Decompositionalization and Partial Recompositionalization: The Emergence of by the Same Token as a Polyfunctional Discourse Marker

Abstract

This paper documents the constructionalization of by the same token. Originally, the word same in this phrase did not encode similarity but functioned as an identification emphasizer and a marker of syntactic dependency between the evidential noun token and the clause that followed it. By the same token then acted as a complex subordinator introducing a justification. The data suggest that two non-compositional uses developed from the evidential subordinator during the seventeenth century: a digressive discourse marker and a subordinator combining high degree and consequence. Faced with polysemy and lack of transparency, speakers/hearers then reintroduced some compositionality to the phrase by assigning to same the meaning of similarity that it had elsewhere. From this partial recompositionalization stems today’s elaborative discourse marker. Originally used to connect two consequences of the same premise, it then extended its connective value. It is now polyfunctional and is even often used to connect two contrasting statements. There are signs that it is now often treated by speakers as a member of a constructional network of adversative discourse markers.

Keywords

discourse marker constructionalization compositionality digressive elaborative adversative

1. Introduction

The phrase by the same token is included in Fraser’s (2009) list of so-called “elaborative discourse markers,” together with other connectives like and, also, besides, equally, in addition, and similarly. These discourse markers “signal an elaboration in [discourse segment] 2 to what is contained in [discourse segment] 1” (Fraser 2009:301). In Present-Day English, by the same token is said to be “a general way to connect statements that have some logical association with one another” (Ammer 1997, s.v. by the same token). This use is illustrated in (1).

(1) “Explain to your family what you will be doing and that this is important to you,” he says. “Make it clear that this means there might not be Oreos in the house or you might not be at home all the time because you will be out exercising.” #By the same token, family members need to support dieters not just with their words, but through their actions, as well. (COHA, Magazine, 1996)

The evolution of elaborative markers is discussed in detail by Lenker (2010), who calls them “additive adverbial connectors.” She argues that all equative additive markers are fairly transparent (such as in the same way, coined during the Late Modern English period) (Lenker 2010:221), and she devotes comparatively few pages to this type of markers in her comprehensive diachronic study of adverbial connectors. However, while by the same token, which she does not mention, clearly belongs to this class of connectors (see Quirk, Greenbaum, Leech & Svartvik 1985:635), it is not transparent. It is analyzable morphologically, but its degree of “chunking” is higher in the sense that it is more of a fixed phrase than other phrasal additive adverbials and its semantic trajectory is much more complex than that of most other additive/elaborative markers. For instance, the degree of chunking of by the same token is higher than that of in the same way. COHA yields five different phrases with a comparable meaning based on the same pattern as in the same way: in the same way (277 occurrences), in the same manner (fifty occurrences), in the same vein (twenty-five occurrences), in the same spirit (nineteen occurrences), and in the same fashion (two occurrences). In contrast, by the same token has virtually no alternative forms: by the same token (156 occurrences), versus by the same logic (two occurrences) and by the same reasoning (two occurrences).

By the same token is a lexicalized chunk: it is stored, processed, and produced as a unit rather than component by component; from a semantic point of view, its meaning does not equal the sum of the meaning of its parts. This explains why the phrase is included in several dictionaries of idioms (e.g., Ammer 1997; Spears 2006, s.v. by the same token), which are, by definition, non-compositional. As the OED notes, in this phrase “the sense of token becomes vague” (s.v. token, n. def. 15), and, as I argue in this paper, the meaning of same initially had nothing to do with similarity. This study shows that the phrase gradually lost compositionality over the Early Modern English period, and that it is only during the Late Modern English period that it regained a degree of compositionality under the influence of the modern meaning of same. Although it is often assumed by Present-day speakers that similarity is the core value of the phrase by the same token, it is in fact a rather recent development.¹

I propose the term “decompositionalization” to refer to the loss of compositionality and contrast it with the term “recompositionalization,” which, conversely, refers to an increase in compositionality. These terms function as a pair and have the advantage of highlighting the dynamic nature and potential reversibility of these two processes.

This paper explores how the semantics of by the same token has evolved over time. After a description of the background and the corpora used in this study (sections 2 and 3), the paper presents an overview of the evolution of by the same token (section 4). It then documents its evolution in Early Modern English. The phrase first underwent a phase of fixation, during which the alternative phrases by this/that (adjective) token, by these/those (adjective) tokens, and by the (adjective) token(s) receded. The phrase by the same token took over and became a complex subordinator (section 5). The subordinator then went through a process of decompositionalization, exhibiting two different semantic shifts. The subordinator shifted toward the expression of high degree and consequence on the one hand and toward the encoding of digression on the other. The next stage of the constructionalization of the phrase exhibits partial recompositionalization, leading to the elaborative discourse marker (section 6). Finally, before a conclusion in section 8, section 7 addresses the polyfunctionality of the discourse marker in Present-Day English, with a focus on its recent adversative value.

2. Background

Discourse markers have been the focus of much research since the 1980s (e.g., Schourup 1985, 1999; Schiffrin 1987; Fraser 1990; Brinton 1996). They constitute a rather fuzzy category, but they are all characterized by their procedural rather than representational function. They do not contribute significantly to the semantic content of the sentence but provide an indication of how to process meaning.

There is no absolute consensus on the precise delimitation of discourse markers. For instance, Lutzky (2006:8) explicitly excludes by the same token from discourse markers on the grounds that it is not multifunctional (cf. Lenk 1998:50), short, informal, or typical of oral language (cf. Brinton 1996:33-35). While by the same token is today a feature of formal written English (e.g., 2.45 occurrences per million words in the academic section of COCA, as opposed to 1.1 on average in the whole of COCA), it was once an informal discourse marker (section 6) and is multifunctional in Present-Day English (sections 7 and 8). In addition, the length is actually quite typical of additive connectors, as it iconically signals their reinforcing value (Lenker 2010:219).

The process leading to the emergence of discourse markers has given rise to several theoretical interpretations, the most recent to be invoked being the constructionalization framework (Hilpert 2013; Traugott & Trousdale 2013). This approach has the advantage of removing the separation between lexicon and grammar, which is useful for the study of discourse markers. A construction is a learned form-meaning pair whose meaning is not entirely inferable from the sum of its parts (Goldberg 1995). This is why constructionalization is particularly relevant for the study of phrases that are not fully compositional like by the same token. The concept of constructionalization refers to “the creation of form_new-meaning_new (combinations of) signs” (Traugott & Trousdale 2013:22), affecting both syntax (form) and semantics (meaning). Constructional change, by contrast, refers to those cases where the change is only syntactic or only semantic.

The concept of constructionalization is a recent development of the grammaticalization approach, which has led to numerous insights regarding recurrent cross-linguistic aspects of syntactic and semantic changes. Researchers working within the grammaticalization framework have notably documented prosodic detachment and increase in syntactic scope (e.g., Brinton 2008). They have analyzed how discourse markers gradually acquire metatextual functions (e.g., Traugott 1995 on indeed and in fact; Brinton 1996 on Old English gan or Middle English anon; Haselow 2015 on anyway). Through these studies, researchers have established typical pathways of semantic change; they have explored the tendency for expressions to develop subjective and intersubjective meanings (Traugott & Dasher 2002) and have demonstrated that semantic change is brought about by pragmatic inferencing, followed by conventionalization (e.g., Traugott 1989; Traugott & Dasher 2002). These findings have laid the groundwork for my analysis of by the same token. The application of the constructionalization approach to discourse markers has also proved critical to this study (cf. Lewis 2011; Traugott 2017, 2018, 2020, 2022).

Within this larger body of work, it is the emergence of adversative/concessive markers that has garnered the most interest: König and Traugott (1982) worked on yet and still, Lenker (2010:184-197) analyzed the history of though, however, yet, and still, and Lewis studied the evolution of after all (2007), instead (2011), and still (2019). By contrast, the emergence of elaborative markers, as defined by Fraser (2009), has not been treated in great detail, except besides (Traugott 1995) and what is more (Brinton 2008). Traugott’s (2022:104) study on discourse structuring markers mentions by the same token in passing in her chapter on elaborative markers. She states that “whether native (also) or borrowed (alternatively), many [discourse structuring markers] derive from the conceptual domain of similarity and difference (also, alternatively, analogously, by the same token, correspondingly, likewise, in other words, otherwise, similarly).” As I argue here, however, the notion of similarity was not originally part of the meaning of by the same token.

Crucial for my exploration of the development of by the same token are the concepts of “decompositionalization” and “recompositionalization,” which I use to refer to loss of compositionality and increase of compositionality respectively. The term recompositionalization is related to but not synonymous with “folk etymology,” “spelling pronunciation,” and “antilexicalization.” Folk etymology (Förstemann 1852)—also known as “analogical reformation” or “etymological reinterpretation” (Cienkowski 1969)—refers to the process whereby language users turn opaque forms into more familiar ones, based on erroneous analogy. By doing so, they typically increase the compositionality of words. Folk etymology tends to affect loanwords (e.g., penthouse from the French appentis) and obsolete terms (e.g., crayfish from Middle English crevis). It is usually caused by “paronymic attraction,” that is, the influence of quasi-homonyms on existing words. Unlike folk etymology, recompositionalization is not necessarily based on incorrect assumptions. In the case addressed in this paper, the etymology of same is not incorrect. Rather, the original meaning of same (i.e., identification emphasizer; see section 5) was no longer accessible to language users in this construction, which is why they attributed to same its contemporary meaning instead (i.e., similarity; see section 7). The concept of folk etymology would therefore be inadequate in the case of by the same token.²

Recompositionalization also concerns the case of lexicalized compounds whose pronunciation is altered because of their spelling, a phenomenon usually called spelling pronunciation (Algeo 2009:46). This is notably what happened to the compound forehead. The word lost compositionality when it came to be pronounced [ˈfɒrɪd], but many British and American speakers are now reintroducing compositionality to the word by pronouncing it according to its spelling, [ˈfɔː(r)hed] (Jones 1997:196).

Folk etymology and spelling pronunciation only partly overlap with recompositionalization. The first obvious specificity of recompositionalization is that it only concerns compounds, polymorphemic words, or multi-word expressions. Consequently, not all cases of folk etymology exhibit recompositionalization. For instance, the sixteenth-century addition of <s> in island, based on the erroneous notion that it is related to Latin insula, is a case of folk etymology but not of recompositionalization, as the revised spelling does not make island more compositional. Similarly, spelling pronunciation is often disconnected from recompositionalization, as in the case of monomorphemic often when pronounced with an epenthesized [t]. It seems fruitful to make the distinction between cases like this one and cases like forehead, whose altered pronunciation may be motivated by a desire to increase semantic transparency. This is why the term recompositionalization seems better suited than spelling pronounciation to describe such cases.

Folk etymology is a case of antilexicalization, as it goes against the tendency toward greater opacity that is exhibited by lexicalization (Brinton & Traugott 2005:103). The same could be said about the spelling pronunciation of compounds. However, loss of semantic transparency is also a feature of grammaticalization (Brinton & Traugott 2005:105), and the emergence of discourse markers has been treated both as a case of lexicalization and of grammaticalization. This is why the term recompositionalization, which is more neutral and transparent than antilexicalization (or antigrammaticalization), seems more appropriate in the case of by the same token.

3. Material and Method

I have used a variety of corpora to document and analyze the evolution of the low-frequency item by the same token from Early Modern English to Present-Day English. These corpora include the Early English Books Online (henceforth EEBO), the Corpus of Late Modern English Texts 3.0 (henceforth CLMET 3.0), the Old Bailey Proceedings, and the Corpus of Historical American English (COHA). EEBO contains mostly British texts amounting to 755 million words (Davies 2017) and covers the years 1473-1700. CLMET 3.0 (De Smet, Diller & Tyrkkö 2013) is British. It comprises 34 million words and ranges from 1710 to 1920. The Old Bailey Proceedings (Hitchcock, Shoemaker, Emsley, Howard, McLaughlin et al. 2012) include 127 million words collected in London between 1674 and 1913.³ For American English, COHA (Davies 2010) includes 400 million words from 1820 to 2009. The Corpus of Contemporary American English (COCA) (Davies 2008) was also consulted to document ongoing changes (section 8).

The choice of these corpora was dictated by the nature of the study. By the same token is a low-frequency item, necessitating large corpora, and to explore the various stages of its evolution, data covering a long stretch of time was necessary. These two aspects explain why corpora from different geographical origins were selected: a focus on American English would have limited the time span of the study, and a focus on British English would have limited the size of the data. Diachronic British corpora are comparatively small, apart for the Hansard Corpus (Davies 2015). Although this corpus covers roughly the same period as COHA, it yields only one occurrence of by the same token for the whole of the nineteenth century, rendering it unhelpful for examining language change over time.

Since CLMET 3.0 and the Old Bailey Proceedings yielded relatively few examples, I supplemented the data by accessing mainly fictional works through Google Books in November 2021. I collected a sample of around a hundred occurrences from the 1700 to 1900 period. I removed quotes from earlier periods and examples from dictionaries or metatextual comments. Even though Google Books does not qualify as a corpus because it is not representative and normalized, it proved useful to compensate for the paucity of the eighteenth and nineteenth century data. These searches, which were not geographically restricted, also made it possible to uncover a distinctive Irish usage which would not have been found otherwise. One of the main problems with Google Books identified by Pechenick, Danforth, and Dodds (2015) is an over-representation of some specific prolific authors. No such bias was observed in my data.

The lack of diatopic consistency of the data is thus a caveat to bear in mind when interpreting the results presented here. Another caveat is that the corpora selected are not genre-specific. Like most additive markers (Biber, Johansson, Leech, Conrad & Finegan 1999:880; Lenker 2010:45), by the same token is a rather formal marker today, predominantly used in academic English. However, by the same token did not function as an additive marker in Early Modern English and was used in much more varied text types, including narratives. This is why I have not restricted the search to specific genres, as it would have obscured some of the changes at play.⁴

The form of the phrase was initially more varied as well. As such, all the prepositional phrases involving by and the noun token were collected in EEBO and classified according to their internal characteristics (see section 5). The searches were conducted using the strings by this/these [adj.] token(s), by that/those [adj.] token(s), by the [adj.] token(s). All the spellings mentioned in the Middle English Compendium Online (see Lewis et al. 1952-2001) were tested, to ensure that all the relevant data were included, even those with obsolete spellings. The only alternative spellings that yielded results were tokyn(s), tokene(s), taken(s), and takin(s).

Regarding CLMET 3.0, the Old Bailey Proceedings, and COHA, which cover a period when the phrase had reached a higher level of formal fixity, the searches were limited to the string by the same token.⁵ The alternative spellings yielded no results. All the results were analyzed and categorized according to semantic and discursive criteria (sections 5 and 6).

Table 1 gives the number of occurrences of by the same token (including its variants in EEB0) that have been included in the quantitative analyses.

Table 1.

Number of Occurrences of by the Same Token

	1480-1549	1550-1599	1600-1649	1650-1699	1700-1749	1750-1799	1800-1849	1850-1899	1900-1949	1950-2010
EEBO (Including Variants)	47	129	164	278	23
Old Bailey				1	19	1
CLMET 3.0						4	3	2
Google Books					2	16	27	27
COHA							14	15	103	221

4. Overview of the Evolution of by the Same Token

As the evolution of by the same token is quite complex, it seems useful to start with a summary of the various stages that are addressed in detail in sections 5-8. The noun token, meaning ‘proof, sign,’ initially occurred in various evidential prepositional phrases introduced by by. The form that gradually prevailed was a version including same and followed by a noun clause (by the same token that. . .). The word same did not encode similarity but functioned as an identification emphasizer (meaning ‘this one in particular’), as well as a marker of syntactic dependency between the noun token and the noun clause that followed it.

Figure 1 synthesizes the general evolution of by the same token, with one example for each use. Cases of constructionalization are indicated with “CXN” and cases of constructional changes are represented as “CC.” A semantic constructional change (“CC_SEM”) followed by a syntactic constructional change (“CC_SYNT”) also corresponds to a process of constructionalization overall.

Figure 1.

Overall Evolution of by the Same Token

As can be seen in Figure 1 (all examples used in Figure 1 are discussed in subsequent sections), the prepositional phrase expressing the means of access to knowledge developed into a complex subordinator encoding justification. It gave way to two non-compositional uses during the seventeenth century: a digressive discourse marker (meaning ‘incidentally’) and a subordinator combining high degree and consequence. As the phrase was polysemous and had lost its compositionality, it probably appeared opaque to language users. During the nineteenth century, some compositionality was reintroduced to the phrase by attributing to same its contemporary meaning of similarity. This partial recompositionalization is probably what led to the emergence of the elaborative/additive discourse marker. Originally used to connect two consequences of the same premise (‘for the same reason’), it then extended its connective value (‘moreover’). It is now polyfunctional and is even used to connect two contrasting statements (‘conversely’).

5. Early Modern English: The Emergence of by the Same Token (That)

5.1. Evidential Prepositional Phrases

As indicated by the OED (s.v. token), a token is ‘something that serves to indicate a fact, event, object, feeling, etc.; a sign, a symbol,’ and the phrase by the same token (OED, s.v. token, n. def. 15) is a calque translation of French. The corresponding French word is the noun enseigne. It currently refers to a ‘shop sign’ or ‘signboard,’ but it used to mean ‘a piece of evidence,’ a ‘clue that helps the recognition of something’ (TLFi, s.v. enseigne; my translation). It is related to the verb enseigner, ‘to teach’ (TLFi, s.v. enseigne) and comes from the Latin insignia, from in and signum, meaning ‘sign.’ Both the French and the English nouns originally had an evidential meaning, illustrated in (2).

(2) [T]hy blessyd moder behelde the place where as thou stode, and she sawe it replete with thy blode, and she folowynge then knewe where thou had gone by the tokens and steppes of blode, for the grounde where thou had gone appered infuded with thy blode. (EEBO, The fruyte of redemcyon, 1514)

The “tokens” in (2) are what enables the character to have access to knowledge, that is, the sensory stimuli from which past events can be deduced. As such, the noun token is used to express an inferential type of evidentiality. Evidentiality is the expression of the type of evidence for a given statement (Aikhenvald 2004). In many languages, such as Tibetan and Quechua, marking evidentiality is obligatory, and it is expressed through grammatical morphemes. In Germanic and Romance languages, by contrast, the expression of the information source is optional and is mostly coded through lexical items. The use of the term evidentiality in such cases is controversial but is becoming widespread among specialists in Germanic and Romance languages (e.g., Cornillie 2009; Whitt 2010; Nuyts 2017).

Prepositional phrases including the noun token initially had a compositional meaning and they appear in various forms in EEBO. They typically involve a demonstrative, as in (3)-(5); the noun is occasionally modified by an adjective (4) or is pluralized (5).

(3) Mary, saith S. Paul, if he doth his Masters will and commaundement [. . .], then is he a faithful Minister and a trustie Steward. By this shall you know him, by this token, & none other shall you soone discerne, whether he be a faithfull Steward. (EEBO, Seuen godly and learned sermons, 1607)

(4) [Y]ou shoulde not only vnderstande, that you ought not to caste downe your hartes hartes, as though god had forsaken you, but rather that you ought with all youre harte to reioyce, that by this evident token god loueth you and careth for you, in that he tryeth youre pacience. (EEBO, Paraphrase of Erasmus vpon the Newe Testament, 1549)

(5) Whan the boulke is like to suffer any sickenes it is perceiued by these tokens, all the body is in a sweatte, the bulke most specially, the tonge waxeth thicke, the spettyll is eyther salt or bitter, or cholerycke [. . .]. (EEBO, The castel of helth, 1535)

As the presence of by suggests, these prepositional phrases constitute adverbials of means. They are “circumstancial adverbials” (Biber, Johansson, Leech, Conrad & Finegan 1999:776), also known as “adjuncts” (Quirk, Greenbaum, Leech & Svartvik 1985:503), and can therefore be it-clefted, negated, and questioned (Mittwoch, Huddleston & Collins 2002:666-668). Their evidential nature is shown by the co-occurrence of the cognitive verbs know, discerne, understand, and perceive, the epistemic adjectives evident and like, and the presence of an indirect question (whether he be a faithfull Steward). The use of demonstratives and of the plural was frequent until the end of the sixteenth century, resulting from a calque of the French equivalent à ces enseignes, illustrated in (6).

(6) Galaor [. . .] luy demanda quelles armes portoit celuy qui auoit fait tel efort. Vn escu vermeil & deux Lyons de sable, respondit il & autant en a sur son armet, & chevauche vn cheual rouen. A ces enseignes Galaor cogneut que c’estoit celuy qu’il cherchoit [. . .]. (Herberay des Essarts 1555)

‘Galaor asked him what coats of arms the man who had accomplished this feat was bearing. A vermilion shield with two sable lions, he replied; he has the same on his helmet and rides a roan horse. By these tokens Galaor knew that he was the one he was looking for.’ (my translation)

In (6) the cognitive verb cogneut shows that the physical characteristics mentioned in the preceding sentence constitute evidence of the identity of the said knight.

The internal composition of by-token phrases is illustrated in Table 2, which presents the data from the EEBO corpus. As can be seen, the proportion of the phrase by the same token steadily increases at the expense of other forms over the course of the Early Modern English period.

Table 2.

The Internal Composition of Prepositional Phrases with by + Token in EEBO

Prepositional phrases	1480-1549	1550-1599	1600-1649	1650-1699	Total
By Demonstrative token	19% (18)	18.8% (40)	27% (58)	20.7% (69)	185
By Demonstrative Adjective token(s)	2.1% (2)	4.2% (9)	7.9% (17)	3.6% (12)	40
By Demonstrative tokens	12.6% (12)	16% (34)	8.8% (19)	5.1% (17)	82
By the token	3.1% (3)	2.8% (6)	2.3% (5)	5.7% (19)	33
By (the) Adjective token(s)	52.6% (50)	42% (89)	25.1% (54)	21.3% (71)	264
By the tokens	8.4% (8)	10.3% (22)	7.4% (16)	2.7% (9)	55
By the same token	2.1% (2)	5.6% (12)	21.4% (46)	40.8% (136)	196
Total	100% (95)	100% (212)	100% (215)	100% (333)	855

5.2. The Value of Same: Identification Emphasizer and Marker of Syntactic Dependency

In Early Modern English, same typically played the role of an emphasizer (Breban 2010a, 2010b); its function was to stress identification (Ho-Cheong Leung & van der Wurff 2018). It is comparable to the use of the adjective very in Present-Day English and can be glossed by phrases like ‘this one in particular’ and ‘this one precisely.’ It is illustrated in (7).

(7) Þe same nexte nyʒt [. . .] it sempte in þe hiʒe hevene þe cataratis hadde bene vndo.

(MED s.v. same; Lydgate, c1425; cited from Breban [2010a:298])

‘The/that very next night, it seemed that in the high heavens the floodgates had been opened.’

This is the earliest meaning of same; it combines with the definite article to stress the identification of the referent of the noun phrase, and it does not express similarity or equivalence. According to the OED (s.v. same def. A. I. 1. a), this early use of same was prototypically followed by a clause, which suggests that same helped to explicitly signal the presence of the clause to come. This use is illustrated by several Middle English examples; a seventeenth century example is given in (8).

(8) The old saying is as good, Doe that euery day, which thou wouldest doe the same day that thou dyest. (OED, s.v. same; 1631, Earl of Manchester Contemplatio Mortis 89;)

Quite clearly, same is not used to compare two referents in (8). It emphasizes the connection between the noun and the restrictive relative clause that follows it. In the adverbials of means under study, the noun token was also often modified by a that-clause, which probably explains why the use of same became so frequent. The role of same as a marker of syntactic dependency is suggested by the fact that by the same token is followed by a noun clause with an overt that in 46.9 percent of the occurrences retrieved in EEBO (92/196), while by this token is accompanied by this type of clause only 16.5 percent of the time (19/114).⁶ The use of same was therefore a way to highlight the syntactic dependency between the following clause and its antecedent.

Unlike the examples mentioned by the OED, however, the that-clauses used after by the same token are not relative clauses but noun clauses which describe what the evidence of the claim is. This is illustrated in (9).

(9) [T]he capitayne commaundes you by me to open the gates and let them come in / or els gyue me the keyes and I shall let them in / by the same token that he was nat at the watche him selfe this nyght / but sent his sonne: that is trewe quod the porters abyde a lytell / thou shalt haue the keyes incontynent. (EEBO, Jean Froissart’s Chronicles, 1525)

By expressing what the evidence consists in, the speaker provides a justification of his directive speech act. He has to prove that he really is acting on the captain’s behalf, by providing details on the captain’s whereabouts. The fact that the porters’ answer includes the adjective true reinforces the justification interpretation.

When token is followed by a noun clause, it becomes a “shell noun” (Schmid 2000), also known as a “signaling noun” (Flowerdew & Forest 2015). This means that the noun signals “an inherent semantic gap that must be filled by information from the linguistic context” (Schmid 2018:111). In this case, the nature of the evidence is encoded in the noun clause that follows it.

5.3. By the Same Token as a Complex Subordinator

In (9), by the same token that + clause can be interpreted as a prepositional phrase including a nominal clause or as a complex subordinator introducing an adverbial clause. When the use of token as a shell noun became frequent, the construction most likely underwent a clause boundary reanalysis: while the adverbial of means was part of the verb phrase of the main clause—that is, [by [the same token [that sentence]_CLAUSE]_NP]_PP—the new complex subordinator was reanalyzed as introducing a separate subordinate clause—that is, [[by the same token that] [sentence]]_CLAUSE. This phenomenon is frequent crosslinguistically and has contributed to the large inventory of adverbial subordinators in English, such as in case that, in the sense that, and in the hope that (Brems & Davidse 2010). This is also what happened to because, originally by cause that.

The fact that by the same token is no longer a prepositional phrase expressing an adverbial of means but has become a subordinator introducing a justification can be illustrated by the test of the how-question. This test is possible in the case of (5), cited in section 5.1: How is it perceiued whan the boulke is like to suffer any sickenes? > It is perceiued by these tokens, all the body is in a sweatte [. . .], the tonge waxeth thicke [. . .]. It is no longer possible in the case of (9): How does the capitayne commaund you by me to open the gates and let them come in? > *He commands you by the same token that he was nat at the watche him selfe this nyght but sent his sonne.

This change is accompanied by a change of scope: as an adverbial prepositional phrase of means, the phrase was prototypically associated with predicates like know, perceive, show, testify, remember, or be apparent, while all sorts of predicates are compatible with the complex subordinator, which introduces a justification (the proof of this being that). The new syntactic bracketing may be reflected in the presence of a punctuation mark directly before the subordinator. Even though punctuation practices were far from consistent at the time, there is a clear distinction in EEBO between the proportion of punctuation marks before by this token, namely 20.9 percent (24/115), versus by the same token, that is, 94.9 percent (186/196).

The phrase by the same token (that) became a chunk during the seventeenth century, and its progressive formal fixity was accompanied by a decrease in compositionality. The complex subordinator gave rise to two new constructions in which the meaning of token is bleached. The version with an overt that probably gave rise to a subordinator combining high degree and consequence (comparable to ‘so much so that’; see section 5.5), while the version with an omitted that was presumably reanalyzed as a discourse marker with a digressive function (meaning ‘incidentally’; see section 5.4). The fact that the omission of the complementizer can have an impact on the prosodic detachment of what precedes it and ultimately turn it into a parenthetical/discourse marker has been demonstrated for other constructions, such as the “Noun is” construction, that is, (the) fact is/thing is/truth is (Mantlik & Schmid 2018).

5.4. Digressive Discourse Marker (‘Incidentally’)

The inconsistency of Early Modern English punctuation (e.g., Moore 2017:294) coupled with the omissibility of that means that it is difficult to identify with certainty early uses of by the same token as a discourse marker. However, by the late seventeenth century, the that-less version of the complex subordinator had given rise to a new use, in which the notion of evidence or justification is bleached. An example is given in (10).

(10) Up in the morning, and had some red herrings to our breakfast, while my boot-heel was a-mending, by the same token the boy left the hole as big as it was before. Then to horse, and for London through the forest, where we found the way good, but only in one path, which we kept as if we had rode through a canal all the way. (OED, s.v. token; Samuel Pepys’s Diary 1889 [1660])

Quite clearly, if the hole in the boot-heel is as big as it was before, then the fact the hole was still there cannot be taken as a piece of evidence that the mending was completed. A contrastive meaning was not possible at the time (the adversative value emerged three centuries later). The meaning is therefore digressive here. The clause assumes a parenthetical value and is given as an aside. It does not constitute a justification for the repair of the shoe but provides a discursive interruption in the narrative of the day’s events. It is this metatextual function that suggests that by the same token is a discourse marker here and would probably be detached prosodically from the rest of the sentence. The function of by the same token is similar to the digressive markers mentioned by Mittwoch, Huddleston, and Collins (2002:779): by the way, incidentally, and parenthetically. This use may have been favored by an analogy with by the way and by the by, which developed similar uses around the same time (see Traugott 2020). This tends to confirm Traugott’s (2022:50) hypothesis on the emergence of a digressive subschema of connectors based on generalizations made over “by the x” microconstructions.

The digressive use of by the same token persisted until the nineteenth century and had several metatextual functions, including that of topic shifter, as in (11), where it introduces a new stretch of discourse only loosely connected with the main point of the discussion.

(11) “Who did you get them [= pound notes] from, Lanty?” said the O’Donoghue.

“Malachi Glynn, sir, of Cabernavorra, and, by the same token, I got a hearty laugh at the same house once before.”

“How was that?” said the old man, for he saw by the twinkle of Lanty’s eye that a story was coming. (Google Books; Lever 1858)

The reaction of the old man (“he saw by the twinkle of Lanty’s eye that a story was coming”) clearly attests to the change in topic introduced by by the same token. The connection between the upcoming story and the preceding topic—that is, financial negotiations for a horse sale—is very loose and only hinges on the mention of a particular place. Following Traugott and Trousdale’s (2013) graphic convention, the syntactic and semantic constructionalization that can be observed so far may be formalized as such, where the double arrows represent the mapping between the form (on the left) and the meaning (on the right): [prepositional phrase ⬄ means of access to knowledge] > [discourse marker ⬄ digression].

The digressive use is considered by the OED as a bleaching of the evidential meaning: by the same token “introduce[es] a corroborating circumstance, often weakened down to a mere associated fact that helps the memory or is recalled to mind by the main fact” (s.v. token, n. def. 15). The link between the digressive discourse marker and the evidential subordinator revolves around the notion of remembrance, which is one of the cognitive processes often associated with evidential expressions (see section 5.3.). In the metatextual use, the notion of evidence is backgrounded while the notion of memory remains, enriched by a digressive pragmatic function. According to Traugott (2020:126), this development reflects an incipient subjectification, as this meaning “become[s] increasingly based in the speaker’s subjective belief state/attitude toward the proposition” (Traugott 1989:35). The digressive nature of the discourse marker is said to imply an evaluation on the part of the speaker, that is, that the points introduced by digressive markers are considered less important than the preceding discourse. They are only tangentially connected to the previous discourse and do not contribute to the overall pragmatic function of the conversation (buying a horse in this case).

The digressive use persisted in non-standard dialects (Ogilvie 1883) and even became characteristic of Irish English (Joyce 1910:228; Clark 1917:34). The majority of the nineteenth century examples found in my corpus (12/22) are produced by Irish speakers. It even became stereotypically associated with rural Irish English, as illustrated in (12).

(12) The little anecdotes introduced here [. . .] will remind the reader of those odd collaterals which are sometimes introduced in the stories or arguments of our Irish peasantry with a preface of ‘by the same token,’ or something equally inappropriate. (The Dublin University Magazine 1843:473).

By the early twentieth century, this use was said to be “extinct outside of Ireland” (Clark 1917:34).

5.5. Subordinator Combining High Degree and Consequence (‘So Much So That’)

The evidential subordinator did not only give rise to the now-extinct digressive discourse marker. The version with an overt that remained a subordinator but underwent a semantic shift at the turn of the eighteenth century. Possibly following the evolution of its French counterpart (à telle(s) enseigne(s) que), by the same token that started to express the fact that a property or event reaches such a high degree that it triggers a particular consequence. The clauses introduced by such subordinators combine the notion of high degree with the notion of result (Quirk, Greenbaum, Leech & Svartvik 1985:1142-1143).

The semantic shift from evidence to high degree and result is understandable, as many actions can be attested to only through their consequences. The potential ambiguity between evidence and result is perceptible in (13); the ambiguity is clarified in the paraphrases in (14), which is evidential, and (15), which is high degree + consequence.

(13) [My] rump-bone sticks out like the ace of spades, by the same token that it perforates all my breeches[.] (EEBO, The Comical Works of Scarron, 1700)

(14) [My] rump-bone sticks out like the ace of spades, the proof being that it perforates all my breeches[.]

(15) [My] rump-bone sticks out like the ace of spades, so much so that it perforates all my breeches[.]

These manipulations show how the evidential subordinator could be interpreted as a marker of high degree and consequence: the holes in the breeches could be both the proof and the result of the shape of the man’s coccyx.

This shift from a variety of evidential prepositional phrases, functioning as means adverbials, to a complex subordinator whose meaning combines high degree and consequence is a case of constructionalization. It can be formalized as follows: [prepositional phrase ⬄ means of access to knowledge] > [complex subordinator ⬄ high degree + consequence].

Table 3 illustrates the semantic values of by the same token in the eighteenth-century portion of my corpus. As can be seen, the proportion of evidential uses decreases while those of the digressive meaning and the ‘high degree and consequence’ meaning increase.

Table 3.

Semantic Evolution in the Eighteenth-Century Section of the Corpus

	1700-1749	1750-1799
Evidential	75% (33)	55% (11)
Digressive	13.6% (6)	20% (4)
High Degree + Consequence	11.4% (5)	25% (5)
Consequence	0% (0)	5.9% (1)
Total	100% (44)	100% (21)

6. Late Modern English

6.1. Nineteenth Century: Partial Recompositionalization

As can be seen in Table 3, by the same token was polysemous in the late eighteenth century. On average, the original evidential meaning still accounted for almost 55 percent of cases. The digressive discourse marker remained in use (around 20 percent in my corpus), but was quite possibly deemed stereotypically Irish and rural (cf. Clark 1917:34, and example (12) above). It was therefore unlikely to gain ground in the standard language. The subordinator expressing high degree and consequence was still in use (25 percent in the current materials), but was also the source construction for a discourse marker, through omission of that and prosodic detachment. This new discourse marker was more or less synonymous with consequently, illustrated in (16).

(16) The blaze there has thawed all the snow from your cloak; by the same token, it has streamed on to my floor, and made it like a trampled street. (CLMET 3.0, Charlotte Brontë, Jane Eyre, 1847)

In (16), the punctuation, which had reached a higher level of standardization by the nineteenth century, suggests that by the same token is prosodically detached from the rest of the sentence. Its meaning highlights the chain of cause and effect that led to a wet floor. The blaze is the cause; its consequence is the melting of the snow; the consequence of the melting of the snow is the flooded floor. For this reason, by the same token can be labeled an inferential discourse marker in this example, following Fraser’s (2009:296) terminology.

In view of the variety of meanings encoded by by the same token, speakers were not always able to interpret it adequately:

[By the same token] is common enough in Scott, e.g., Anne of Geierstein, vol. iii p. 259, ed. 1829, where Sigismund, describing the rout of Charles’s army and the contents of his camp, says: “Here were hosts of laced lackeys, and grooms, and pages [&c.]; and thousands, for what I know, of pretty maidens. By the same token, both menials and maidens placed themselves at the disposal of the victors,” &c. Here the phrase seems to mean, “upon the same testimony”; but I doubt. (Google Books; C.M.I. Athenænum Club, Notes and Queries: A Medium of Intercommunication for Literary Men, General Readers, 1882)

The polysemy is problematic, because discourse management markers “have to be as unambiguous metalinguistic indicators of the text structure as possible, since they explicitly state the voice of the speaker, their opinion on the state of affairs” (Lenker 2010:140). The polysemy probably explains why partial recompositionalization occurred. Because of the difficulty of interpreting the phrase and because of the obsolescence of the original meaning of token and of same (‘this one in particular’), speakers reintroduced some compositionality to the phrase by assigning to same the meaning of similarity that it had elsewhere. This evolution may have also been influenced by the existence of the additive connector in the same way, which was coined in the early nineteenth century (Lenker 2010:221). This new usage led to examples in which by the same token is used to introduce the second consequence of a state of affairs, as in (17).

(17) I did not wish to let her know that her father had died in utter poverty, [therefore I assumed the office of executor, deceiving her, but with no evil intention, into the notion that some wreck remained of her father’s property]_C1. [By the same token also I paid the wages of that sad varlet Peter Longstaff]_C2 [. . .]. (Google Books; Scargill, 1833)

In this new usage, by the same token is used to connect two comparable statements. The passage is clearly organized, and the discourse marker is used to link two consequences of the same premise. The speaker’s desire to hide the father’s poverty brings about consequence one, marked C1 in (17), introduced by therefore. By the same token is then used to introduce the second logical consequence of the same premise, marked C2 in (17). The connection between the two results is highlighted by the adverb also.

The meaning of the discourse marker, ‘connection between two consequences,’ remains in Present-Day English, although it is becoming rarer (see Table 6 in section 8.2). This meaning is reflected in many dictionary definitions today, which include for the same reason as a paraphrase for by the same token (e.g., Cobuild dictionary; Cambridge dictionary, s.v. token).⁷

Table 4 presents the semantic evolution of by the same token in the nineteenth-century section of my corpus. It illustrates the burgeoning of the ‘second consequence’ meaning (‘for the same reason’) and of the elaborative meaning (‘similarly’), as well as the decrease of most other meanings.

Table 4.

Semantic Evolution in the Nineteenth-Century Section of the Corpus

	1800-1849	1850-1899
Evidential	45.4% (20)	34.1% (15)
Digressive	27.3% (12)	22.7% (10)
High Degree + Consequence	2.3% (1)	4.5% (2)
Consequence	9.1% (4)	6.8% (3)
2nd Consequence	6.8% (3)	13.6% (6)
Elaborative	2.3% (1)	13.6% (6)
Other (Cf. Section 8.2)	0% (0)	4.5% (2)
Total	100% (41)	100% (44)

6.2. Twentieth Century: Additive/Elaborative Marker

After connecting two consequences, the discourse marker started to be used to link more varied stretches of connected discourse. It can now be paraphrased by similarly or moreover. Semantic broadening (here from ‘similarity between two consequences’ to ‘similarity’) is typical of this type of connectives (see Pinson 2018). By the same token can now connect any type of comparable ideas; its role is crucial in the organization of the passage in (18), which enumerates alleged reasons why women may want to have children.

(18) First let us take the imaginative woman who is deeply in love with her husband. She will want a child as a completion or a “symbol” of the passion act, a poetic aspiration. The possessive instinct will also make her want a small edition of her beloved, which she can be sure of possessing wholly and for all time. On the other hand, the woman whose husband shows dangerous signs of philandering will seize upon a child as the obvious means of holding him. By the same token, the woman who is not greatly in love with her husband may crave a child as a substitute emotional outlet. (COHA, Magazine, 1929)

In (18), each type of woman is introduced by a discourse connector: first, on the other hand, and by the same token. By the same token cannot be paraphrased by for the same reason here, as it does not introduce the second consequence of a previously-mentioned premise. It is used to connect comparable sub-cases, as part of an enumeration within a more general topic. The widening connective value of the discourse marker explains why some dictionaries consider the phrase “a general way to connect statements that have some logical association with one another” (Ammer 1997, s.v. by the same token).

The development of the additive/elaborative value of the discourse marker went hand in hand with an increase in initial position placement. Initial position is defined as the very beginning of a sentence/independent clause or after a sentence-initial coordinating conjunction; final position is located at the very end of a sentence. Any other position is considered medial (cf. e.g., Quirk, Greenbaum, Leech & Svartvik 1985:643; Biber, Johansson, Leech, Conrad & Finegan 1999:891; Lenker 2014). Cases of medial position are varied. They notably include three of the positions outlined by Greenbaum (1969): between auxiliary and verb, between be and its complement, and between a transitive verb and its object. Other cases considered medial here are not mentioned in Greenbaum’s typology. Most of the time, medial occurrences of by the same token are related to coordination. Any case of coordination where by the same token does not introduce a fully independent clause is considered medial for the purpose of this study.

As can be seen in Table 5, the percentage of initial placement of by the same token has surged from 7.1 percent to 76.5 percent in 150 years. The increase in initial placement is consistent with the general observation that linking adverbials favor initial position (Biber, Johansson, Leech, Conrad & Finegan 1999:770-774), which tends to facilitate discourse processing (Lenker 2010:44). The change from clause-initial position to sentence-initial position correlates with the increase in scope of by the same token: while it only had scope over a clause when it was a subordinator, it now has scope over a whole sentence or even a larger discourse segment.

Table 5.

Evolution of the Position of by the Same Token in COHA

Position	1800-1849	1850-1899	1900-1949	1950-2010
Initial	7.1% (1)	40% (6)	62.1% (64)	76.5% (169)
Medial	85.7% (12)	60% (9)	35.9% (37)	23.5% (52)
Final	7.1% (1)	0% (0)	1.9% (2)	0% (0)
Total	100% (14)	100% (15)	100% (103)	100% (221)

7. A Polyfunctional Discourse Marker

The extended notion of similarity which is associated with the phrase has turned it into a versatile connective. It is still compatible with the notion of consequence but is increasingly associated with two contrasting statements.

7.1. Consequence

The discourse marker remains compatible with a result interpretation (‘consequently’), which is frequently found in non-sentence-initial position, as in (19).

(19) In his late work Davis continually experimented with design and technique. [. . .]. Only half obscured by the tangle of trees which he may depict, there rise the great slanting structural lines of the hilly formations which evidently have a great attraction for him. They introduce a sharply rhythmical element into his compositions and, by the same token, give to these a new unity and a new force. (COHA, Magazine, 1995)

This passage depicts the work of Charles Harold Davis (1856-1933), who was famous for his landscape paintings. The argument is that “the great slanting structural lines of the hilly formations” constitute a “sharply rhythmical element,” which results in “a new unity and a new force.” The compositional strength of the painting is clearly presented as the consequence of the presence of the slanting hills.

The data gathered from COHA show that the initial position is favored when the discourse marker is additive/elaborative (by a ratio of 161:43). When the discourse marker can have a result interpretation, however, the non-sentence-initial position is preferred (by a ratio of 26:12). This probably results from an analogy with therefore, which is the second most frequent adverbial connector in medial position (after however) (Lenker 2014:24).

7.2. Contrast

Additionally, the discourse marker is becoming increasingly compatible with two contrasting statements. The notion of contrast arose through the connection between two contrasting consequences of the same premise, as in (20), which addresses two sides of a healthcare reform, its advantageous consequences, and their disadvantageous counterparts.

(20) Under the new system, a hospital can keep all the money it saves by holding its costs below the prices fixed by the Federal Government. By the same token, a hospital stands to lose every dollar it spends on a Medicare patient in excess of the amount set by the Government. (COHA, News, 1983)

This new use emerged in the late twentieth century: the most frequent left collocate of by the same token in COHA was and, but in COCA it is but (COHA: but 34 percent, N = 20 versus and 66 percent N = 38; COCA: but 75 percent, N = 162 versus and 25 percent, N = 46). This shows that the discourse marker is increasingly compatible with the notion of contrast, but it also suggests that the adversative meaning is context-dependent, appearing mostly when other cues also indicate contrast. The adversative notion is likely due to pragmatic inference and not yet part of the semantics of the discourse marker per se. In fact, few dictionaries mention this meaning, apart from Spears (2006, s.v. by the same token): by the same token is “a phrase indicating that the speaker is introducing parallel or closely contrasting information.” Conversely, many contributors to online fora mention the adversative value of by the same token: “It is usually used to express that something else is also true, especially something different and surprising,” and “It’s used to identify corollary, often contrasting outlooks on the same circumstance or issue.”⁸ The fact that speakers give definitions that differ from those provided by most dictionaries may suggest that the adversative meaning of by the same token is not yet fully semanticized and conventionalized. The adversative value still relies on additional cues and is not felt by all users to be part of the semantics of the phrase itself. The adversative meaning of by the same token may have arisen due to the influence of other adversative markers including same, such as at the same time (coined in the eighteenth century), and all the same (late nineteenth century) (see Lenker 2010:180).

Semantic shifts from a temporal to an adversative meaning are very frequent and have been studied in detail (e.g., Lenker 2010:109). But other types of connectors can also give rise to adversative values, as is the case of the originally spatial subordinator whereas (see Lenker [2010:112] for a list of originally spatial adverbial connectors, like anyway). These shifts typically involve “an inference from co-existence in time (while) and space (whereas) to juxtaposition and contrast” (Traugott 2012:555) and reflect subjectification. The case of by the same token is quite similar, as it is the concomitance between two facts that leads to the adversative value (König 1985:269). The overall semantic evolution of by the same token is presented in Table 6; these data are also reproduced in Figure 2 for the sake of readability.

Figure 2.

The Overall Semantic Evolution of by the Same Token in the Eighteenth Century through Twentieth Century

Table 6.

The Overall Semantic Evolution of by the Same Token in the Eighteenth Century through Twentieth Century

Meanings	1700-1749	1750-1799	1800-1849	1850-1899	1900-1949	1950-2010
Evidential	75% (33)	55% (11)	45.4% (20)	34.1% (15)	2.8% (3)	0% (0)
Digressive	13.6% (6)	20% (4)	27.3% (12)	22.7% (10)	8.3% (9)	0% (0)
High degree + consequence	11.4% (5)	25% (5)	2.3% (1)	4.5% (2)	0% (0)	0% (0)
Consequence	0% (0)	5.9% (1)	9.1% (4)	6.8% (3)	15.7% (17)	8.1% (18)
2nd consequence	0% (0)	0% (0)	6.8% (3)	13.6% (6)	7.4% (8)	2.2% (5)
Elaborative	0% (0)	0% (0)	2.3% (1)	13.6% (6)	53.7% (58)	63.7% (142)
Adversative	0% (0)	0% (0)	0% (0)	4.5% (2)	12% (13)	26% (58)
Total	100% (44)	100% (21)	100% (41)	100% (44)	100% (108)	100% (223)

As can be seen in Table 6 and Figure 2, the data suggest that there has been complete semantic renewal of by the same token within less than two centuries. The three meanings that prevailed during the eighteenth century (evidential, digressive, and high degree + consequence) appear to have obsolesced. After a period of complex polysemy during the nineteenth century (up to seven different senses), by the same token was used primarily for the elaborative value, with the recent addition of the adversative meaning.

7.3. Constructional Interference

As Figure 3 shows, there was a peak in the overall use of by the same token during the 1960s; its use has gradually receded since then.⁹

Figure 3.

Frequency Per Million Words of by the Same Token in COHA

In fact, there are signs that the phrase may be unfamiliar to some present-day English users. Although rare, altered forms such as at the same token, in the same token, and on the same token are attested in COCA (seventeen, fourteen, and thirty-one occurrences respectively), especially in the spoken sections. This is presumably due to constructional interference with other prepositional phrases functioning as discourse markers, such as at the same time and on the other hand. Like these two discourse markers, the alternative versions of by the same token have a high tendency to express contrast. In COCA, 82.3 percent (14/17) of occurrences of at the same token and 74.2 percent (23/31) of occurrences of on the same token connect two contrasted statements, as in (21) and (22).

(21) Being positive doesn’t preclude the odd obscenity-laced motivational speech, according to Billick. # “You have to dog-cuss them a little bit to make sure they understand what it takes to win,” he said. “But at the same token, it has to be in a teaching mode. You have to be comforting to a degree but not to the point where you condone it.” (COCA, News, 2002)

(22) I think if someone is just blindly confident from birth, and you never get to sort of see where that came from or what challenges brought the character to that place, then you don’t necessarily find a way into them or a way to relate to them or to love them. And on the same token, I don’t think you necessarily want to see a character that’s purely weak and terrified and never can overcome that to accomplish anything. (COCA, Spoken, 2019)

The constructional interference is illustrated in several educational websites, which prescribe against confusion of by the same token with on the other hand, as, for example, in (23).

(23) When we compare things with each other, we often say “on the one hand” and “on the other hand.” These phrases mean “on this side” and “on the other side.” But it is a mistake to say “on the same token,” meaning “in the same regard.” The standard expression is “by the same token.” (https://brians.wsu.edu/2016/05/19/on-the-same-token-by-the-same-token/; accessed 14 August 2020)

The spoken section of COCA also includes hesitations (24) and hybrid correlative constructions (25) which confirm that by the same token is now strongly associated with at the same time and on the other hand in speakers’ minds.

(24) SMITH: They have ombudsmen, or people like that, to do that sort of stuff, that – they – maybe that would save us having this discussion. But at the same time. . . Ms. STASI: More court time. More court time. SMITH: Well, you’re right. But will this one end up on Court TV? By the same time – or by the same token, isn’t a man punished if a man is a workaholic[.] (COCA, Spoken, 1995)

(25) I had affection on one hand from somebody I wanted to care desperately, and on the same token, it was – I finally had control over what was going on. (COCA, Spoken, 1993)

As an adversative discourse marker, by the same token becomes semantically opaque once more, as none of its components indicates that it can be interpreted as meaning ‘conversely.’ As a new member of the constructional network of adversative discourse markers, it becomes influenced by other adversative discourse markers with the same composition, thereby initiating a nascent subschema: {[preposition + definite article + similarity/otherness adjective + noun] ⬄ adversative DM}.

8. Conclusion

Even if the data lack in diatopic and generic consistency, a clear picture of the trajectory of by the same token seems to emerge. The phrase by the same token appears to have gone through several phases of constructionalization and constructional change. When Early Modern English users wished to explain how they had inferred knowledge, they could resort to a variety of evidential prepositional phrases involving the noun token and functioning as means adverbials. The noun token in these prepositional phrases was either singular or plural; it could be modified by any adjective and was typically determined by a demonstrative. As the noun token was often followed by a noun clause expressing what the evidence consisted of, speakers gradually favored the use of a definite article and of the adjective same, which acted both as an identification emphasizer and as an explicit marker of syntactic dependency. In this use, by the same token that seems then to have been reanalyzed as a complex subordinator, meaning the proof being that, via a rebracketing of clause boundaries. In turn, the complex subordinator began to undergo a process of decompositionalization during the seventeenth century. The version with an omitted that constructionalized into a discourse marker with digressive functions which appears to have persisted longer in rural Irish dialects, while the subordinator with an overt that shifted toward the expression of high degree and consequence.

At this stage, by the same token had lost its compositionality and must have appeared rather opaque to language users, notably because same had long lost its identification emphasizing value. Users therefore reintroduced some compositionality to the phrase by assigning a similarity meaning to the adjective same. This partial recompositionalization is likely what brought about the emergence of the additive/elaborative discourse marker. Originally used to connect two consequences of the same premise, the discourse marker then extended its connective value and increasingly appeared in sentence-initial position. It is now increasingly associated with an adversative meaning, based on an inference from concomitance to contrast. This new meaning likely remains for the moment at the level of pragmatic inferences and is not yet fully semanticized. The discourse marker is now rather infrequent and undergoes interference from other prepositional phrases, which suggests that it is now perceived by many as a member of a constructional network of adversative discourse markers.

The contemporary use of by the same token is the result of a complex history of syntactic and semantic changes, making it distinct among equative connectives like in the same way, in the same vein, or similarly. Its history involves calque, several phases of constructionalization and constructional change, decompositionalization, and partial recompositionalization. The semantics of the phrase is constantly caught between two opposing forces, the drive toward chunking on the one hand and the preference for semantic transparency on the other. These two tendencies have opposite results and yet they both stem from the same underlying purpose: facilitating cognitive processing and storage.

It is hoped that this case study has helped to shed some light on the processes at play in the emergence of discourse markers, and that “decompositionalization” and “recompositionalization” can add to the repertoire of concepts used for describing language change.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Mathilde Pinson

Notes

Author Biography

Mathilde Pinson is a senior lecturer in English linguistics at Sorbonne Nouvelle University, France, specializing in English historical linguistics, construction grammar, and language variation and change. She has written articles on the representation of Scottish specificities in literature, the evolution of verbal complementation, the constructionalization of complementizers, the grammaticalization of adverbial subordinators, and the emergence of discourse markers.

References

Davies

Mark

. 2008. The corpus of contemporary American English (COCA): One billion words, 1990-2019. https://www.english-corpora.org/coca/

Davies

Mark

. 2010. The corpus of historical American English: 400 million words, 1810-2009. http://corpus.byu.edu/coha/

Davies

Mark.

2015. Hansard corpus. https://www.english-corpora.org/hansard/

Davies

Mark

. 2017. Early English books online corpus. https://www.english-corpora.org/eebo/

De Smet

Hendrik

Diller

Hans-Jürgen

Tyrkkö

Jukka

. 2013. A corpus of Late Modern English texts (version 3.0). Department of Linguistics, University of Leuven.

Hitchcock

Tim

Shoemaker

Robert

Emsley

Clive

Howard

Sharon

McLaughlin

Jamie

, et al. The Old Bailey proceedings online, 1674-1913 (www.oldbaileyonline.orgversion 7.0, 8 March 2012). Accessed on 7 August, 2021.

Huber

Magnus

Nissel

Magnus

Puga

Karin

. 2016. Old Bailey corpus 2.0. hdl:11858/00-246C-0000-0023-8CFB-2

Aikhenvald

Alexandra

. 2004. Evidentiality. Oxford: Oxford University Press.

Algeo

John

. 2009. The origin and development of the English language. 6th edn. Boston: Wadsworth.

10.

Ammer

Christine

. 1997. The American heritage dictionary of idioms. Boston: Houghton Mifflin Harcourt.

11.

Biber

Douglas

Johansson

Stig

Leech

Geoffrey

Conrad

Susan

Finegan

Edward

(eds.). 1999. Longman grammar of spoken and written English. Harlow: Pearson Education.

12.

Breban

Tine

. 2010a. English adjectives of comparison: Lexical and grammaticalized uses. Berlin: Mouton de Gruyter.

13.

Breban

Tine

. 2010b. Reconstructing paths of secondary grammaticalisation of same from emphasizing to phoricity and single-referent-marking postdeterminer uses. Transactions of the Philological Society 108. 68-87.

14.

Brems

Liselotte

Davidse

Kristin

. 2010. Complex subordinators derived from noun complement clauses: Grammaticalization and paradigmaticity. Acta Linguistica Hafniensia: International Journal of Linguistics 42(1). 101-116.

15.

Brinton

Laurel

. 1996. Pragmatic markers in English: Grammaticalization and discourse functions. Berlin: Mouton de Gruyter.

16.

Brinton

Laurel

. 2008. The comment clause in English: Syntactic origins and pragmatic developments. Cambridge: Cambridge University Press.

17.

Brinton

Laurel J.

Traugott

Elizabeth Closs

. 2005. Lexicalization and language change. Cambridge: Cambridge University Press.

18.

Butterfield

Jeremy

(ed.). 2015. Fowlers’ dictionary of Modern English usage. 4th edn. Oxford: Oxford University Press.

19.

Cambridge Dictionary . Cambridge: Cambridge University Press. https://dictionary.cambridge.org/fr/dictionnaire/anglais/by-the-same-token (12 August, 2020).

20.

Cienkowski

Witold

. 1969. The initial stimuli in the processes of etymological reinterpretation (so-called folk etymology). Scando-Slavica 15(1). 237-245.

21.

Clark

James Midgley

. 1917. The vocabulary of Anglo-Irish. St Gall: Zollikofer & Cie.

22.

Cobuild Advanced English Dictionary . HarperCollins. https://www.collinsdictionary.com/dictionary/english/by-the-same-token (14 August, 2020).

23.

Cornillie

Bert

. 2009. Evidentiality and epistemic modality. Functions of Language 16(1). 44-62.

24.

The Dublin University Magazine . 1843. Dublin: William Curry Jun. and Company.

25.

Flowerdew

John

Forest

Richard W.

2015. Signalling nouns in English: A corpus-based discourse approach. Cambridge: Cambridge University Press.

26.

Förstemann

Ernst

. 1852. Über deutsche Volksetymologie. Zeitschrift für vergleichende Sprachforschung auf dem Gebiete des Deutschen, Griechischen und Lateinischen 1(1). 1-25.

27.

Fraser

Bruce

. 1990. An approach to discourse markers. Journal of Pragmatics 14(3). 383-395.

28.

Fraser

Bruce

. 2009. An account of discourse markers. International Review of Pragmatics 1(2). 293-320.

29.

Goldberg

Adele

. 1995. Constructions: A construction grammar approach to argument structure. Chicago/London: University of Chicago Press.

30.

Greenbaum

Sidney

. 1969. Studies in adverbial usage. London: Longmans.

31.

Haselow

Alexander

. 2015. Left vs. right periphery in grammaticalization: The case of anyway. In Smith

Andrew D. M.

Trousdale

Graeme

Waltereit

Richard

(eds.), New directions in grammaticalization research, 157-186. Amsterdam: John Benjamins.

32.

Herberay des Essarts

Nicolas

. 1555. Amadis de Gaule. Paris: Groulleau.

33.

Hilpert

Martin

. 2013. Constructional change in English: Developments in allomorphy, word formation, and syntax. Cambridge: Cambridge University Press.

34.

Ho-Cheong Leung

Alex

van der Wurff

Wim

. 2018. Anaphoric reference in Early Modern English: The case of said and same. In Leung

Alex Ho-Cheong

van der Wurff

Wim

(eds.), The noun phrase in English: Past and present, 143-186. Amsterdam: John Benjamins.

35.

Jones

Daniel

. 1997. English pronouncing dictionary. 15th edn. Cambridge: Cambridge University Press.

36.

Joyce

Patrick Weston

. 1910. English as we speak it in Ireland. London: Longmans, Green & Co.

37.

König

Ekkehard

. 1985. Where do concessives come from? On the development of concessive connectives. In Fisiak

Jacek

(ed.), Historical semantics, historical word-formation, 263-282. Berlin: Mouton de Gruyter.

38.

König

Ekkehard

Traugott

Elizabeth Closs

. 1982. Divergence and apparent convergence in the development of yet and still. In Macaulay

Monica

Gensler

Orin

(eds.), Proceedings of the eighth annual meeting of the Berkeley Linguistics Society, 170-179. Berkeley, CA: University of California Berkeley.

39.

Lenk

Uta

. 1998. Marking discourse coherence: Functions of discourse markers in spoken English. Tübingen: Narr.

40.

Lenker

Ursula

. 2010. Argument and rhetoric: Adverbial connectors in the history of English. Berlin: Mouton de Gruyter.

41.

Lenker

Ursula

. 2014. Knitting and splitting information: Medial placement of linking adverbials in the history of English. In Pfenninger

Simone E.

Timofeeva

Olga

Gardner

Anne-Christine

Honkapohja

Alpo

Hundt

Marianne

Schreier

Daniel

(eds.), Contact, variation, and change in the history of English, 11-38. Amsterdam: John Benjamins.

42.

Lever

Charles James

. 1858. The O’Donoghue: A tale of Ireland fifty years ago. London: Chapman and Hall.

43.

Lewis

Diana

. 2007. From temporal to contrastive and causal: The emergence of connective after all. In Celle

Agnès

Huart

Ruth

(eds.), Connectives as discourse landmarks, 89-99. Amsterdam: John Benjamins.

44.

Lewis

Diana

. 2011. A discourse-constructional approach to the emergence of discourse markers in English. Linguistics 49(2). 415-443.

45.

Lewis

Diana

. 2019. Grammaticalizing adverbs of English: The case of still. In Núñez-Pertejo

Paloma

López-Couso

María José

Méndez-Naya

Belén

Pérez-Guerra

Javier

(eds.), Crossing linguistic boundaries: Systemic, synchronic and diachronic variation in English, 127-150. London: Bloomsbury.

46.

Lewis

Robert E.

, et al. 1952-2001. Middle English dictionary. Ann Arbor: University of Michigan Press. Online edition in Middle English compendium. McSparran

Frances

, et al. (eds.), 2000-2018. Ann Arbor: University of Michigan Library. http://quod.lib.umich.edu/m/middle-english-dictionary/ (4 September, 2020).

47.

Lutzky

Ursula

. 2006. Discourse markers? Well. . . Views 15(1). 3-24.

48.

Mantlik

Annette

Schmid

Hans-Jörg

. 2018. That-complementiser omission in N + be + that-clauses. In Ho-Cheong Leung

Alex

van der Wurff

Wim

(eds.), The noun phrase in English: Past and present, 187-222. Amsterdam: John Benjamins.

49.

Mittwoch

Anita

Huddleston

Rodney

Collins

Peter

. 2002. The clause adjuncts. In Huddleston

Rodney

Pullum

Geoffrey K.

(eds.), The Cambridge grammar of the English language, 663-784. Cambridge: Cambridge University Press.

50.

Moore

Colette

. 2017. Literary language. In Bergs

Alexander

Brinton

Laurel

(eds.), The history of English, vol. 4, Early Modern English, 287-308. Berlin: Mouton de Gruyter.

51.

Notes and queries: A medium of intercommunication for literary men, general readers, etc . 134, 1882. London: G. Bell.

52.

Nuyts, Jan. 2017. Evidentiality reconsidered. In Juana I.

Marín-Arrese

Haßler

Gerda

Carretero

Marta

(eds.), Evidentiality revisited: Cognitive grammar, functional and discourse-pragmatic perspectives, 57-83. Amsterdam: John Benjamins.

53.

Ogilvie

John

. 1883. The imperial dictionary of the English language new edition. London: Blackie & Son.

54.

OED = Oxford English Dictionary . Online edition. Oxford: Oxford University Press.

55.

Pechenick

Eitan Adam

Danforth

Christopher M.

Dodds

Peter

. 2015. Characterizing the Google Books Corpus: Strong limits of socio-cultural and linguistic evolution. PLoS One 10(10): e0137041.

56.

Pepys

Samuel

. 1889 [1660]. Diary and correspondence of Samuel Pepys. New Edition. London: George Bells and Sons.

57.

Pinson

Mathilde

. 2018. Constructionalization of in the same vein: From an intrapredicative manner adverbial to a topic transition marker. Paper presented at the LPTS Conference, Linguistic approaches to text structuring, 18-20 January 2018, ENS, Paris, France.

58.

Quirk

Randolph

Greenbaum

Sidney

Leech

Geoffrey

Svartvik

Jan

. 1985. A comprehensive grammar of the English language. London: Longman.

59.

Scargill

M. A.

1833. The Puritan’s Grave. London: Saunders & Otley.

60.

Schiffrin

Deborah

. 1987. Discourse markers. Cambridge: Cambridge University Press.

61.

Schmid

Hans-Jörg

. 2000. English abstract nouns as conceptual shells: From corpus to cognition. Berlin: Mouton de Gruyter.

62.

Schmid

Hans-Jörg

. 2018. Shell nouns in English–a personal roundup. Caplletra, Revista Internacional de Filologia 64(1). 109-128.

63.

Schourup

Lawrence C.

1985. Common discourse particles in English conversation. New York, NY: Garland.

64.

Schourup

Lawrence C

. 1999. Discourse markers. Lingua 107(3-4). 227-265.

65.

Spears

Richard A.

2006. McGraw-Hill dictionary of American idioms and phrasal verbs. New York, NY: McGraw-Hill.

66.

TLFI Dictionary = Trésor de la langue française informatisé, http://www.atilf.fr/tlfi, ATILF - CNRS & Université de Lorraine (10 August, 2020).

67.

Traugott

Elizabeth Closs

. 1989. On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 65(1). 31-55.

68.

Traugott

Elizabeth Closs

. 1995. The role of the development of discourse markers in a theory of grammaticalization. Paper presented at International Conference on Historical Linguistics (ICHL) 12, Manchester.

69.

Traugott

Elizabeth Closs

. 2012. Pragmatics and language change. In Allan

Keith

Jaszczolt

Kasia

(eds.), Cambridge handbook of pragmatics, 539-565. Cambridge: Cambridge University Press.

70.

Traugott

Elizabeth Closs

. 2017. A constructional exploration into “clausal periphery” and the pragmatic markers that occur there. In Onodera

Noriko

(ed.), Periphery: Where pragmatic meaning is negotiated, 55-73. Tokyo: Hituzi Syobo.

71.

Traugott

Elizabeth Closs

. 2018. Modeling language change with constructional networks. In Bordería

Salvador Pons

Lamas

Óscar Loureda

(eds.), Beyond grammaticalization and discourse markers, 17-50. Leiden: Brill.

72.

Traugott

Elizabeth Closs

. 2020. The development of “digressive” discourse-topic shift markers in English. Journal of Pragmatics 156. 121-135.

73.

Traugott

Elizabeth Closs

. 2022. Discourse structuring markers in English: A historical constructionist perspective on pragmatics. Amsterdam: John Benjamins.

74.

Traugott

Elizabeth Closs

Dasher.

Richard B.

2002. Regularity in semantic change. Cambridge: Cambridge University Press.

75.

Traugott

Elizabeth Closs

Trousdale

Graeme

. 2013. Constructionalization and constructional changes. Oxford: Oxford University Press.

76.

Whitt

Richard, J.

2010. Evidentiality and perception verbs in English and German. Bern: Peter Lang.