3.1 Introduction

The study of pragmatic markers was presented as a case study in historical pragmatics in Chapter 1. This chapter takes up the topic in more detail. We first provide a nuanced definition of pragmatic markers and their characteristics, examining how they compare with or differ from related “short” forms. We then examine how pragmatic markers are studied historically. In the sections that follow, we explore the origins of pragmatic markers, using representative examples of one-word, phrasal, and clausal pragmatic markers in the history of English, and the pathways that they follow in their development. The question of what processes of change explain the rise of pragmatic markers – a topic that has been hotly debated – is explored in the last section of the chapter.

3.2 Characteristics of Pragmatic Markers

As you saw in Chapter 1, pragmatic markers – often called discourse markers – are forms that are set off syntactically from their clause (their “host”) and lack clear lexical or propositional content, such as well, now, you see, after all, I mean, so, in fact, and you say in the following examples:

1. 1. a. Ah, well, now, you see, they’re me best customers. (2000 Fantasy & Science Fiction; COHA)

   2. b. After all, I mean, a thing can be just so lousy. (1942 Hale, Prodigal Women; COHA)

   3. c. So, in fact, you say, let us forget that war we started, and the defeat, … (1961 The Atlantic Monthly; COHA)

As the names suggest, however, these forms are understood as serving textual (“discourse”) and interpersonal (“pragmatic”) functions: “they situate their host unit with respect to the surrounding discourse and with respect to the speaker–hearer relationship” (Reference WaltereitWaltereit 2006: 64). Rather than propositional or referential meaning, they have what is called “procedural” meaning. That is, they guide the hearer/reader to the correct interpretation of the surrounding discourse. Pragmatic markers have a wide range of functions and often serve multiple functions at the same time. Textually, they may initiate or end discourse, mark a boundary in discourse, introduce, resume, or shift the topic of discourse, acquire, hold, or relinquish the floor in conversations, or denote new or old information. Subjectively, they may express a response or attitude to the preceding or following discourse, signal understanding or continued attendance to what is being said, or hedge an opinion. Interpersonally, they may confirm shared knowledge, effect cooperation or sharing, request confirmation, evoke politeness by expressing deference or showing intimacy. Pragmatic markers thus form a functional class rather than a clear morphosyntactic class. As a result, membership in the category of pragmatic marker is rather fluid, and scholars do not agree upon a definitive set of forms belonging to the category.

Pragmatic markers can be identified by a number of characteristics that are summarized in Table 3.1. While often “short” one-word items (well, now, so, like), they may also be phrases (of course, indeed, by the way) and (reduced) clauses (I guess, you know, it seems). They are often identical in form to fully lexical and compositional forms (e.g., like as a preposition or conjunction as well as a pragmatic marker). They occur preferentially at the clause boundaries (right or left) outside the core syntactic structure (“parenthetically”), often in a separate intonation group. At the same time, they may be moveable and also occur in sentence-medial position. They lack semantic content; that is, they are non-propositional or non-referential in meaning. For this reason, they are not easily glossed or translated. Because they are syntactically and semantically optional: their absence “does not render a sentence ungrammatical and/or unintelligible” (Reference FraserFraser 1988: 22). But it can be argued that they are communicatively and pragmatically necessary, contributing in critical ways to the interpretation of the discourse. Pragmatic markers are frequent and salient in oral discourse and hence colloquial and frequently stigmatized, though a different set of pragmatic markers may also be found in written discourse.

Pragmatic markers overlap with a number of classes of other short, independent forms, such as interjections, conversational routines, hesitation and backchannelling markers, and general extenders. Interjections (e.g., ouch, oops, gee), like many pragmatic markers, are invariable and typically a single morpheme. Some interjections – most notably oh and hwæt (see below, §3.4) – have been analyzed as both interjections and pragmatic markers. However, there are differences between the two categories. Some interjections express the speaker’s wishes (e.g., shh, psst) while others establish contact between speaker and hearer (yeah, mhm), but most are essentially emotive, uttered by the speaker as a spontaneous emotional reaction to a situation (ugh, ouch, gosh). Unlike pragmatic markers, which must be adjoined to a phrase or clause, interjections form a complete utterance on their own. The majority of interjections – the so-called “primary interjections” such as wow, oho, oops – are not identical with any other part of speech; however, some interjections – the “secondary interjections” such as Goddammit, my goodness, thank God – have both independent uses and conventional uses as interjections (see Reference AmekaAmeka 1992). For the most part, primary interjections cannot be traced back to full-content forms, though there are a few exceptions: gee < Jesus (Reference GehweilerGehweiler 2008), marry < by Mary (Reference LutzkyLutzky 2012a), pardie < par Deu ‘by God’, alas < ha/a las ‘miserable’. The inventory of interjections in English has changed in significant ways over time. OE interjections such eala, la/ lo, efne, wa and ME interjections such as a(h), alas, fie, hei, tush, parde, weilawei look very foreign to us today. Many of the most common interjections of Present-day English arise only in the modern period (oops, wow, ouch, yuck). Despite their intrinsic interest, interjections will be omitted from the discussion in this chapter for the sake of space.

Conversational routines, “formulaic speech acts such as thanking, apologizing, requesting, offering, greetings, complimenting” (Reference AijmerAijmer 1996: 2), such as thank you, I’m sorry, how are you?, excuse me, while they bear some resemblance to pragmatic markers in their fixed form and interpersonal meaning, retain their content meaning to a much greater extent than do pragmatic markers, can stand alone as independent utterances, and serve primarily illocutionary functions. A number of these forms are thus treated in Chapter 6 on speech acts (i.e., greetings, thanks) and in Chapter 5 on politeness (i.e., responses to thanks). Backchannel devices and hesitation markers (mhm, uh, um, er) are similarly short, fixed, and syntactically free forms used in oral discourse, but they have strictly defined functions and do not necessarily occupy the clause boundaries. Like primary interjections, they do not derive from content forms and thus do not overlap with other more lexical forms. Forms such as mhm, as you can see, are variously interpreted as either interjections or back channels – thus pointing to the fuzzy boundaries between categories and definition. Finally, “general extenders” such as and stuff, or something (like that), and and so forth, like pragmatic markers, have interpersonal and other discourse-pragmatic functions, occupy the right sentence periphery, and have high frequency in speech, but they are more syntactically and prosodically incorporated than pragmatic markers and form a fairly homogeneous set of forms generally following a fixed pattern (see Reference Overstreet and YuleOverstreet and Yule 2021). See Table 3.2 for a summary of the differences between pragmatic markers and other “short” forms.

3.3 Studying Pragmatic Markers

Once identified, pragmatic markers lend themselves most readily to the form-to-function approach discussed in Chapter 2. But identification can be a challenge. Pragmatic markers may be transitory, and over time the set of pragmatic markers in English has undergone considerable change. For instance, we have lost pragmatic markers such as hwæt, witodlice, anon, iwis, for the nones, what ho, whilom, pray/prithee, faith, marry, I thee bihote, trusteth me wel, forsooth, peradventure, God woot, and I warrant, though at the same time, some pragmatic markers are of long standing, for instance that is to say (from Old English), and I think/guess/believe, you know, you see/say (all from Middle English). It is also necessary to recognize that pragmatic markers may be highly multifunctional, and distinguishing between content uses and pragmatic uses of a particular form is an ongoing challenge.

Pragmatic markers may be studied using a historical pragmatics (proper) or diachronic pragmatics approach, as was discussed in Chapter 2. That is, the use and function of a pragmatic marker in a particular period (or text, or genre, or author) can be the focus of study, or the origin and development of a pragmatic marker over time, from its lexical source, can be explored.

An example of the historical pragmatic approach is Reference Lutzky, Markus, Iyeiri and HeubergerLutzky’s (2012b) study of why and what in EModE drama. These are markers that in Present-day English seem to be restricted to American English. Within the EModE period, Lutzky also considers change over time, from 1500 to 1760, thus incorporating a diachronic pragmatic approach. She finds that while both have continuing uses as interrogatives in this period, why functions as a pragmatic marker 70 percent of the time and what 12 percent of the time, with why increasing in this use over time, as shown in Figure 3.1

Figure 3.1 Distribution of the pragmatic markers what and why in Early Modern English by period

(adapted from Reference Lutzky, Markus, Iyeiri and HeubergerLutzky 2012b: 181) (Ursula Reference Lutzky, Markus, Iyeiri and HeubergerLutzky. 2012. Why and what in Early Modern English drama. In Manfred Markus, Yoko Iyeiri, and Reinhard Heuberger (eds.), Middle and Modern English corpus linguistics: A multi-dimensional approach. Amsterdam: John Benjamins, 181. https://benjamins.com/catalog/scl.50.16lut. Reprinted with permission.)

In their pragmatic use, when introducing interrogatives, both what and why express speaker surprise, irritation, or even contempt; why can also have the implication of disbelief and serve to challenge what the interlocutor has said (2a–b) (see Culpeper and Reference KytöKytö 2010: 383). When introducing declaratives, why is more common and has an explanatory or hedging function (2c), while what displays impatience. When used in conjunction with a vocative, they can also serve to attract the attention of the interlocutor.

1. 1. a. Medley Get thee gone for the prettiest Hero that ever was shown on any Stage.

      Exit Pistol

      Sowrwit Short and sweet, faith, what, are we to have no more of him? (1737 Fielding, The Historical Register 22–3; CED; Reference Lutzky, Markus, Iyeiri and HeubergerLutzky 2012b: 182)

   2. b. Ranger … Here have I been at it these three Hours – but the Wenches will never let me alone –

      Bellamy Three Hours! – Why, do you usually study in such Shoes and Stockings? (1747 Hoadley, The Suspicious Husband 4; CED; Reference Lutzky, Markus, Iyeiri and HeubergerLutzky 2012b: 186)

   3. c. Countesse … vnlesse you will sweare to me, you will neither court nor kisse a dame in any sort, till you come home againe.

      Moren Why I sweare I will not. (1599 Chapman, An Humerous Dayes Myrth D3V–D4 R; CED; Reference Lutzky, Markus, Iyeiri and HeubergerLutzky 2012b: 185)

In sum, both are primarily interpersonal markers, expressing the speaker’s attitude toward the interlocutor’s utterances (Reference Jucker, Watts and TrudgillJucker 2002: 221).

An example of the more obviously diachronic pragmatic approach is Reference Davidse, Simon and LindenDavidse et al.’s (2015) study of the development of no doubt/I have no doubt. The original epistemic meaning (of certainty and predictability) of this form has eroded, and as a pragmatic marker, it has come to express interactional relations and interpersonal stance, sometimes with irony and ridicule, as in these examples:

1. 1. a. “Well, I’m sorry for it,” replied he, with more of sulkiness than contrition: “what more would you have?”

      “You are sorry that I saw you, no doubt,” I answered, coldly.

      “If you had not seen me,” he muttered, fixing his eyes on the carpet, “it would have done no harm.” (1848 A. Brontë, The Tenant of Wildfell Hall; The Corpus of Late Modern English Texts (extended version) [CLMETEV]; Reference Davidse, Simon and LindenDavidse et al. 2015: 49)

   2. b. “My father is a man who seldom gives way to any elation of mind.”

      “Ah, indeed! A philosopher, I have no doubt, like his son.”

      “I have no claims to the title of philosopher, … ” (1826 Disraeli, Vivian Grey; CLMETEV; Reference Davidse, Simon and LindenDavidse et al. 2015: 45)

The pragmatic marker “can be used to invite a response from the addressee … or to express agreement to the preceding utterance” (Reference Davidse, Simon and Linden2015: 55). Davidse et al. argue that a set of structures predated the pragmatic marker use, including existential there/it/∅ be (no) doubt, personal I have/make (no) doubt, and adverbial no doubt (perhaps modeled on without doubt/out of doubt), all of which developed epistemic meaning from the original lexical meaning. The pragmatic marker use arose in Late Modern English in the case of the latter two structures, the adverbial form with its greater positional flexibility leading the way. But the existential form did not develop into a pragmatic marker.

The assumption behind diachronic pragmatic studies is that pragmatic markers derive from fully compositional, lexical structures that undergo processes of change in the evolution toward pragmatic marker status. As Reference Traugott and DasherTraugott and Dasher (2002: 156) argue, pragmatic markers “typically arise out of conceptual meanings and uses … Over time, they not only acquire pragmatic meanings (which typically coexist for some time with earlier, less pragmatic meanings) but also come to have scope over propositions.” The semantic development of pragmatic markers is fairly well understood and follows general principles of semantic change (see Reference Traugott and DasherTraugott and Dasher 2002: Ch. 4):

• Content meaning evolves into procedural meanings through invited inferences that arise in the context of use. Invited inferencing occurs in many types of semantic change. For example, hearing a sentence such as I have been unhappy since they left, you might infer a causal meaning (“because they left”) from a meaning that is originally temporal (“after they left”). The causal meaning can (and in this case does) then become part of the conventional meaning of the word.

• Meanings which are neutral and objective develop into meanings expressing speaker attitude or subjective belief (“subjectification”) as well as meanings focused on the speaker–hearer interaction (“intersubjectification”). For example, the change in while from its temporal meaning (I listen to podcasts while I cook) to a concessive meaning (While the book is well written, I didn’t like the plot) shows a change from an objective meaning of temporal simultaneity to one which expresses my subjective opinion of contrast, or subjectivity. We can see a change to intersubjective meaning in surely: from the subjective meaning (We surely wouldn’t be paid for that day – i.e., ‘I am confident/I believe we wouldn’t be paid’) to intersubjective meaning (Surely, she of all people understands – i.e., ‘I believe she understands and I believe you do too’).

• Meanings within the scope of the sentence come to have meanings related to the sentence as a whole or to larger chunks of discourse (Reference Traugott and DasherTraugott and Dasher 2002: 40, 281).

In contrast, the syntactic pathways that pragmatic markers follow may be quite varied. Pragmatic markers have been found to originate in a wide variety of sources, including adverbials, conjunctions, prepositions, nouns, reduced main clauses, adverbial and relative clauses, nonfinite clauses, imperative clauses, pseudo-cleft clauses, and other structures (see Reference Heine, Kaltenböck, Kuteva and LongHeine et al. 2021: 245). The syntactic pathways followed by pragmatic markers are examined in the next three sections.

3.4 Single-Word Pragmatic Markers

An early hypothesis concerning the development of pragmatic markers was that they develop from adverbs via conjunctive uses: adverb > conjunction > pragmatic marker (Reference Traugott, Lehmann and MalkielTraugott 1982). This pathway might explain the origin of the OE pragmatic marker hwæt. Hwæt functions as an interrogative adverb meaning ‘what, why’ introducing direct questions (4a) and as a conjunction introducing indirect questions (4b). It can also be syntactically independent from the sentence that it is adjoined to and serve as an introductory particle with vague or indeterminate meaning (4c). This last usage appears at the beginning of Beowulf as well as a number of other OE poems and often introduces important sections within the poems. Translators have struggled to gloss this form, suggesting a range possibilities, including ‘indeed’, ‘certainly’, ‘lo’, ‘listen’, ‘hear me’, and ‘what ho’.

1. 1. a. Hwæt sceal ic ma riman yfel endeleas? (Jul 505; DOEC; Reference BrintonBrinton 2017b:14)

      ‘why must I recount more endless evil’

   2. b. God ana wat hwæt him weaxendum winter bringað. (Fort 8; DOEC; Reference BrintonBrinton 2017b:14)

      ‘God alone knows what the winter brings for the growing one’

   3. c. Hwæt, we for dryhtene iu dreamas hefdon,/ song on swegle selrum tidum, (Sat 44–5; DOEC; Reference BrintonBrinton 2017b: 48)

      ‘Why, we before had joys in front of the lord, song in the heavens in better times’

Because of its syntactic independence and apparently expressive or exclamatory force, the usage in (4c) has traditionally been treated as an “interjection,” though this view has been contested (Reference WalkdenWalkden 2013). If we carefully examine the contexts in which this form occurs, we see that hwæt almost always occurs in the context of first- or second-person pronouns (i.e., an interpersonal context) in which the poet or character in the poem indicates how information has been acquired, reminds the hearer of what has been said or narrated, recalls past events (as in 4c), or points out what is evident or visible. What follows hwæt is common knowledge because it is part of the cultural or historical tradition, it is part of the immediate past of the hearer, or it has been given in the text. That is, hwæt serves to call attention to the following discourse and to establish common ground or knowledge between speaker or hearer. It may also serve to bring information to consciousness by renewing interest in it and making it salient. Thus, OE hwæt serves the same functions that have been identified for the pragmatic marker you know/y’know in Present-day English. This pragmatic use of hwæt declines and falls out of use in Middle English. A common pragmatic marker in Middle English is what ho, used as an attention-getting device. The PDE use of y’know what? and guess what? as an attention-getting device and narrative opening, especially in children’s language, may show remnants of both the OE and ME uses. The EModE use of what to express surprise (discussed above) can also be traced back to Middle English.

The question of whether the pragmatic marker develops out of the conjunctive use, following the pattern adverb > conjunction > pragmatic marker must be answered in the negative, however. While the conjunctive use develops from the use of hwæt as an interrogative adjective/adverb, it is not clear that the pragmatic use derives from the conjunctive use. In fact, the pragmatic use is contemporary with, or predates, the conjunctive use.

Addressing problems of this sort, Reference TraugottTraugott (1995/97: 13) proposes a different sequence of development: clause-internal adverb > sentential adverb > pragmatic marker. That is, an adverb such as a manner adverb moves from its position in the predicate of the sentence where it has narrow scope to the position of a sentential adverb (typically clause-initial position) where it has scope over the entire proposition. In this position it then acquires pragmatic meanings (such as those relating to the upcoming discourse or to speaker–hearer relations) and ceases to evaluate the proposition. Reference LenkerLenker (2010) discusses such a development for the ME adverb truly. It begins as a manner adverbial meaning ‘faithfully, loyally, with steadfast allegiance’, as in (5a). It then acquires epistemic meaning and begins to function as a sentence adverbial, expressing a subjective perspective ‘I am being truthful when I say’, ‘it is true (in my opinion) that’, as in (5b). Finally, it loses its epistemic meaning and acquires procedural meaning, serving as a pragmatic marker, or what Lenker calls a “transition marker”; it highlights the connective function of and and links to the following discourse (5c).

1. 1. a. He was also a lerned man, a clerk,/ That Cristes gospel trewely wolde preche; (1387 Chaucer, Canterbury Tales A.Prol. 480–1; Reference LenkerLenker 2010: 124)

      ‘He was also a learned man, a clerk who would loyally preach Christ’s gospel …’

   2. b. For trewli, fader I love oon (1393 Gower, Confessio Amantis 5.2536; Reference LenkerLenker 2010: 125)

      ‘For indeed, father, I love one’

   3. c. The which a long were to devyse./ And trewely, as to my juggement,/ Me thynketh it a thyng impertinent, (1387 Chaucer, Canterbury Tales E.Cl 52–4; Reference LenkerLenker 2010: 126)

      ‘ … it would take a long time to describe this thing to you and [TRANSITION marker], in my mind it is an irrelevant thing’

Another pragmatic marker that seems to follow this pattern is ME whilom (see Reference BrintonBrinton 2017b: Ch. 3). This form is the dative plural of OE hwīl ‘a while, space of time’. In Old English it is a sentence-internal adverb meaning ‘at times, for a time, sometimes’, as in (6a); this usage becomes rare in Middle English and obsolete by Early Modern English. Whilom begins to function as a sentence adverbial with the meaning ‘formerly, at some past time’; this is its usual meaning from Middle English through Early Modern English (6b). In (6c) we see whilom functioning as a pragmatic marker with global scope marking textual structure. Specifically, it serves to introduce an episode or story and, as Reference Jucker and TaavitsainenJucker and Taavitsainen note (2013: 61), it also has genre-specific functions, setting up the expectation that a romance or fabliau will follow.

1. 1. a. Þinceð him on mode þæt he his mondryhten clyppe ond cysse, … swa he hwilum ær in geardagum giefstolas breac. (Wan 41; DOEC)

      ‘he thinks in his mind that he embraces and kisses his lord, … just as sometimes before in days gone by he enjoyed the throne’

   2. b. & he answered þus:/ Whilen hit wes iseid; inne soð spelle,/ þat moni mon deð muchel vuel (a1225 Layamon, Brut 4129–31; Corpus of Middle English Prose and Verse [CMEPV]; Reference BrintonBrinton 2017b: 81)

      ‘and he answered thus: formerly it was said, in a true story, that many men did much evil’

   3. c. Whilom ther was dwellynge in my contree/ An erchedeken, a man of heigh degree, … (1387 Chaucer, Canterbury Tales D.Fri 1301–2; CMEPV)

      ‘Once upon a time there was dwelling in my country, an archdeacon, a man of high degree … ’

While these two examples, truly and whilom, seem to show the progression from clause-internal adverb to sentential adverb to pragmatic marker, the historical data are often messy and sometimes scant. We see in the case of truly, for example, that all of the uses cited in (5) are roughly contemporaneous (fourteenth century). For whilom, the sentence adverbial and pragmatic marker uses (6b–c) seem to overlap in time; the Middle English Dictionary (MED) (s.v. whīlom, adv.) does not distinguish clearly between the two uses: “At one time, formerly, once; also, in narrative contexts: once upon a time” (def. a). Moreover, the examples of hwilom in the DOEC are already overwhelmingly clause-initial, suggesting that the sentence adverbial use is already possible. Thus, while the progressions look plausible, it is difficult to establish firm dating in these and similar cases.

3.5 Phrasal Pragmatic Markers

A number of pragmatic markers derive from prepositional phrases that are “univerbated” (coalesced) and “lexicalized” (comprehended as a single word, though not always written as such). These include forms such as ME forsooth, peradventure, EModE perchance, perhaps, indeed, and LModE in fact, in truth, instead, after all, anyway, and of course.

Reference Traugott and DasherTraugott and Dasher (2002: 159–165) trace the development of indeed from sentence-internal adverb to sentential adverb to pragmatic marker. In Middle English, it serves as an adverbial meaning ‘in action, practice’ (7a). By the sixteenth century, it appears in initial position as a sentential epistemic adverb meaning ‘in truth’ (7b). Finally, it is recruited as a pragmatic marker meaning something like ‘what’s more’ (7c) where it “signals that what follows is not only in agreement with what precedes, but is additional evidence being brought to bear on the argument” (Reference Traugott and Dasher2002: 164).

1. 1. a. Fals witnesse is in word and eek in dede. (1387 Chaucer, Canterbury Tales I.Pars 795; CMEPV).

      ‘false witness is in word and also in deed’

   2. b. Dan.: … I pray you tell me. I thinke you take me for your friend.

      Sam.: In deede I haue always found you my very good friend … (1593 Witches, p. A4V; Reference Traugott and DasherTraugott and Dasher 2002: 163)

   3. c. last of all came in his youngest sonne, riding in a chariot, that it was drawne with buffeles [‘buffalos’], which me thought was very vnseemly: in deed they haue but few horses, the which are smal nags, i neuer saw any of them put to draw, but only to ride on, (1606 Scott, An Exact Discourse … ; EEBO)

As in the case of truly and whilom above, the matter of timing here casts some doubt on the proposed sequence. The sentence adverbial use and the pragmatic marker use are almost contemporaneous. Moreover, distinguishing semantically between the sentence adverbial and pragmatic marker use is often very difficult.

In more recent work, Reference TraugottTraugott (2020, Reference Traugott2022) does not argue for this simple linear progression, though she still traces a pathway from sentence-internal adverbial to pragmatic marker. The “digressive” pragmatic marker by the way originates in an adverbial in Middle English with the literal meaning ‘along the way’, in either a static or dynamic sense, as in (8a). Frequent use of this form with verbs of speaking (e.g., talkynge by the way, told him by the way) contributed to acquisition of figurative meaning (based on the schema that an argument is a journey). We see an example of this in (8b), where by the way means ‘in passing’. Traugott notes that this meaning is related to discourse organization but its scope is still local, not global, and hence it is not a fully formed pragmatic marker. Structures such as those in (8c), where the form occurs with a relative clause that the speaker marks as an unimportant elaboration, express a high degree of subjectivity and represent an important step toward pragmatic marker status. The form achieves this status in examples such as (8d), where by the way occurs in clause-initial position, makes no mention of a path, either literal or textual, and expresses a high degree of subjectivity; that is, the upcoming discourse is only partially relevant in the speaker’s opinion.

1. 1. a. And by the weye his wif Creusa he les. (c. 1430 (c.1386) Chaucer, LGW 945; MED; Reference TraugottTraugott 2020: 123)

      ‘And along the way his wife Creusa he lost’

   2. b. plato in his dialogue … maketh mention by the way of a wonderfull earthquake, (1563 Fulke, A Goodly Gallery; EEBO; Reference TraugottTraugott 2020: 124)

   3. c. it may be also, that within these two or three hundred yeares one of their great Auncestors, whom by the way they repeat in their genealogies from their demigorgons, i would say demigods, might come in at the window indirectly. (1630 Vaughn, The Arraignment of Slander; EEBO; Reference TraugottTraugott 2020: 125)

   4. d. bees turn not droanes, nor courages ever abate or degenerate: by the way, I observe that none have ever arrived to an eminent grandeur, but who began very young. (1661 Argyll, Instruction to a Son; EEBO; Reference TraugottTraugott 2020: 126)

What seems clear is that strict pathways do not capture the many and varied ways in which sentence-internal adverbs develop into pragmatic markers and each form follows its own unique path.

3.6 Clausal Pragmatic Markers

In English, there exist a wide variety of clausal pragmatic markers, also known as “comment clauses.” In form, they represent a number of different structures, as shown in Table 3.3.

Declarative main clause pragmatic markers have received the most attention. In an important article Reference Thompson and MulacThompson and Mulac (1991), using synchronic data, propose a sequence of development for parenthetical I think and I guess, forms which they call “epistemic parentheticals.” They observe that guess and think together constitute 65 percent of all complement-taking verbs, occur overwhelmingly with first-person subjects, and are the most common verbs in epistemic parentheticals in their corpus of conversational English. They argue that epistemic parentheticals originate as main clauses followed by that complements, as in (9a).

1. 1. a. I think that we can do it this time. (2002 LaBute, The Mercy Seat; COHA)

   2. b. I think ∅ the neighbors complained. (2001 Glancy, American Gypsy; COHA)

   3. c. It’s too late for that, I think. (2000 Cruz, Yellow Eyes; COHA)

      We’ve spoken only two or three times, I think, in all our years of school. (2002 Fantasy & Science Fiction; COHA)

In Thompson and Mulac’s data, I think and I guess occur over 90 percent of the time without the complementizer that, as in (9b). Here the status of I think is indeterminate, allowing for its reanalysis as a parenthetical. This results in the reversal of the main clause/complement clause structure: the original main clause (I think) assumes parenthetical status and the complement clause (the neighbors complained) takes on main clause status. Once I think assumes parenthetical status, it becomes moveable, as in (9c). I have termed this the “matrix clause hypothesis” (Reference BrintonBrinton 2008: 36).

Reference Boye and HarderBoye and Harder (2007) propose a development pathway incorporating both structure and usage which is compatible with Thompson and Mulac’s. They argue that the source construction consists of a main lexical clause, I think, with “primary discourse status” (it is the main point of the utterance and is inherently addressable – one could ask really?) followed by a complement clause. This undergoes usage reanalysis in which the main clause assumes secondary status. It is no longer the main point of the utterance but rather has a “modifying status.” I think has been reanalyzed at the level of usage but not of structure. Finally, I think is structurally reanalyzed as “grammatical” (inherently non-addressable) and secondary to the point of the utterance (non-addressable, and moveable).

The matrix clause hypothesis has intuitive appeal. But does the synchronic progression proposed by Reference Thompson and MulacThompson and Mulac (1991) account for the diachronic development of main clause pragmatic markers? Do they all originate as complement-taking main clauses? Extensive studies of these forms in English suggest that in some cases, the matrix clause hypothesis is borne out and in other cases it is not. I provide an example of each case.

Methinks/methinketh ‘it seems to me’ is a pragmatic marker formed from univerbation of the dative personal pronoun me and impersonal uses of the verb thinken. (ME thinken is a conflation of two verbs in Old English, one personal and one impersonal.) This form can be followed by a that clause (10a) or a that-less clause (10b), and it may also function parenthetically, in either the present or past tense (10c–d).

1. 1. a. Me þincð þæt hit hæbbe geboht sume swiðe leaslice mærðe. (King Alfred tr. Boethius De Consol. Philos. (Otho) xxiv.54; OED)

      ‘it seems to me that it has bought some very false fame’

   2. b. With such gladnesse I daunce and skippe, Me thenkth I touche noght the flor. (a1393 J. Gower, Confessio Amantis (Fairf.) iv.2785; MED)

      ‘with such gladness I dance and skip, it seems I do not touch the floor’

   3. c. ’Twere good, me thinks, to steal our marriage (1590–1 Shakespeare, Taming of the Shrew III.ii; OSS)

   4. d. Nay, by the mass that he did not; he beat him most unpitifully, me thought. (1597 Shakespeare, Merry Wives of Windsor IV.ii; OSS)

Reference WischerWischer (2000) argues that in Old English and Middle English me thinks was a free syntactic construction that could occur with other pronouns, in a full range of tenses and moods, and with variant word orders. However, over time it came to have a more fixed order (with the pronoun immediately preceding the verb) and to be followed by a that-less complement. Impersonal uses of the verb declined in the fifteenth century, with the exception of those with the first-person pronoun. The impersonal use was reserved for expressing ‘opinion’, while the personal use could express ‘have in mind’ as well as ‘opinion’ (Reference Palander-Collin, Rissanen, Kytö and HeikkonenPalander-Collin 1997). These changes allowed for the complement clause to be reinterpreted as the main clause, and me thinketh to be reanalyzed as an epistemic or evidential parenthetical, a change that began in the fifteenth century. Parenthetical methinks continued to be used until the eighteenth century but is now archaic, poetic, or regional (OED, s.v. methinks, v.). The progression outlined in Wischer thus conforms to the “matrix clause hypothesis.” Reference Palander-Collin, Rissanen, Kytö and HeikkonenPalander-Collin’s (1997: 387) statistical data also shows a progression from that clause to ∅-clause to parenthetical for impersonal thinken, but she notes that the earliest parentheticals in her data are second person, not first person.

Diachronic evidence suggests that the very forms that Reference Thompson and MulacThompson and Mulac (1991) use to illustrate the matrix clause hypothesis, I think and other epistemic parentheticals, arise historically in a different way. As I have argued (Reference BrintonBrinton 2017b: Ch. 5), first-person epistemic parentheticals are well established in Middle English, formed with a variety of verbs. They may occur in final position (11a), in medial position (11b), and ambiguously in initial position (11c).

1. 1. a. “And that ye weten wel yourself, I leve.” (1380–86 Chaucer, Troilus and Criseyde II 238; Reference BrintonBrinton 2017b: 142)

      ‘And that you know well yourself, I believe’

   2. b. “I nas, ne nevere mo to ben, I thynke,/ Ayeins a thing that myghte the forthynke.” (1380–86 Chaucer, Troilus and Criseyde II 1413–14; Reference BrintonBrinton 2017b: 142)

      ‘I wasn’t, nor never more to be, I think, against a thing that might displease you’

   3. c. “I woot right wel, thou darst it nat withseyn.” (1387 Chaucer, Canterbury Tales A.Kn. 1140; Reference BrintonBrinton 2017b: 148)

      ‘I know quite well, you dare not deny it’

Thompson and Mulac’s matrix clause hypothesis rests on the frequency of ambiguous that-less clauses as well as the predominance of first-person subjects. Middle English data on that-deletion and the use of cognitive verbs with first-person subjects are not, however, persuasive (see Reference BrintonBrinton 2017b: 158–159). Moreover, if we look back to Old English, we find that although it has a large set of cognitive verbs that occur frequently with an adjoined þæt-clause, the omission of the complementizer is rare. A much more likely OE source for epistemic parentheticals is an adverbial clause with þæs þe, as shown in (12a). The complementizer þæs þe functions as an adverbial connective meaning ‘the measure in/the extent to which’ (Reference FischerFischer 2007). This construction is continued by ME adverbial clauses with as or so (12b).

1. 1. a. se hæfde ænne sunu nu for þrym gærum, & se wæs, þæs þe ic wene, V wintre, (GDPI and 4(C) 19.289.3–4; Reference BrintonBrinton 2017b: 156)

      ‘he had one son now for three years, and he was, so I know, five winters old’

   2. b. “For thrittene is a covent, as I gesse.” (1387 Chaucer, Canterbury Tales D.Sum. 2259; Reference BrintonBrinton 2017b: 160)

      ‘For thirteen is a convent, as I guess’

The development of epistemic parentheticals from these adverbial clauses is straightforward. As adverbial clauses they are by nature more or less loosely adjoined to the main clause. Deletion of the complementizer leads to greater independence and parenthetical status and allows for movement to clause-medial and initial position. Unlike the matrix clause hypothesis, there is no reorganization of the syntactic hierarchy. Another type of epistemic parenthetical, the impersonal it seems (to me) seems to arise in the same way. Reference López-Couso and Méndez-NayaLópez-Couso and Méndez-Naya (2014) find that adverbial as/so it seems/thinks clauses predate and give rise to “bare” parentheticals. Furthermore, they find that that-deletion with it seems becomes frequent only well after the development of parenthetical forms, concluding that it is the parentheticals which motivate the deletion of that, not the reverse.

Adverbial clauses are a frequent source of clausal pragmatic markers (see Table 3.3). An example of an adverbial clause development is as it were, which functions exclusively as a metalinguistic PDE pragmatic marker meaning that “a word or statement is perhaps not formally exact though practically right” (OED, s.v. as, adv. and conj., def. P2a); it hedges the linguistic adequacy of the expression (Reference ClaridgeClaridge 2013: 162). Unlike other pragmatic markers, this marker is equally common in speech and in some types of writing. The remnant subjunctive were undoubtedly contributes to its more “literary” quality and its use in academic writing as well as fiction. Also, unlike other pragmatic markers, as it were has a more local scope, often qualifying the adjacent word or phrase. For example, it may accompany figures of speech, clichés, words set off by quotation marks, foreign words or phrases, unusual native formations, or a word that is unexpected in context. The development of this form has been studied by Reference BrintonBrinton (2008: 171–177, 180–182) and Reference ClaridgeClaridge (2013), who arrive at similar if not identical conclusions. The form swa/swilc hit wære is rare in Old English; here and in Middle English it is followed by a subjective complement and forms a complete clause. It is a clause of hypothetical comparison meaning ‘as if it were X’. It refers back to a noun in context (13a). This usage dies out after the sixteenth century. By the first half of the fourteenth century, as it were occurs without a complement and has metalinguistic meaning; it is no longer referential (13b). In Early Modern English, pragmatic as it were is the majority form, becoming the only possible use by c. 1700 (Reference ClaridgeClaridge 2013: 166). As it were occurs in EModE contexts similar to those in modern English, for example with figurative language (13c).

1. 1. a. His comb was redder than the fyn coral,/ And batailled as it were a castel wal; (1387 Chaucer, Canterbury Tales B.NP 2859–60; Reference BrintonBrinton 2008: 172)

      ‘His comb was redder than fine coral and notched with crenellations as if it were a castle wall’

   2. b. Hi seye an aungel … a ffayr ȝong man as hit were, Ycloþed in white cloþes. (a1325 (c1280) South English Legendary: Temporale (Passion of Christ) (Pep2344) 1824; MED; Reference BrintonBrinton 2008: 173)

      ‘They saw an angel … a fair young man as it were clothed in white clothes’

   3. c. Sir, he hath never fed of the dainties that are bred in a book; he hath not eat paper, as it were; he hath not drunk ink. (1594–95 Shakespeare, Love’s Labours Lost IV.ii; OSS)

Pragmatic as it were is a native construction, though its frequency in Middle English is likely influenced by Latin quasi, adopted through French, which is often translated as it were. The development postulated for as it were again (unlike the matrix clause hypothesis) does not involve any change in syntactic hierarchy, since the adverbial clause is intrinsically semi-independent and subordinate to the main clause. The adverbial complementizer is preserved in this case but the clause is truncated. How is the complement lost? One scenario sees the complement being reanalyzed as an appositive (e.g., (13a) could be read “his comb was redder than fine coral and crenelated (as it were), a castle wall”). Another scenario sees change as occurring in ambiguous contexts such as (14), where as it were can be read as a comparative clause with the complement belonging to the as it were clause (“which is to God as if it were an enemy”), or it can be read as parenthetical with the complement belonging to the main clause and hence omissible (“which is to God an enemy”) (Reference ClaridgeClaridge 2013: 164).

1. (14) þe whiche is to God as it were an enmye … (1350–1420 cmcloud; HC; Reference ClaridgeClaridge 2013: 164)

   ‘which is to God as (if) it were an enemy’

The metalinguistic meaning preserves the comparative ‘similar to’ sense of the source construction but the form also acquires subjective meaning since it expresses the speaker’s assessment of the state of affairs (Reference ClaridgeClaridge 2013: 167). There is general fossilization of the form:

• as expresses comparison, whereas in other contexts this meaning becomes obsolete (OED, s.v. as, adv. and conj., def. BI1b), replaced by as if or as though;

• the subjunctive is preserved; and

• it becomes non-referential and non-anaphoric.

As a result of clausal truncation and ossification, as it were loses its clausal qualities and becomes particle-like – a “fossilized formula” in Claridge’s terminology. It becomes fully independent or parenthetical (often set off by commas in contemporary texts) and becomes moveable.

Imperative and free relative clauses/absolutes are much less common as sources of pragmatic markers (see Table 3.3). Like declarative Subject Verb constructions such as I think, imperatives such as look, say, and see are frequently followed by complement clauses (including both that clauses and wh-clauses). In their development as pragmatic markers, the imperative forms undergo the same reversal of syntactic hierarchy we see in the matrix clause hypothesis, as in:

• _main[Look] _subordinate[that you love your wife] > _{parenthetical}[Look (you)] _main[love your wife]

A free relative clause such as what I’m saying also undergoes a reversal in syntactic hierarchy. It begins as the subject of a pseudo-cleft clause (as in 15a) and is “demoted” to a parenthetical pragmatic marker attached to the original complement (as in 15b). Interestingly, we see a case of syntactic misanalysis (in 15c) where the free relative in its parenthetic use is accompanied by the original main verb is.

1. 1. a. What I’m saying is that there should be a diversity of dollars spent. (2017 Fox: Ingraham Angle; COCA)

   2. b. What I’m saying, I think it’s something genetic. (1997 The Massachusetts Review; COCA)

   3. c. All I’m saying is, you looked with me and it took them about five minutes. (2005 CNN_Dolans; COCA)

3.7 Processes of Change

The preceding sections have looked at the different historical sources of pragmatic markers. While scholars are generally agreed on the sources of many pragmatic markers and the pathways they follow in their (often complex) development, there is one area which is still much disputed: the process of change which best explains the changes. A number of well-known processes of change have been suggested, including lexicalization, grammaticalization, pragmaticalization, and, most recently, cooptation. We will examine this controversy below.

3.8 Chapter Summary

This chapter covered the following topics:

• the characteristics of pragmatic markers as syntactically independent, grammatically optional words, clauses, and phrases with little or no conceptual meaning that are frequent in oral discourse;

• the relation of pragmatic markers to other “short” items such as interjections, conversational routines, backchannel devices, and general extenders;

• the study of pragmatic markers from a form-to-function approach either in a historical pragmatics (proper) or diachronic pragmatics perspective;

• the assumption that pragmatic markers derive diachronically from fully compositional, lexical constructions;

• the semantic development of pragmatic markers by general principles of semantic change, including invited inferencing, (inter)subjectification, and extension of scope;

• the syntactic pathways followed by single-word pragmatic markers, phrasal pragmatic markers, and clausal pragmatic markers (including declarative main clauses, adverbial clauses, imperatives and free relatives);

• the viability of the “matrix clause hypothesis,” especially in relation to epistemic parentheticals; and

• the choice of processes of change responsible for the development of pragmatic markers, with grammaticalization favored over lexicalization, pragmaticalization, and cooptation, if an expanded definition of grammar is accepted.