the texts originated or how they were used as part of court procedure" (Kytö, Grund and Walker 2011: 285).For one thing, the fact that Latin is a prominent feature in ecclesiastical depositions but not in criminal depositions might influence do usage.A further, potentially influential, factor has to do with the organization of the testimony proper.In the criminal court depositions the witness's testimony tends to be rendered as one unified narrative, which is similar in structure to oral narratives (Grund and Walker 2011: 26), whereas in the ecclesiastical court depositions the testimony is often given in the form of answers to articles, or a set of questions (Grund and Walker 2011: 36-37).
Offering eye-witness accounts by both men and women, depositions purportedly record 'authentic' speech events.However, it should be kept in mind that as these testimonies were taken down by a scribe in written form, they share certain characteristics, one of which is that a deposition contains different levels of discourse (Grund and Walker 2011).At one level, there is a message from the scribe to the court in which the scribe makes use of formulaic expressions such as "and further this deponent sayeth not" as seen in ( 1).At a further level, there is the message from the deponent to the scribe.Here, the scribe tends to render the deponent's testimony as a third person narrative (Grund and Walker 2011: 45).This is evident in "this deponent was at the howse of Thomas Wolston" in example (1).Lastly, the deponent's citations of earlier speech events may be rendered by the scribe in the form of direct speech as in "yf I do not paye you your money naybo r Tompson", also seen in ( 1).According to Grund and Walker (2011: 45), 8.9 percent of the deposition material consists of direct speech.

Methodology
As mentioned in the introduction, this corpus-based study is a quantitative study.Apart from counting and calculating the number of occurrences of do-periphrasis versus non-auxiliary V as frequencies and percentages, I have made a thorough manual analysis of the examples retrieved from the texts.It must be pointed out that the present study essentially adopts a do versus V approach and therefore mainly discusses the data in terms of relative frequencies.Previous research differs in this respect; for example, Ellegård (1953) uses relative frequencies, whereas Nurmi (1999) mainly uses normalized frequencies per 10,000 words.
The examples of do-support versus simple V were searched for in the corpus with the help of WordSmith Tools (Scott 2012).However, as ETED is untagged, it was not possible to automatically search for all examples of a certain linguistic category in one search.Instead, based on the complete wordlist for the corpus, all non-auxiliary verbs in the present tense and the past tense had to be identified manually.To be certain of finding all examples, each verb, including its variant spellings, had to be searched for separately.The examples were then inserted into an Excel spreadsheet.
Before any results could be analyzed, however, a manual editing of the data in the spreadsheet had to be carried out.This means that unwanted examples, e.g.those constructions containing auxiliary verbs other than do, were identified manually and excluded from the data.Once this was done, in the spreadsheet I manually assigned the examples (both do and simple V) that were left according to sentence type (affirmative, negative, interrogative or imperative), time period, region and type of court.The data filter and pivot table functions in Excel allowed me to calculate the frequencies of do-support versus simple V according to these categories.
The examination of the linguistic factors that might govern the use of do-support was carried out using a quantitative method.This involved analyzing each example of do-support in the data and assigning it to different linguistic features in the spreadsheet, e.g.do serving as an operator of a long or short verb (see Section 3), after which I created pivot tables in order to calculate and compare the frequencies of the different linguistic categories (e.g.long versus short verb, and do+V versus do+adv+V).

The do-construction
Establishing the criteria of do-periphrasis and non-auxiliary V is highly relevant for my results to be comparable with previous findings (see Section 3).There are four main uses of periphrastic do in present-day English according to Quirk et al. (1985: 133-134).Firstly, auxiliary do occurs in negative declarative sentences (He does not know) and negative imperatives (Don't cry!) where not is used as the negative marker.Secondly, the construction is present in structures involving subject-auxiliary inversion.Such structures include questions (Does he know?), clauses beginning with a negative expression (Neither did he see anything), and tag-questions (You know, don't you?).Thirdly, dosupport appears in affirmative statements (He did want to go) and affirmative imperatives (Do come in!) where emphasis is needed.Finally, do is used as a substitute for a complete verb phrase (ellipsis) (He does not know where it is, but I do).In all four usages, do might occur in the simple present or past tense.This investigation is concerned with auxiliary do in the first (negatives), second (interrogatives), and third (affirmatives) usage.In the fourth usage (ellipses), it is difficult to determine what the corresponding simple V construction would be, or whether do was ever optional in such constructions in the early modern period.This is why instances of do preceding ellipses in the corpus have been omitted from the data.Moreover, it should be pointed out that cases where inversion occurs after an initial negative element (Neither did he hear anything), as illustrated in example (2) have also been omitted from the data, because it is not likely that a corresponding simple V construction (Neither heard he/he heard anything) was possible even in the early modern period, at least such instances cannot be found in the corpus.
The criteria of inclusion and exclusion for negative declarative sentences are largely the same as those applied by Nurmi (1999: 142-143); included in my data are clauses containing a lexical verb preceded by an auxiliary do and the negation marker not, as illustrated in example (3), as well as those with non-auxiliary V+not that might have been used with do, as shown in (4).Moreover, the marginal modal used to, illustrated in (5), is included as it can be used with or without do-support.All cases where a clause contains another auxiliary verb have been omitted from the data, and so have the marginal modals dare, need, and ought to.Finally, as Nurmi (1999)  As regards affirmative declarative sentences, I have included instances of verbs in the simple present and past tense that are operated by auxiliary do, as seen in ( 6), or might have been operated by auxiliary do, illustrated in (7).In other words, in order to give an accurate picture of the relative frequencies of the two constructions, each lexical verb that is operated by do is counted as an example of do.This means that cases with only one do may generate two examples of do-support in the data, as illustrated in (8).Had these main verbs occurred without do, they would have been counted as two separate cases of simple V, just like the two verbs in example (9).Included in the category of imperative sentences are clauses which begin with an auxiliary do negated by 'not' followed by the main verb ( 13), or affirmative imperatives as seen in ( 14).Also, included are clauses that could potentially have been used with do, illustrated in (15).Example ( 16), taken from Shakespeare, illustrates that do was not obligatory in negative imperatives in the early modern period.
( Further examples in each sentence type will be shown in Section 4, in which the results of the investigation are presented.Section 3 presents the findings of previous research on the development of do-periphrasis in the early modern period.

Previous research
There is an abundance of research on the use and development of doperiphrasis in the early modern period.Most previous research has focused on the occurrence of do in so-called 'constructed' speech genres (e.g.Ellegård 1953), i.e. fiction, poetry and plays.These contain a great deal of dialogue but are not based on 'authentic' speech-events as are trials or depositions.Using literary works from the early modern period by authors such as Shakespeare, previous research has shown that the process of regulation of periphrastic do occurred from the middle of the sixteenth century onwards (Barber 1997: 196).However, the development differs depending on sentence type.As regards doperiphrasis in affirmatives, the general view is that the use of the construction reaches its peak in the latter half of the sixteenth century, after which a rapid decline takes place in the seventeenth century (Rissanen 1999: 240-42).The occurrence of do in negative sentences increases rapidly in the sixteenth century, and from the seventeenth century on, "its proportionate share increases steadily in comparison with the combination of the simple verb+negative, and the usage is established in the following century" (Rissanen 1999: 245).The conclusions above are to a great extent based on Ellegård (1953), which was the first study to examine the development and regulation of auxiliary do using a relatively large corpus of texts.However, Ellegård's material does not include depositions but consists of literary texts, mainly prose, written by various authors.Given that Ellegård's results are referred to in most research on auxiliary do, it is of interest to compare them with the development of do-periphrasis in depositions.
Figure 1 illustrates the development of do-support in the different sentence types in Ellegård's study.7) As seen in Figure 1, affirmative declaratives have a comparatively low level of frequency.Based on Ellegård's raw figures, the rates are 9.3 percent for 1550-1575, 6.3 percent for 1575-1600 and 3 percent for 1600-1625, which thus shows a steady decline starting from the middle of the sixteenth century (Ellegård 1953: 161).Moreover, Ellegård concludes that compared to affirmatives, the "relative frequency of do is higher in negative sentences and questions all the time, even in the 15 th century" (Ellegård 1953: 161).For negative declarative sentences, Ellegård gives the frequency of auxiliary do versus non-auxiliary V where the use of do goes up in the period 1535-1550 (28.9 percent) and in 1550-1575 (38 percent), and then there is a drop in 1575-1600 (24.4 percent) followed by a rise again for 1600-1650 (37 percent), and the peak (46.9 percent) is reached in the last period 1650-1700 (Ellegård 1953: 161).Ellegård (1953: 199) also calls attention to the frequency of do-periphrasis in negatives in the so-called know group and the main group of verbs, the former of which consists of 9 verbs that are more rarely used with do-periphrasis (see Section 4.2.2).Ellegård concludes that the frequency of do in the know group mirrors the general development of do-periphrasis across time, but with a much lower relative frequency throughout the early modern period; the percentages for do-support range from around 5 to 25 percent (1953: 161, 199).
In order to conform to my own study, I have put Ellegård's figures for affirmative questions and negative questions together into one category, simply interrogatives.In this sentence type in his data there is a rise more straightforward than that of negative declaratives, and with a considerably higher frequency throughout the whole period.The highest frequency is seen in the period 1625-1650 (83.1%).Ellegård also highlights dialectal differentiation in do usage and concludes that the construction was originally used in the South, after which it spread to other regions, and that in the sixteenth and seventeenth centuries, "the do-form continues to be used much less often in the North than elsewhere" (Ellegård 1953: 164).In imperatives, Ellegård's data demonstrate that do usage increases sharply from the mid-seventeenth century and becomes the regular construction in negative imperatives (Ellegård 1953: 162).
A more recent study is Nurmi (1999) which investigates the doconstruction in the Helsinki Corpus and personal letters.The letter material is provided by the Corpus of Early English Correspondence (CEEC) (for details see Nurmi 1999).The findings of Nurmi's study correspond in many respects to those of Ellegård.However, the decline in the use of do-support in affirmative sentences in the Helsinki Corpus comes towards the end of seventeenth century rather than the sixteenth century (Nurmi 1999: 128).As regards do-support in negatives in the Helsinki Corpus, Nurmi establishes that there is a steady rise over time (1999: 146), but when looking at letters, she finds that there is a decline around the latter half of the sixteenth century (1999: 149).With respect to do in the main group and the know group in negatives, Nurmi's results follow those of Ellegård in that there is no great difference in the general trends of development between the two groups, even though the level of frequency is always lower in the know group (Nurmi 1999: 146).Nurmi also pays attention to the relationship between extra-linguistic factors and the use of do-support.With regard to regional variation, Nurmi finds that do in affirmatives is rarely used in the North (1999: 93).
To my knowledge, the development of do in the genre of depositions has not been studied in any depth before.However, some researchers have included samples of deposition material in their investigations.For example, Rissanen (1985) looks at depositions in his study of periphrastic do in affirmatives in early American English and finds that the rate of do occurrence in depositions at times reaches over 12 percent in texts dating back to the 1670s (Rissanen 1985: 176).Furthermore, Culpeper's and Kytö's case study of do-periphrasis in early modern dialogues (2010: 194-98) offers certain insights into the use of do in negative declaratives in depositions and trials.Their data show that the relative frequency of do-periphrasis in trials and depositions ranges from around 13 percent in the period 1560-1599 to 97.3 percent in the period 1720-1760, while the frequency rates in plays, fiction and didactic works range from 25 percent to 79 percent in period 1560-1599 and 1720-1760, respectively (Culpeper and Kytö 2010: 196).
The findings of previous research on do-support described above suggest certain hypotheses as regards general trends in the development of do-periphrasis.The results of Nurmi (1999) and Rissanen (1985) suggest that a marked decline in the use of affirmative do should occur around the latter half of the seventeenth century also in depositions.Furthermore, do-periphrasis in negative declaratives and interrogatives is expected to increase steadily across the early modern period (Ellegård 1953, Nurmi 1999, Culpeper and Kytö 2010).Regarding negative do, the development in the know group and the main group is likely to be similar, but the frequency of the former should be lower than that of the latter in the period 1560-1760 (Ellegård 1953, Nurmi 1999).Moreover, the use of do-periphrasis in the deposition material should correspond to the 'authentic' speech-related group of genres (trials, depositions) rather than the 'constructed' speech-related group of genres (plays, fiction and didactic works) in terms of relative frequency (Culpeper and Kytö 2010).Concerning imperative do, there should be a marked increase from the beginning of the seventeenth century onwards (Ellegård 1953).
Linguistic factors that might influence the use of do-support are discussed in previous research with regard to a number of features, two of which are dealt with in this study.Ellegård (1953) highlights the occurrence of do-periphrasis in affirmatives in connection with adverbs.Among other things, he focuses on do+adv+V (He did often see her), where an adverb is placed between do and the main verb, and do+V (He did see her), in which do is directly followed by the main verb.Ellegård finds that while the use of do+adv+V gradually increases over time, do+V decreases, even though the relative frequency of do+V is always higher than that of do+adv+V (1953: 182).It can be hypothesized that these two constructions should show a similar development in the deposition material.A further linguistic feature of interest is the occurrence of affirmative do with so-called monosyllabic (short) and polysyllabic (long) verbs, as discussed by Rissanen (1985).Based on written records of church meetings in the early American period (1635-1643), Rissanen reaches the conclusion that auxiliary do is more frequently used with long verbs, and that many short high-frequency verbs (e.g.say, think, get) are hardly ever used with do-periphrasis (1985: 175).
Finally, with respect to extra-linguistic factors, both type of court (see Section 2.1) and region might influence do usage.As for region, the do-construction is expected to be particularly common in the South (Ellegård 1953) and less used in the North than in other regions (Ellegård 1953, Nurmi 1999).

Results
In order to demonstrate the development of do-periphrasis in depositions in the period 1560-1760 I will here present the quantitative data across time and sentence type.The results will be compared with the findings of previous research discussed in Section 3. Section 4.1 presents the overall frequency of do-support versus simple V across time.In Section 4.2, I examine the use of do-periphrasis over time in the four sentence types.A summary of the results is given in Section 4.3.

Overall frequency of do-periphrasis over time
This section presents the frequency of do-support versus simple V in the period 1560-1760 and aims to outline the overall development of doperiphrasis in depositions over time.Table 1 illustrates the overall distribution between do and simple V across the four sub-periods.All four sentence types, i.e. affirmatives, negatives, interrogatives and imperatives, are included in the figures, as are examples belonging to the so-called main group and know group of verbs (see Section 3).According to the chi-square test, the difference between the frequency of do-support and the frequency of simple V is statistically significant (χ 2 = 62.5, p = 0.0000, df = 3).Table 1 shows that the total percentage for doperiphrasis is 9.9 percent in the period 1560-1760.It is difficult to compare this result with that of Ellegård (1953), not least because the corpus material that Ellegård uses contains prose texts stretching as far back as to fifteenth century.What is more, Ellegård's corpus is divided into shorter sub-periods than is the deposition material and includes no sub-period for the first half of the eighteenth century as does the deposition material.Nevertheless, if Ellegård's later sub-periods are fused into one period, 1550-1700, and the relative frequency of do is calculated based on the raw figures for the whole period, the result is 7.1 percent for do-support, which is somewhat lower than my results (Ellegård 1953: 159).In terms of general development, the figures in Table 1 reveal that the percentage of do-periphrasis in depositions increases steadily up to period 3 (1650-1699) where it reaches its peak (13.6%).This is followed by a drop to 8.1 percent in the last period (1700-1760).By comparison, Ellegård's results show a decreasing rather than increasing development; the peak can be seen in the period 1550-1600 (9.6 percent), after which there is a drop to 8.5 percent in the period 1600-1650, and then a further drop to 6.2 percent in the period 1650-1700 (1953: 159), which is considerably lower than the 13.6 percent for depositions around the same time.Unlike Ellegård's data, then, my results show a rather even development of the do-construction.It is true that the decline does come in the period 1700-1760, but the percentage still exceeds that of the first period.It must be stressed that the rather low overall frequency of do in depositions throughout the whole early modern period is very much due to the fact that simple V affirmatives are included in data, the most common construction in the corpus.Naturally, this greatly influences the overall ratio of do-support to simple V.
The next section deals with do across time according to the different sentence types and describes more precisely the development of the construction.Also, in Section 4.2.3 the data regarding the development of periphrastic do in negative declaratives in the main group and know group of verbs are presented.

Frequency of do-periphrasis according to sentence type
This section is concerned with the frequency of do-periphrasis versus simple V across the period 1560-1760 in affirmative declaratives, negative declaratives, interrogatives and imperative sentences.The results for each sentence type will be shown in a section of its own (Sections 4.2.1-4.2.4).The figures for these categories are presented in Table 2 and include both the main group and the know group of verbs.Figure 2 further illustrates the development of do in the different types of sentences.It should also be pointed out that with regard to the know and main group of verbs (see Section 4.2.2),I have not intended to compare the raw figures of the two groups but rather the relative frequency of do in each.In addition, the comparison made is for negatives only, for two reasons.Firstly, the simple V construction is expected to be too dominant in affirmatives, so comparative results for this category would not be interesting.Secondly, the number of cases with interrogative and imperative know verbs is too few to be studied separately.
The results in Table 2 and Figure 2 show that there is great variation in the use of do in the different sentence types, which is discussed in Sections 4.2.1-4.2.4 below.The observed difference is statistically significant in affirmatives (χ 2 = 96.3,p = 0.0000, df = 3) and negatives (χ 2 = 94.1, p = 0.0000, df = 3).With regard to interrogatives and imperatives, statistical testing would be unreliable as the expected frequencies are too low in these categories.

Affirmative declaratives
As expected, the results shown in Table 2 reveal that the relative frequency of do-support is considerably lower in affirmatives than in negatives and interrogatives, but exceeds that of imperatives.The percentage of periphrastic do in affirmative declaratives increases from 6.4 percent in period 1 to 10.5 percent in period 2 and stays more or less the same in period 3, after which it drops to around 4.4 percent in period 4, as illustrated in Table 2 and Figure 2. The relatively high percentage for period 3 (10.3%)and the drop in period 4 support the hypothesis suggested in Section 3, that do-support is not expected to decline until the end of the seventeenth century.As discussed in Section 3, this hypothesis is based on Rissanen's study of early modern American English (1985) and Nurmi's (1999) investigation of the Helsinki Corpus.Most cases of affirmative do in the corpus occur in the running testimony proper in which the scribe frames the deponent's account in the third person (see also Section 2.1), as illustrated in (17).However, there are also instances of affirmative do in the form of direct speech in the first person, as illustrated by ( 18).About 84 percent of all instances of affirmative do occurs in the past tense, but as depositions consist of testimony about past speech events, this is hardly surprising.There are other linguistic variables that possibly influence do usage, two of which are discussed in Sections 5.1.1-5.1.2.

Negative declaratives
As is illustrated in Table 2 and Figure 2, do in negative declaratives shows a pattern where the construction gradually increases over the four sub-periods; the percentages are 19.5 percent in period 1, 28.2 percent in period 2, 43.2 percent in period 3 and 75.4 percent in period 4. In other words, negative do in depositions does not become the preferred construction until period 4 (1700-1760).To enable a comparison with Ellegård's data, the last five of Ellegård's sub-periods have been fused into three periods, and the percentage for do-periphrasis in each period has been calculated based on his raw figures.Ellegård's data show 28.6 percent (1550-1600), 33.9 percent (1600-1650) and 45 percent (1650-1700) for do in negative declaratives (Ellegård 1953: 161).In other words, the overall development of periphrastic do in negatives for period 1-3 in depositions corresponds well to Ellegård's data, although the percentage for do in depositions is always somewhat lower than it is in Ellegård's material.As Ellegård's study includes no sub-period that covers the first half of the eighteenth century, no comparison can be made for the last period.
As regards the general trend of negative do, my results also correspond to those of Nurmi (1999) who found that negative do increases gradually across time in the Helsinki Corpus (see Section 3).However, the hypothesis that depositions should correspond to other 'authentic' speech related genres, as discussed in Section 3, does not seem to hold.As my results reveal that the relative frequency of negative do varies between 19.5 and 75.4 percent, it can be argued that depositions are rather similar to the 'constructed' speech-related group of genres in which the rates for negative do range from 25 to 79 percent (Culpeper and Kytö 2010).
Example ( 19) shows an instance of negative do in a deposition connected to an ecclesiastical court.As mentioned in Section 3, previous research has paid attention to the division of lexical verbs into two types, namely main verbs and know verbs (e.g.Ellegård 1953, Nurmi 1999).The know group comprises 9 verbs that are less frequently used with do-periphrasis; know, boot, trow, care, doubt, mistake, fear, skill and list.Of these, know is by far the most frequent one with 137 cases in total in negative sentences followed by care (15 cases in total), while fear in negatives only occurs twice in the corpus, both examples with do-periphrasis.There are no occurrences of the other know group verbs in negatives.The main group of verbs simply refers to those verbs that do not belong to the know group.Table 3 and Figure 3 show the results for the two verb groups.The difference in frequency found in the know group (χ 2 = 39.8, p = 0.0000, df = 3) and the main group (χ 2 = 49.8,p = 0.0000, df = 3) is statistically significant.As illustrated in Table 3 and Figure 3, doperiphrasis with the know group shows a steady development.In Section 3, I hypothesized that the rate for do-support should be higher in the main group than in the know group throughout the whole period (Ellegård 1953, Nurmi 1999).However, the results in Table 3 do not fully support this hypothesis, because even though the construction has a lower relative frequency with the know group than it does with the main group in the first three periods, in the last period the percentage for do is in fact somewhat higher in the know group than it is in the main group.It was also hypothesized that the know group should approximately follow the development of the main group (Ellegård 1953, Nurmi 1999).This trend can be seen in depositions as well, but the change over the first three periods in the main group is more gradual than the development found in the know group; the use of do in the know group increases from 15.2 percent in period 2 to 41.9 percent in period 3.However, it is only in the last period that the percentage goes up considerably in both groups (75.2 and 76.2% in the main group and know group respectively), which suggests that regardless of verb group, the do-construction gains ground over the alternative simple V construction in negatives in the eighteenth century.
It is not unproblematic to contrast depositions with the Helsinki Corpus with regard to the know group as the division into sub-periods differs between the two corpora.However, the data for periods 1-3 in depositions have been fused into two periods, 1560-1649 and 1650-1700, in order to compare with Nurmi's data for the periods 1570-1640 and 1640-1710 in the Helsinki Corpus.The development is illustrated in Table 4.
As Table 4 reveals, the percentages for negative do in the know group are fairly similar in depositions and the Helsinki Corpus.Of course, as the two periods in the two corpora do not match completely, the results may not be fully comparable.Even so, it appears that the development of negative do in the know group in depositions follows the expected trend insofar as the construction increases its relative frequency over time (see Section 3).

Interrogatives
The do-construction in interrogative sentences, including both negative and affirmative questions, clearly dominates over the simple V construction throughout the early modern period; the percentages range from 60 percent to 100 percent.However, as mentioned in Section 4.2, the expected frequencies in Table 2 for interrogatives are too low for the data to be tested for statistical significance.There are, on the whole, relatively few instances in the corpus in this category (89 in total for both do and simple V).In comparison, Ellegård's study reveals a more gradual development for interrogatives; if both negative and affirmative questions are taken into account for the corresponding three sub-periods in his study, the results are 62.2 percent (1550-1600), 74.8 percent (1600-1650), and 81.9 percent (1650-1700) (Ellegård 1953: 161).Therefore the hypothesis that the construction should demonstrate a more or less steady increase in this category, as discussed in Section 3, does not seem to be supported by the data drawn from the deposition material, because the number of cases is too few to show the development with any certainty.Nevertheless, the data indicate that in terms of relative frequency, the do-construction is preferred to the simple V construction in questions.Example (20) shows a typical case where interrogative do in a direct question is used in direct speech.
(20) he the said Barn ~d loked vpon her & h asked her / why dost thou wepe / (ETED: Winchester 1566-1577: F_1SD_Winchester_005) Section 5.1.2sheds more light on the use of interrogative do in connection with short and long verbs.

Imperatives
As regards imperatives, auxiliary do is very rare both in terms of number of cases found in the corpus and relative frequency; in total, there are only 8 cases in the corpus.As mentioned in 4.2, any statistical test of the data in Table 2 for this sentence type would be unreliable.This makes it difficult to draw any conclusions about development over time, even though the percentage goes up to almost 12 percent in the last period.It is, however, reasonable to suggest that auxiliary do is not preferred, given the relatively high number of hits for the alternative simple V construction.In contrast, Ellegård (1953) shows that do is the preferred construction from the mid-seventeenth century onwards (see Section 3).In my data, three of the examples of imperative do are negated by 'not', as illustrated in (21), while the other five are imperative affirmatives, illustrated in ( 22).
(21) don't take any notice, for there is a woman to come in the afternoon (ETED: Norwich 1700-1754: F_4EC_Norwich_016) (22) God will bless thee, and mend thy life; so do my Dear, go # out of the Room, (ETED: Henley 1751: F_4SC_Henley_002) As mentioned in Section 2.1, direct speech is relatively rare in depositions.Little direct speech may explain the low number of instances of imperative do in the corpus, but it does not explain why there is such a great difference in relative frequency between do-periphrasis and the simple V construction in this sentence type.More data would be needed to enable me to reach any conclusions regarding imperatives.

Summary of the results
As expected, do-periphrasis in depositions has a relatively low frequency throughout the early modern period.However, the construction shows a rather even development over time compared to Ellegård's results (1953).Looking more closely at the development in the four sentence types, however, there is much more variation.In affirmatives, there is a notable drop in use of do in the last period (1700-1760), which is some 100 years later than the decline found in Ellegård's material, but coincides approximately with the decline in the data from the Helsinki Corpus.In negative declaratives, the do-construction increases in each sub-period and dominates over the simple V construction by the last period, which was expected based on previous research.Negative do in the know group demonstrates an increasing development, but the change is more dramatic than was expected.In the category of interrogatives, it appears that do-support is more frequent than simple V throughout the early modern period.In this respect, my results correspond to Ellegård's findings.However, the data in this category are not enough to allow reliable statistical tests regarding how do-support usage changed over time.Lastly, while do-support in imperatives shows a steady increase in Ellegård (1953), instances of this category are too rare in depositions to allow any conclusions about the development over time.However, it appears that the alternative construction V is preferred in this category.Section 5 will deal with both linguistic and non-linguistic variables that possibly influence the occurrence of do-support in depositions.

Factors influencing do-support usage
In order to investigate the occurrence of do-periphrasis in depositions further, this section examines selected linguistic factors (Section 5.1) and extra-linguistic factors (Section 5.2) that might have an influence on the use of do-periphrasis.Section 5.3 sums up the results of the main points of Section 5.

Linguistic factors
This section deals with linguistic factors that may influence the use of do-periphrasis in depositions.As stated in Section 3, this investigation mainly pays attention to two linguistic features, both of which have been highlighted in previous research (Ellegård 1953, Rissanen 1985).The first factor to be studied is instances of do-support in affirmatives where a pre-verbal adverbial is placed between do and the main verb (do+adv+V) in contrast to cases where do is directly followed by the main verb (do+V).As mentioned in Section 3, Ellegård (1953) shows that the former type of construction increases its relative frequency over time, and it was hypothesized that this trend would also be found in the deposition material.The results concerning this category are presented in Section 5.1.1.Secondly, as discussed in Section 3, this study investigates the use of do-periphrasis as an operator of so-called long verbs and short verbs (Rissanen 1985), the results of which are presented in Section 5.1.2.

Do-periphrasis with an adverbial phrase
This section deals with the use of do-periphrasis in do+adv+V and do+V in affirmative declaratives in the period 1560-1760.The figures are presented in Table 5.The observed difference between the frequency of do+adv+V and the frequency of do+V in Table 5 is statistically significant (χ 2 = 33.1,p = 0.0000, df = 3).The figures show that the pre-verbal adverbial type (do+adv+V) increases its relative frequency significantly from period 1 (5.8%) to period 2 (16.1%), after which it decreases to 13 percent in period 3.In period 4, the percentage for this type goes up considerably (30.7%).In other words, the relative frequency of do+adv+V increases over time in depositions, as it does in Ellegård's study.In Ellegård's material, however, the relative frequency of the do+adv+V is considerably higher than it is in the deposition material; 17.8 percent in the period 1550-1600, 32.8 percent in the period 1600-1650, and 38.2 percent in the period 1650-1700 (Ellegård 1953: 182, Table 9).Rissanen (1985) does not show the development of affirmative do over time, but rather gives the overall percentage for the different types of do affirmative in different church meeting records from the early American English period.Rissanen concludes that the relative frequency of do+adv+V construction varies between 23 and 38 percent in texts from the period 1630-1650 (1985: 173), which is considerable higher compared to the rate in depositions in the period 1600-1649 (16.1%).
The results show that structures with the pre-verbal adverbial in do affirmatives are by no means a dominant feature in the deposition material.However, the hypothesis that there should be a gradual increase in the use of do+adv+V over time is supported by the results in Table 5. Example (23) shows a typical instance where do is used with an adverbial phrase: (23) this informant doth further say that he heard […] (ETED: Northern 1654-1699: F_3NC_Northern_024)

Do-periphrasis with long and short verbs
This section presents the results with regard to the use of auxiliary do with long and short verbs in each of the four sentence types.Table 6 shows the total frequency of each of the two variants, do+longV and do+shortV.
Excluding the category of imperatives, the observed difference in frequency between do+longV and do+shortV is statistically significant (χ 2 = 14.7 p = 0.001, df = 2).As the number of imperative do is too low (the expected frequency is less than 5), this category cannot be included in the chi-square test.The results in Table 6 show that the relative frequency of do+shortV ranges from 63.9 percent (negatives) to 87.7 percent (interrogatives).Thus do-periphrasis co-occurs more often with short verbs than with long verbs.In contrast, Rissanen (1985: 175) finds that affirmative do in church meeting records from the early American English period is preferred with long verbs (57%) rather than with short verbs (43%).It was suggested that this would be the case for depositions as well (see Section 3), but this hypothesis is not supported by the results in Table 6.Example (24) shows a typical case in the affirmative where doperiphrasis governs a high-frequency short verb.The short verbs that occur most frequently with do-support in the corpus are see, know, hear and go (287 cases in total).In other words, these alone constitute around 38 percent of the 747 instances of do+shortV.This result does not correspond to that of Rissanen who suggests that short high-frequency verbs are rarely used with do-support (1985: 175).

Extra-linguistic factors
In this section, I will investigate to what extent extra-linguistic factors influence do usage.In order to do so, the frequency of do-support versus simple V will be presented according to two parameters, type of court (Section 5.3.1) and region (Section 5.3.2).Section 5.3.3sums up the results for the extra-linguistic parameters.

Type of court
This section compares the results concerning the use of do-support in criminal and ecclesiastical depositions.As discussed in Section 2.1, it was suggested that the type of court might play a role in do usage.Table 7 presents the overall frequency of do versus V according to type of court, while Table 8 shows the frequency of do versus V in the two types of court in affirmatives, negatives, interrogatives and imperatives.The figures do not reveal any striking difference between the two types of court in terms of overall relative frequency of do-support; as seen in Table 7, the rates are 9.4 percent (criminal) and 10.7 percent (ecclesiastical).The two types of courts differ most obviously with respect to interrogatives where the percentages for do-support are 88.2 percent (criminal) and 73.7 percent (ecclesiastical).As regards affirmative do, the difference between criminal and ecclesiastical depositions is small; the rates are 7 percent (criminal) and 8.4 percent (ecclesiastical).In negatives, the difference between the two courts is even smaller; the relative frequency of do-periphrasis is 42 percent and 41.6 percent in criminal depositions and ecclesiastical depositions, respectively.The data for affirmatives show a significant difference in distribution (χ 2 = 6.088 p = 0.00136, df = 1), but not the data for negatives (χ 2 = 0.000 p = 0.986, df = 1) and interrogatives (χ 2 = 2.218 p = 0.1364, df = 1).Concerning imperatives, the expected frequencies are too low to allow statistical testing.These results suggest that type of court is not an influencing factor on do-support usage.

Region
In order to determine to what extent the parameter of region influences the use of do-support, this section examines the frequency of do versus V in each of the five regions represented in ETED (North, South, East, West and London).On the basis of previous research, do-periphrasis is expected to be less common in the North and more frequently used in the South than in other regions (see Section 3).The overall distribution of do versus V in the five regions is presented in Table 9, while Table 10 presents the figures for do and V in each sentence type according to the different regions.The difference in frequency between do-periphrasis and simple V in Table 9 (χ 2 = 28.2p = 0.000, df = 4) is statistically significant.As regards sentence type, the difference in Table 10 is statistically significant in affirmatives (χ 2 = 29.9,p = 0.0000, df = 4) and negatives (χ 2 = 34.9,p = 0.0000, df = 4).No statistical testing can be performed for interrogatives and imperatives as the expected frequencies are below 5. Table 9 shows that overall relative frequency of do-support is practically the same in the East, South, West and London (ranging from only 10.6 to 11.4%), while the rate for do in the North is somewhat lower, 7.5 percent.As seen in Table 10, the percentages of do with regard to affirmatives are approximately the same in the East, West, South and London (7.8-9.3%)whereas the rate for do in the North is only 5.4 percent.Moreover, the percentage of negative do is considerably lower in the North (31.8 percent) than it is in the other regions.The regions with the highest rates of do-support in negatives are the South (66.7%) and London (54.3%).As the difference found in the data for interrogatives and imperatives cannot be tested for significance, no definite conclusions can be drawn.However, also in these categories the trend is that do-support is relatively rare in the North.My data show that the use of do seems to be preferred in the South and London and suggest that regional variation plays an important role in the use of do-support.Thus, the results of this study seem to correspond to those of Nurmi (1999) in that the do-periphrasis in affirmatives is more rarely used in the North.In addition, the results are in line with Ellegård's findings (1953: 164) insofar as the do-support in negatives and interrogatives is most common in the South.

Summary of influencing factors
The results in Section 5 showed that as regards linguistic factors, the use of an adverbial phrase with do-periphrasis in affirmatives (do+adv+V) increases over time, but this construction is never preferred to the do+V construction.Moreover, contrary to what previous research has found (Rissanen 1985), it is clearly the case that do is favored with short verbs rather than with long verbs, at least in affirmatives, negatives and interrogatives.Thus this appears to be the most influential linguistic factor on the use of do-support.As regards extra-linguistic factors, there is no evidence to suggest that the parameter of court influences do usage.With respect to region, the expected frequencies in interrogatives and imperatives are too low to allow any conclusions, yet this parameter does appear to be an extra-linguistic factor that influences the use of doperiphrasis; the construction in negative declaratives is clearly preferred in the South and London, while it is more rarely used in the North with regard to both affirmatives and negatives, a result that corresponds to that of previous research (Nurmi 1999, Ellegård 1953).

Conclusion
This corpus-based study of periphrastic do in depositions in the period 1560-1760 has shown that the construction undergoes a change in usage over time which in many respects corresponds to the results of previous research.My results provided evidence for the hypothesis that do usage in negative declaratives would increase its relative frequency across the period 1560-1760 and be clearly preferred to the simple V construction by the last sub-period (Ellegård 1953, Nurmi 1999).In addition, as expected, it was found that do-support usage in affirmatives increases up to the beginning of the seventeenth century while the decline comes towards the end of the seventeenth century (Nurmi 1999;Rissanen 1985).In interrogatives, do seems to be preferred to the simple V construction throughout the early modern period, which is in line with previous findings (Ellegård 1953), although there are not enough data to provide reliable statistical results concerning the development over time.
Unlike previous findings (Culpeper and Kytö 2010), however, this investigation shows that depositions correspond to 'constructed' speechrelated genres (e.g.fiction) rather than to 'authentic' speech-related genres (e.g.trials) as regards do usage in negatives.Furthermore, with respect to imperatives, the hypothesis that do-periphrasis would be increasingly common (Ellegård 1953) could not be confirmed, as the construction in this sentence type is too rare in the corpus.As for do usage in the know group, my results do not fully correspond to those of previous research as the development is more marked than expected (Ellegård 1953, Nurmi 1999).
When it comes to the influence of linguistic factors, it was concluded that the occurrence of an adverbial phrase with do (do+adv+V) in affirmatives increases over time but is never favored over the construction without an adverbial phrase (do+V), which was predicted (Ellegård 1953).A much more influential linguistic factor is that dosupport tends to co-occur more frequently with short verbs than with long verbs (excluding the category of imperatives, for which there are insufficient data), a result not in line with previous findings (Rissanen 1985).
Concerning the influence of extra-linguistic factors, type of court plays no major role in do usage, but region does; as previous research has shown, the construction is most frequent in the South and London and less common in the North.This is evident in the case of affirmatives and negatives, a result that corresponds to previous research (Ellegård 1953, Nurmi 1999), while the relatively low data for interrogatives and imperatives are not sufficient to allow firm conclusions.This investigation has not included sociolinguistic factors such as sex, rank and occupation, but further studies would reveal to what degree these variables influence the use of do-periphrasis in depositions.

(
24) she this Deponant did then see the said John Sangar.(ETED: Somerset 1706-1716: F_4WC_Somerset_021) Thoms Algood of Witon in # the Com ~ of Norff husbondeman of thage of xliiij ti yeares Sworne and examined the daye and yeare Aboue written saythe / That vpon the Sondaye after Twelfe Daye laste paste this deponent was at the howse of Thomas Wolston and at that tyme one Willm Tompson of Welles was there and Requyred a certeyne dett of hym {Wolston} / And # Wolston sayde vnto Tompson y t yf I do not paye you your money naybo r Tompson to morrowe That then I wyll forffett all my nete vnto you and further this deponent sayeth not / Be me Thomas allgood (ETED: Norwich 1560-1566: F_1EC_NorwichA_003)

Table 1 .
Frequency of do-support versus simple V in depositions across time (raw figures and percentages)

Table 2 .
Frequency of do-periphrasis versus simple V in depositions across time according to sentence type (raw figures and percentages)

Table 3 .
Frequency of do-periphrasis versus simple V in negative declaratives in depositions: main group vs. know group of verbs (raw figures and percentages) Figure 3. Relative frequency of do-periphrasis in negative declaratives in depositions: main group vs. know group of verbs (percentages)

Table 4 .
Frequency of do-periphrasis versus simple V in negatives in the know group: depositions (based on Table2) and the Helsinki Corpus(after Nurmi  1999: 146, Table 9.1)

Table 5 .
Frequency of do+adv-V versus do+V in affirmative declaratives in depositions (raw figures and percentages)

Table 7 .
Frequency of do-support versus simple V in depositions according to type of court (raw figures and percentages)

Table 8 .
Frequency of do-support versus simple V in depositions according to type of court in each sentence type (raw figures and percentages)

Table 10 .
Frequency of do-support versus simple V in depositions in each sentence type according to region (raw figures and percentages)