Meta-Communicative Expressions and Situational Variation of Stance Marking : I say and I tell ( you ) in Early Modern English Dialogues

In this study, I compare the functions of I say and I tell (you) in the different text types in A Corpus of English Dialogues 1560–1760 (CED). The text types differ with respect to a number of parameters, such as the authenticity of the dialogue, the stability of the participant roles, the formality of the setting, and the dominant verbal activities. I show that the two expressions are used with a range of different functions, most of which express the speaker’s stance, and I argue that the differences in function are related to differences in the communicative setting and the roles of the participants.


Introduction
There are many ways in which speakers and writers can express an attitude towards what they say.This includes meta-communicative expressions, i.e. expressions that refer to the communicative event or process in which they occur.I say and I tell (you) are two metacommunicative expressions that are used quite frequently in Early Modern English dialogues.From a semantic point of view, they clearly relate to speaking and, therefore, they can be expected to be typical of speech-related texts.When looking at some instances in context, as in examples (1)-( 4), two aspects become clear.First, the expressions occur in a variety of constructions with various degrees of integration with cooccurring clauses, including negated instances and instances that are modified by modal verbs.Second, they tend to express commitment to 1 I would like to thank the anonymous reviewer, the editors of this special issue, and the audience at the 18 th International Conference on English Historical Linguistics in Leuven for their valuable comments.I would also like to acknowledge the financial support of the University of Zurich in the form of a Forschungskredit research grant (grant no.FK-14-074).
what is said, which makes them relevant for the study of epistemic stance.
(1) Mr. Com.Serj.: I tell you I caught hold of him, and the Rabble got him from me.(Trials, Tryal ... of Tho. Pilkington, 1682/1683) (2) Hon.: [...] Take heed I say, for if I catch you once, Your bodies shall be meat for Crowes [...] (Drama Comedy, A Knacke To Knowe a Knaue, 1594) (3) William Dawes sworn.:All that I can say is this; William Turner was at my house on Wednesday was seven-night last; comes in, and called for a Flaggon of Beer; there comes in a Customer of mine, sayes he to me, What do you with such a Fellow in your house?have a care of him, he is a dangerous fellow; presently after Col. Turner came in and paid for a Flaggon of Beer, and went away.(Trials, Tryal … of Col. Iames Turner, 1663) (4) Ld.T: How! Where?When?Ldy T.: That I can't tell; nay, I don't say there was --I am willing to believe as favourably of my Nephew as I can.
(Drama Comedy, The Double-Dealer, 1694) Both stance and meta-communication have attracted the interest of researchers in recent years.Meta-communication has been studied particularly well in the context of academic writing (e.g.Hyland 2000;Ifantidou 2005).Concerning historical data, the settings that have been analysed with respect to stance and meta-communication (although not in all cases covering and combining both concepts) include legal discourse (e.g.Grund 2012, 2013, and in the present volume), medical writing (e.g.Taavitsainen 2000;Taavitsainen and Hiltunen 2012), religious discourse (e.g.Boggel 2004Boggel , 2009) ) and correspondence (e.g.Dossena 2012;Fitzmaurice 2003Fitzmaurice , 2012)).These studies investigate texts from one setting in great detail and their findings suggest that the dominant strategies found in these texts are related to the setting in which the text is produced.For instance, Taavitsainen (2000) relates diachronic changes in the use of meta-discourse to overall changes in scientific thought styles, and Grund (2013) points out that the use of evidential strategies in witness depositions is directly related to the witnesses' need to be perceived as reliable and credible.Comparisons across different historical settings are not easy and have less frequently been made.
Because stance expressions are context-dependent, the range of expressions that are used differs across settings, which poses problems for a direct comparison of detailed qualitative findings.In contrast, purely quantitative evaluations of a large number of stance terms (e.g.Biber 2004) provide only limited insight into how the function of these expressions depends on context.
The aims of this study are two-fold.The first aim is to investigate the functions of I say and I tell (you) in Early Modern English dialogues, based on A Corpus of English Dialogues 1560-1760 (CED).I will analyse the extent to which the functions depend on different structural realisations and on the context in which they are used, and I will also look at similarities and differences between uses with say and uses with tell.In particular, I am interested in how the functions of these expressions relate to stance.This leads me to the second aim of this study, which is to assess what can be learned about the contextdependency of stance marking by looking at meta-communicative expressions.Meta-communicative expressions explicitly refer to aspects of communication and, thus, they are highly dependent on the communicative situation.As such, they are promising candidates for the study of situational variation of stance marking and may help increase our understanding of how stance depends on context.

Meta-communication and stance marking
Meta-communication, as used in this study, can be defined as "communication about (selected aspects of) communication" (Hübler 2011: 108), and is sometimes also referred to as meta-discourse (see, for instance, Boggel 2009;Hyland 2000Hyland , 2005;;Taavitsainen 2000).Metacommunication takes place whenever participants in a communicative situation refer to the communicative activities in which they are engaged.Depending on the function and the communicative constellation, various subtypes of meta-communication can be distinguished.Following Hyland (2000: 110-113), a first distinction is often drawn between textual and interpersonal meta-communication. 2 Textual meta-communication is used to structure and organise a text, to increase cohesion and to make it more accessible.Hyland (2000: 111) gives examples from academic writing, which include logical connectives (e.g. in addition), frame markers (e.g.finally, to repeat), endophoric markers (e.g.noted above, see Fig.), evidentials that mark references to other texts (e.g. according to X), and code glosses (e.g.namely, such as).In contrast, interpersonal meta-communication expresses the attitude of authors and speakers towards the information they present.Hyland's examples from academic writing include hedges (e.g.perhaps), boosters (e.g.definitely), attitude markers (e.g.unfortunately), markers that refer to or build a relationship with the addressee (e.g.you can see) and person markers referring to the author (e.g.I, our).
From the point of view of stance marking, interpersonal metacommunication is of special interest.First person references often mark speakers or writers explicitly as source of information, thus expressing evidential stance.Hedges and boosters are used to express different attitudes with respect to the reliability of the information, and as such they contribute to epistemic stance.I say and I tell (you) contain explicit self-references to the speaker and they can be used as boosters.The emphatic quality of I say has been noted by Goossens (1982: 95) and Brinton (2008: 77), and it has been identified as one of the frequent markers of authorial stance in Early Modern English religious texts (Boggel 2009: 183).
The second distinction of different types of meta-communication depends on the communicative context.In non-dialogic settings, metacommunication is usually directed towards an audience that is not present, i.e. a text-external addressee.Scientific texts, for instance, may contain instances of meta-communication that instruct readers how the text has to be read and understood.Many historical studies on metacommunication focus more or less exclusively on this aspect of metacommunication (e.g.Boggel 2004Boggel , 2009;;Dossena 2012;Taavitsainen 2000).In contrast, meta-communication directed towards a text-internal addressee can take place in dialogic settings when interactants discuss their communicative behaviour, for instance for the purpose of clarifying miscommunication.At least as far as historical texts are concerned, this intra-textual meta-discourse.My study deals only with intra-textual metadiscourse.
aspect has received less attention in the study of meta-communication, but it is included sometimes (see, for instance, Simon-Vandenbergen and Defour 2012).In this study, I will look at meta-communication in dialogic texts and I will only focus on text-internal addressees.

Data and method
The data used in this study come from A Corpus of English Dialogues 1560-1760 (CED).The CED contains roughly 1.2 million words of speech-related texts from five text types: drama comedy, didactic works, prose fiction, witness depositions, and trial proceedings (see Kytö and Walker 2006;Culpeper and Kytö 2010).The texts are from the Early Modern period.The corpus contains meta-textual information for all text files and the text is annotated so that passages of direct speech can be distinguished from the rest of the text.
As a first step in my analysis, all instances of I say and I tell (you) were tagged semi-automatically in an XML version of the corpus.The tagging included spelling variants, like I saie, and, for I tell (you) forms with the pronoun thou (i.e.I tell thee).Only the present simple forms were tagged, thus excluding forms like I said and I am telling you.Using regular expressions, all instances with a maximum of four intervening words between I and say/tell were tagged.This resulted in 1,289 hits, roughly half from each verb.Following this, all instances underwent detailed manual analysis.This included the sorting out of wrong hits and hits that were not relevant for this analysis.Examples of instances that were excluded are cases in which I was not the subject of say or tell (e.g.I've heard say), past tense uses (e.g.I did not tell) and interrogatives (e.g.what shall I say?).After removing these instances, 1,138 hits remained, exactly half of which were cases with say and half with tell.Except for 14 hits all occurred within text marked as direct speech.During the manual analysis, additional information about each instance was added to the XML markup.This included the annotation of structural information, i.e. whether the instance was negated and/or modified by modal verbs and, for unmodified instances, whether it occurred as a parenthetical or in a matrix clause with following thatcomplement clause.For the group of parentheticals, only instances in which the construction was used sentence-medially or sentence-finally were considered.
Sentence-initial instances without overt complementiser are in most cases structurally indeterminate and could be analysed either as parenthetical or as matrix clause with following complement clause with zero complementiser (Brinton 2008: 12).All these sentence-initial uses were included into the matrix clause category.Table 1 gives an overview and examples of each of the categories.These examples provide a first illustration of the fact that the different structural types express stance to varying degrees.As I will further show in Section 5, practically all uses of I say and I tell (you) from the categories of parentheticals and matrix clauses followed by thatcomplement clause express epistemic and evidential stance.In contrast, the uses in the group 'other' are very varied, and many of these instances are not relevant from the point of view of stance.Finally, the modified uses are difficult to generalise, since the meaning depends on the modifier to a considerable extent.I will discuss this group in more detail in Section 5.2.

I say and I tell (you) across CED text types
Before I take a closer look at their functions, I will present an overview of the frequency of I say and I tell (you) across various text types in the CED.Tables 2 and 3 present the frequency of all instances that occur within text marked as direct speech, thus excluding the 14 instances occurring in passages other than direct speech.The frequencies are normalised on the basis of all words occurring in direct speech in the respective text type.Figures 1 and 2 present a visualisation of the same results, excluding the miscellaneous group of texts, which is very heterogeneous and, therefore, more difficult to interpret.
The results show striking parallels for I say and I tell (you).The overall frequency is exactly the same (62.9instances per 100,000 words).For both verbs, there is strong variation in the frequency across the different text types, and the distribution is similar; trial proceedings are the text type with the highest overall frequency, followed by drama comedy.Concerning structural types, modified instances are more frequent than unmodified instances for both verbs.For I tell (you), this holds for all text types except the mixed group of texts labelled 'miscellaneous', whereas for I say some text types have a higher frequency of unmodified instances.Particularly noteworthy is drama comedy, which has a high frequency of parenthetical I say -twice as high as the average of parenthetical I say across the corpus.The group 'other', which is of least relevance from the point of view of stance marking, contributes about 10% of all instances, 67 (12%) in the case of I say and 41 (7%) for I tell (you).Concerning differences between the two verbs, the most notable difference is the higher frequency of modified constructions for I tell (you) compared to I say.This is true for the corpus overall, but also for all of the text types except trial proceedings.There, I say has a higher frequency for almost all structural types, including modified constructions.In contrast, unmodified constructions are generally less frequent for I tell (you) than for I say, and the difference is particularly strong for parenthetical constructions.In order to see how the functions of I say and I tell (you) relate to the context and the communicative situation, it is necessary to have a closer look at the main context parameters of trial proceedings and drama comedy.Table 4 provides an overview of these parameters.Perhaps the most obvious difference is that drama comedy texts are fictional and thus contain constructed dialogue, whereas trial proceedings contain authentic dialogue that was transcribed by a scribe (for a detailed discussion, see Kytö and Walker 2003;Walker 2007: 14-17).In addition, the two settings differ considerably with respect to participant constellation and formality.The trial setting is formal and given that the outcome of a trial was often literally a matter of life and death for the defendants, the situation was very serious.The participant roles in a trial are fixed and the power is distributed asymmetrically between the communicative participants, with judges and court officials being in charge, while defendants and witnesses have to comply.In contrast, the settings in drama comedy are much more varied and the power distribution between the participants can change during the interaction.The difference in setting can also be seen in the different verbal activities that are dominant.For trials, the main activities are questioning and testifying, and each of these activities is clearly assigned to different parties.In drama comedy, it is more difficult to identify dominant verbal activities.Activities that can frequently be found include arguing, fighting, joking and teasing, but they are by no means exclusive.These parameters can influence language use and the function of linguistic expressions.Following Levinson (1979: 370) it is clear that activities like joking and testifying constrain verbal contributions, their functions, and interpretations quite directly.Expressions that are perfectly acceptable in the context of joking can be inappropriate and thus unlikely to occur in the context of giving testimony.Alternatively, the same utterance made when testifying can have a completely different meaning in the context of a joke.Formality constrains expressions in a similar way.In addition to suppressing colloquial expressions, very formal settings are often characterised by specialised meanings and formulaicity.Verbal activities, formality and participant roles are related to each other in systematic ways, and authenticity is related to these other factors, too, given that we do not have a choice between authentic and constructed data for the different settings.This means that it is often not possible to single out one of these parameters as the responsible factor behind differences in linguistic expressions and their functions.
I will start the analysis of I say and I tell (you) with a discussion of unmodified parentheticals and unmodified matrix clauses followed by that-complement clauses.In many ways, these instances are easiest to analyse, since they are more stable in meaning than the other uses.In Section 5.2, I will then take a closer look at the uses in which I say and I tell (you) are modified by modal verbs.By differentiating between the modals that are used, I will show that there are clear parallels with the results from the first part of the analysis.The uses with other constructions will not be included in the detailed analysis.They are very heterogeneous and not all of them are relevant for the expression of stance.
5.1 Main functions of parentheticals and matrix clauses followed by thatcomplement clauses I restrict my discussion of unmodified instances of I say and I tell (you) to functions occurring more than once.Since the analysis is based on a relatively small set of instances, the classification cannot be tested for reliability.As a consequence, I do not provide exact frequencies for each function.Nevertheless, it is still possible to see differences and similarities between the uses of the two constructions by looking at the main functions with which they are used.Table 5 lists the functions that occur repeatedly in trial proceedings, including examples of each function.
By far the most frequent use of I say in trial proceedings is to mark a statement as official testimony.Three examples of this function are given under A in Table 5.Overall, 20 instances of I say in trial proceedings fall into this group.Often, I say is used to introduce facts about the identity of the speaker, as is the case in the first example given in Table 5.In other cases, I say precedes information that is given as part of the testimony.The second example is quite typical of this use.The witness uses I say before providing information as facts.There are only a few instances in which speakers explain how certain they are about the information they provide.The third example is one of these instances.
This first use of I say is very specific to the legal setting of trial proceedings.The expression appears to acquire a specialised meaning in which say does not just stand for 'uttering aloud' but for committing the speaker to the truthfulness of what is said.Speakers on the witness stand want to appear credible, and thus they want to present their statements as having a high degree of reliability.In the examples listed under A in Table 5 the speakers appear to use I say in order to claim such a high degree of reliability for their statements and, as a consequence, a high degree of credibility for themselves.The fact that I say can often be found at the beginning of statements by witnesses and defendants suggests a certain degree of formulaicity that is owed to the formal procedures at court.However, it is important to note that I say is not obligatory and fixed; sometimes I say occurs only in later turns and sometimes it is not used at all.(Trials, Triall, of Lieut. Collonell John Lilburne, 1649) All the other functions of I say do not occur more than two or three times each in the trial proceedings of the CED.The examples given in Table 5 illustrate their use.Function B is closely related to the first function.The uses in this group also occur in turns by defendants and witnesses and I say is again used to emphasise the fact that the statement that follows is the speaker's official testimony.What differentiates this use from A is the fact that it occurs in a context in which the speaker feels that his or her statement has been misunderstood and needs clarification.I say has an even clearer meta-communicative character here compared to A; it helps differentiate between what was said (and intended) and what was not said.In contrast to the first two functions, functions C and D occur in turns by court officials.In C I say is used to insist that a previously posed question should be answered completely.The original question is repeated, followed by I say.Finally, D refers to uses in which court officials instruct witnesses or defendants about the rules and proceedings of the court.
All these uses are clearly tied to specific participant roles of the communicative situation.While the first two uses occur in the turns of witnesses and defendants, the last two uses occur in the turns of judges and other court officials.This is related to the different communicative activities of the participants.What all four uses have in common is that they express a strong commitment to the statement with which they are used.Witnesses and defendants commit to the truth of their statement with functions A and B, while court officials insist on their power and the illucutionary force of their statements with functions C and D. Irrespective of role, all uses clearly help establish the stance of the speaker.
There were only eleven relevant instances of I tell (you) in trial proceedings, which provides a very small basis for investigating the functions of the construction.However, what is noteworthy is that two of the functions that were identified for I say can also be found for I tell (you).There are three instances in which I tell (you) refers to testifying and which are very similar to function A. In contrast to the uses of I say, the uses of I tell (you) occur in contexts in which the same or very similar information has already been provided earlier, so that I tell (you) in these data has a stronger connotation of confirming previous statements than I say.The second function for I tell (you) is instructing witnesses and defendants.There are five instances of this function in my data, and they are very similar to function D for I say.The remaining instances have functions which do not occur repeatedly.
The functions of I say and I tell (you) in drama comedy show differences to those found in trial proceedings.By far the most frequent function of I say is the one listed as G in Table 6.This use occurs together with imperatives and it serves as a booster for the speaker's request of the addressee.About one third of all instances of I say in drama comedy fall into this group.The next two functions, H and I, are closely related to each other.In both cases, speakers use I say when expressing their personal opinion or point of view.What distinguishes the two is that in H I say is used to emphasise information that has been provided before.Often, this occurs in the context of arguments in which both parties insist on their point and in several cases, I say occurs in two consecutive turns, as in the example given in Table 6.This use has some parallels with one of the functions found for trial proceedings, function C, used by court officials when insisting that a question should be answered.The example used to illustrate function I is one of the few cases in which I say occurs in a context in which the degree of commitment to the truth of the statement is hedged.The insertion (for my part) signals that the speaker is going to present his own opinion, with which others might disagree.The last function, J, refers to cases in which speakers use I say to introduce a decision, especially as to whether or not they are complying with a request by the addressee.Like most of the other functions this was not found in trial proceedings.However, there is a certain similarity between this use and use A, testifying, in that both mark that the following statement is binding, be it as official testimony in a court trial or as mutual agreement between the interactants in a drama comedy.complement clause constructions are highly relevant for the study of stance.Second, the functions of I say and I tell (you) show considerable similarities, which tend to be stronger than the similarities of each construction across the two settings.Indeed, in many cases, I say and I tell (you) appear to be more or less interchangeable with respect to their meaning. 3Finally, the context plays an important role for the functions with which I say and I tell (you) are used.Both constructions tend to be used with somewhat different functions in the two settings, and there is more homogeneity in the functions found in trial proceedings than in those found in drama comedy.As discussed above, it is difficult to identify which factors are mainly responsible for the variation.However, the results suggest that participant roles and activity types may be the most important factors.Trial proceedings contain only a limited number of participant roles, each of which has clearly associated activities.This corresponds with the high degree of homogeneity in the observed functions.Moreover, some functions in trial proceedings are used almost exclusively by participants who are being questioned, while other functions are used only by participants who are in power.In contrast, drama comedy contains more varied roles and activity types, which might explain why the functions of the two expressions are more varied.
A closer analysis of the participant roles in other settings and their relation to the functions of meta-communicative expressions would be an interesting starting point for further research.

Frequent constructions with modal verbs
I will now briefly turn to those instances of I say and I tell (you) that are modified by modal verbs.Table 7 gives an overview of the constructions I found, ordered by frequency.Constructions that were observed less than three times are not listed, but the total number of examples this applies to is given in Table 7 at the end of each list.The information in Table 7 shows that the marked differences in frequency between trial proceedings and drama comedy are mainly due to constructions containing the modal can.For say there are 108 instances of I cannot say or I can't say and 32 instances of affirmative I can say.Together, they account for 75 percent of all modified instances of I say in trial proceedings.This is very similar for I tell (you), where 74 percent of all modified instances are cases with cannot, can't or can.The dominance of I cannot say and I cannot tell in trial proceedings can also be seen in the results of a study by Culpeper and Kytö (2010: 116-117).Their analysis of lexical bundles in the CED identified I cannot say and I cannot tell as two of the twenty most frequent lexical bundles in trial proceedings.The remaining modal verbs play a minor role compared to this, and their distribution does not show striking differences across the two settings.Most noteworthy regarding these other modal verbs is, perhaps, that for both I say and I tell (you) only drama comedy has at least three instances with may.
A more detailed study of the uses with cannot in trial proceedings shows that they are used in a very consistent manner, illustrated in examples ( 5)- (7).Again, the uses are to some extent formulaic and they are clearly related to participant roles and activity types.They occur in turns by witnesses defendants and serve to indicate that the speaker does not have sufficient evidence for making a statement.In some cases, the entire turn consists only of I cannot say or I cannot tell, as in the first and third instance in example (5).In other cases, the speaker provides explanations why he or she cannot provide any further information, as in examples ( 6) and ( 7).The affirmative I can say is used to mark that the statement represents the full extent of information available to the speaker, as can be seen in examples ( 8) and ( 9).This is made especially clear by the formulation all I can say used in example ( 9), which can be found repeatedly.considerably more frequent than unmodified I say and I tell, at least for trial proceedings.This means that restricting the analysis to unmodified instances only would result in a very limited view of I say and I tell (you) in these texts.

Conclusion
The findings of this study have shown that both I say and I tell (you) often express stance in Early Modern English dialogues from trial proceedings and drama comedy.Unmodified parentheticals and matrix clauses with that-complement clauses usually express emphasis and a strong commitment to the truth of the proposition.The most frequent modified constructions contain the modal can and these instances are often used to express the extent of information available to the speaker, especially in trial proceedings.Together, modified instances with can and unmodified parentheticals and matrix clauses with that-complement clauses-all of which tend to be directly relevant to stance-make up the large majority of instances in trial proceedings and drama comedy.This means that I say and I tell (you) are relevant for the study of stance in these texts.
The analysis provides various types of evidence for the dependency on the situational context of the two expressions.The quantitative evaluation showed clear variation across the different text types in the corpus.The detailed study of the functions identified differences between trial proceedings and drama comedy and it is noteworthy that the differences between trial proceedings and drama comedy were overall more marked than the differences between I say and I tell (you).To some extent, these differences can be viewed as text-type differences and formulaicity plays a certain role.However, participant roles and activity types appear to be important too.In trial proceedings it becomes particularly clear how different functions correlate with different roles and verbal activities.It would be interesting to extend this study to other data in order to see whether the same functions of I say and I tell (you) can be found in other text types that contain the same or similar verbal activities.
Concerning the suitability of the two expressions for the study of context-dependency of stance marking, some limitations have to be mentioned.The functions vary across the different constructions in which they occur.This makes it necessary to differentiate between the constructions, which cannot be done without manual analysis.In addition, the classification of the functions is not always straightforward and relies on interpretation that is subjective to some extent.Last but not least, it is clear that it is not possible to achieve a comprehensive view on stance by looking at meta-communication only, and even less by restricting the analysis to just two expressions.
Still, the study of meta-communicative expressions like I say and I tell (you) can provide insights into the context-dependency of stance markers.The expressions are very often used to express stance and they strongly depend on the situational context.In addition, they have the advantage that they are quite frequent and relatively easy to retrieve, even from large corpora.Thus, they can reveal differences between different (sub-)corpora relatively quickly.For future research, it would be interesting to expand the analysis to texts from other settings and to include additional meta-communicative expressions.In addition, differentiating between different roles and the identities of the language users would provide further insights into how the expression of stance depends on social factors-an aspect that has not yet been explored in detail.
Figure 1.Frequency of I say across CED categories per 100,000 words of direct speech

Table 1 .
Overview of structural types I tell you, that I mistake it nothing at all (Didactic Works, Surueyors Dialogue, 1607) I say, thou stolest Iupiters Crowne from his head, [...] (Drama Comedy, Menaecmi, 1595) other I know what I say.(Didactic Works, Jack and UUill, 1697) I tell you nothing but what is true.(Didactic Works, A New and Easie French Grammar, 1667) Prim.: Hark, my Lady is just return'd.Do you, Sir, but stretch your self out in your Chair, and feign your self dead, you'll then see the violent Grief, she'll be in, when I tell her the News.(Drama Comedy, The Mother-in-Law, 1734) modified uses I can not yet tell where to begyn [...] (Depositions, Courts of Durham, 1560-88/1845) To that question, I must say there was a letter, [...] (Trials, Triall of Mr Love, 1651/1652)

Table 2 .
Frequency of I say across CED text types per 100,000 words of direct speech (and absolute number of instances)

Table 3
. Frequency of I tell (you) across CED text types per 100,000 words of direct speech (and absolute number of instances)

Table 4 .
Context parameters for Trials and Drama Comedy

Table 6 .
Main functions of I say and I tell (you) in Drama Comedy (parentheticals and matrix clauses followed by that-complement clauses) But sirha: what made you in that tree?My man and I at foot of yonder hill Were by three knaues robd of three hundred pound.But.: A shrewd losse berlady sir, but your good worship may now see the fruit of being miserable: You will ride but with one man to saue hors-meat and mans meat at your Inne at night, & lose three hundred pound in a morning.

Table 7 .
Frequent modal modifications of I say and I tell (you) in Trials and Drama