Daniel Kies Department of English College of DuPage
Modern English Grammar English 2126		Contact Form

Adverbial but

Originally published in The Twentieth LACUS Forum 1993,
Lake Bluff, IL: Jupiter Press, 1994, pp. 315-332.

Current work:   Days remaining this term: Notes: Add Note |  Log in? | Privacy | Change Name & Email Mail this page to a friend

While reading my students' essays, I occasionally discover structures that seem to be "errors" built on analogy, as in sentences like (1a-c).

1. a. But, do parents really have control over their children's television viewing habits?
   b. But, no matter what you say there is bound to be problems with people like that.
   c. But, in reality people rarely say what they feel.

The choice of a sentence-initial conjunction is interesting, given that many students have had years of proscriptive language instruction forbidding that structure. However, the choice of a comma after the conjunction is especially interesting, punctuating the conjunction but as if it were an adverb like however. Is this only a case of confusion — extending the punctuation of conjunctive adverbs to their corresponding coordinators? Or does a structure like that mean something more? I believe it means more.

It seems to me that the coordinator but (with some other conjunctions and conjunctive adverbs) in sentence-initial position is undergoing a semantic reanalysis in contemporary written American English. Imagining a gradient between the word classes COORDINATOR and ADVERB, I hope to show that coordinators and conjunctive adverbs are "moving" along that gradient. Sentence-initial coordinators like but, so, and or increasingly exhibit the distributional, syntactic, and semantic properties of adverbs, while sentence-initial conjunctive adverbs like therefore, however, and plus increasingly exhibit the properties of coordinators and subordinators. Similar gradients exist between SUBORDINATOR and ADVERB as well as between COORDINATOR and SUBORDINATOR.

Before examining this claim in greater detail, however, let's outline the uses of the word but. Modern English exhibits three distinct senses of but.

Prepositional but: but₁

The but used as a preposition is close in semantic force to, and exists in a paradigmatic set with, the preposition except, cf. sentences (2-4) below.

2. Everyone has left but Tom.
3. All but two are missing.
4. There but for the grace of God go I.

Focusing Adverbial but: but₂

Furthermore, there is an adverbial use of but similar in function and meaning to the focusing adverb only, cf. sentences (5-6). Early Modern English also demonstrated a use of but as a relative pronoun that focused emphatically on the information of the following clause, cf. sentences (7-8). This use of but has all but disappeared in contemporary English.

5. A simple sentence contains but one clause.
6. It was but one step on the road to ruin.
7. There is not a university in the area but follows public opinion in fear and trembling.
8. Surely there isn't a mother but faces this problem.

Coordinator but: but₃

Far more frequent, however, is the use of but as a coordinator, cf. sentences (9-11). The usual syntactic hallmarks of the coordinator in English are its restriction to clause-initial position; its ability to fix clauses sequentially; its ability to conjoin constituents of equal rank at different grammatical levels — word, phrase, and clause — as in (9-11); its ability to link subordinate clauses; its ability to link more than two clauses; and, in the case of but, its ability to imply a sense of contrast or negation between the conjoined constituents, as in (9-11) again.

9. She is small but strong.
10. She is painfully strict but extremely fair.
11. She demands a lot, but she gives a lot in return.

The sense of negation is so strong with the coordinator but that it frequently collates with a negative element, as in (12-14).

12. It is rare but not uncommon.
13. The request seems silly but is not unreasonable.
14. He finally left, but not before he spoke his mind.

Most of the literature on but has focused on the theoretical problems presented by the semantic subtleties of this conjunction. Lakoff (1971) outlined the significant syntactic and semantic differences between the 'denial of expectation' and the 'contrast' senses of but. Arguing that context plays a role only in the interpretation of the 'denial of expectation' sense, Lakoff concluded that one cannot maintain a distinction between semantics and pragmatics. In response to Lakoff, Dascal and Katriel (1977) suggested that both interpretations of but must make reference to context, formulating a single rule of interpretation from which the contrast and denial senses form special cases. Wilson (1975) and Blakemore (1989) furthered that discussion by demonstrating that the semantics of but cannot be captured adequately by truth-conditional semantics, opting instead for a different theoretical stance — the principle of relevance.

Indeed, Blakemore (1989) employs the principle of relevance to collapse the denial and contrast senses of but. Both senses, she reasons, are derived from a single implication conveyed by but, the notion of denial: the different senses arise as a consequence of different constraints on the interpretation of that implication. If the constraint is on the relevance of the proposition but introduces, then the result is the 'denial of expectation' interpretation. If the constraint is on the relevance of the conjunction of the two propositions it connects, then the result is the 'contrastive' interpretation.

Coordinators as Adverbials: but₄ and similar structures

It seems to me, however, that the coordinator but₃ no longer appears exclusively in sentence-medial position, where all right-minded prescriptivists have confined it. Indeed, the coordinator but₃ occurs with ever increasing frequency in sentence-initial position, cf. Quirk et al. (1985:1462-63). I suggest though that along with this distributional shift in the coordinator but₃, there are semantic shifts as well. And so but₄ emerges, the adverbial but.

The hallmarks of the adverbial but are syntactic and semantic in addition to the obvious distributional difference. Syntactically, but₄ does allow some minor reordering of clauses (as in 15-16 where columnist Roger Simon is discussing his reactions to a television documentary about drunken driving) and is mobile within the clause for a several English speakers, cf. (17-18). The adverbial but₄ no longer links subordinate clauses because of its distribution in sentence-initial position. Further, but₄ never conjoins constituents below the rank of clause, cf. (19-21). Finally, the adverbial but₄, unlike true coordinators, is not restricted to linking constituents of equal rank, cf. (19-21) again, where the adverbial but₄ demonstrates a link between its clause and the whole of the preceding paragraph.

15. I did not react as Phil Donahue, the host, did when he came on at the end and said: "I was enormously moved by this documentary, as I'm sure you were." ...
         I was plenty moved for the people who were crippled, paralyzed, reduced to vegetables, or killed. But the drunk drivers themselves did not move me. (Roger Simon, "No Compassion for Drunk Drivers")
16. I did not react as Phil Donahue, the host, did when he came on at the end and said: "I was enormously moved by this documentary, as I'm sure you were." ...
         But the drunk drivers themselves did not move me. I was plenty moved for the people who were crippled, paralyzed, reduced to vegetables, or killed.
17. The job's still not done: I'll finish her this avo, but. (informal Australian English, where adverbial but means 'however', cf. Quirk et al. 1985:21)
18. I didn't do it, but. (informal Irish English, where adverbial but means 'all the same', cf. Quirk et al. 1985:644)
19. Sometimes they go to jail and sometimes they lose their licenses and sometimes they lose their jobs, we are told. But, in reality, they rarely do. Most drunk drivers get away with it. (Roger Simon, "No Compassion for Drunk Drivers")
20. College teaches a person many things. Some of the lessons are seemingly unimportant, such as using a verb in a sentence. Others are vital, such as how to tap a keg. But college learning extends beyond the classroom and the bar.... (Pat Healy, "The Poor Slob")
21. Should the content of the research work ... be further specified? No. But there had better be some content, a substance... (faculty memo)

Semantically, the coordinator but₃ strongly implies contrast or negation, whereas the hallmark of the adverbial but₄ is its implied concession. (See Martin (1983) for a systematic description of the concessive sense of several of the coordinators, subordinators, and conjunctive adverbs discussed here, including but.) Through its ability to imply concession, the adverbial but₄ in contemporary American English is moving in the direction of the adverbial but in Australian and Irish English illustrated in (17-18). The easy possibility of paraphrase, substitution, by concessive adverbials like however and all the same or by concessive conjunctions like yet, though, and although lends support to this semantic analysis, cf. (22-26) where I have added the paraphrase for comparison. Compare also (27-29), where the ungrammaticality of paraphrase with coordinator but₃ further suggests that concessive meaning belongs only in the domain of the adverbial but.

22. He was sentenced to two years of unsupervised probation, fined $500 and ordered to produce a documentary on the results of drinking and driving.
    But{However/All the same/Though} having seen his documentary, I get the impression that one of the big results of drinking and driving for Burke was getting exposure on national TV. (Roger Simon, "No Compassion for Drunk Drivers")
23. ..."he didn't mean it; he didn't even remember it happening."
    But{However/All the same} didn't he mean it? Don't all drunk drivers mean it? (Roger Simon, "No Compassion for Drunk Drivers")
24. I admit my reaction to drunken driving is extreme. But{All the same/, although} Burke and I do agree on one thing. (Roger Simon, "No Compassion for Drunk Drivers")
25. I'd like to claim that the paragraph that submits to this kind of structural analysis is thereby a good paragraph and the only good paragraph. But{Yet/All the same/, although} I only claims that the structural relations are real .... (Francis Christensen, "A Generative Rhetoric of the Paragraph")
26. I used to think that we should mold 102 .... But{, though} I'm not so sure anymore. (faculty memo)
27. She is small but{*although/*all the same} strong.
28. She is painfully strict but{?however/*all the same/?though} extremely fair. [The meaning of this example changes from the implied contrast of but to one of implied concession with though or however.]
29. She demands a lot, but{?however} she gives a lot in return. [Again, the meaning changes from implied contrast to implied concession with however here.]

To reinforce the semantic analysis presented here still further, notice that many other sentence-initial conjuncts take on concessive force, as in (30-32).

30. Whatever contributes to the student's developing an independent outlook [should be accomplished in English 102]. Perhaps this cannot be done within an institution. So perhaps the requirement is impossible. (faculty memo)
31. I envision a series of papers written on the way to the research paper — a series of personal essays, anecdotes, events, observations from experience and readings — these papers then can willy nilly be reassembled and rewritten to produce a final major research paper. And that in turn should be subjected to a full global rewriting. (faculty memo)
32. As the world is constituted, the demands of a married state and the care of a posterity require some little regard to what we call circumstances. Yet this provision is greatly increased ... by folly and vanity.... (Henry Fielding, Tom Jones, I, xii)

Furthermore, punctuation, especially the use of a comma after the conjunction, provides the another bit of evidence that these writers are responding to the adverbial, concessive force of the conjunction in informal usage. Compare (1a-c) again and (33-36).

33. And, Stericycle conducts first part of its plant test on mechanical equipment. ("Stericycle Timeline" in The Journal Times)
34. But, one type of student is almost always liked by the other dormmates. (Pat Healy, "The Poor Slob")
35. Yet, to require such content would be difficult in light of our other efforts to incorporate motivation and variety... (faculty memo)
36. We will continue to find reliable students aides to work during the week to make the deliveries. Whereas, every attempt to made [sic] to deliver equipment to your class, we rely on student aides. (faculty memo)

There is one final demonstration of the semantic differences between the conjunct but₃ and the adverbial but₄. Blakemore (1989) and others have noted that the conjunct, expressing a denial of expectation, can be paraphrased by although X, Y if and only if Y expresses a conclusion contradicting the inference based upon X. However, X although Y is not an adequate paraphrase. Using this test, compare the contrast meaning of the conjunct but in (37a-c) to the concessive meaning of the adverbial but in (38a-c).

37. a. She is rich but honest.
    b. = Although she is rich, she is honest.
    c. ≠ She is rich, although she is honest.
38. a. I don't know what the content should be. But there had better be some content.
    b. = Although I don't know what the content should be, there had better be some content.
    AND
    c. = I don't know what the content should be, although there had better be some content.

The acceptability of both (38b and c) demonstrate that the implicational relations that hold true between propositions linked by the coordinator but do not apply to the use of but in (38). This seems to suggest that the but in (38) functions less as a coordinator and more as an sentence-initial adverbial, conveying a sense of concession.

I realize at this point that one might object to the data presented so far: it is too biased, too filled with the informal language of journalism and faculty memoranda. I use those sources for two reasons: first, contemporary journalism is filled with examples of sentence-initial but and so it became the path of least resistance. Secondly, there seems to be some statistical discrepancy between the formal prose of faculty members and their informal prose.

So that I might counter those objections more fully, note the following tables. Table 1 presents the frequency of the word but in a corpus of English 101 essays, prepared and revised by eighty-seven College of DuPage students, with a median age of nineteen, fifty percent female, fifty percent male.

                                TABLE 1:

               Frequency of but in Composition 101 essays

   Position                      Occurrences   Percentages

   sentence-initial but:               68      17.0% of all buts
   other clause-initial but:          192      48.0% of all buts
   total tokens of but:               398       0.3% of total words

   total words in Composition 101 corpus: 114,000

Table 2 presents the frequency of the word but in a corpus of scholarly essays, a conference paper written by a College of DuPage faculty member (Frances Fitch, "Bloom and a Vichican Practice of Advertising," a paper presented at the International Joyce/Vico Conference, Venice, Italy, June 1985), two articles that appeared recently in PMLA (Ellen McCracken, "Metaplagiarism and the Critic's Role as Detective," PMLA 106 (1991) 1071-1082 and Marie Borroff, "Sound Symbolism as Drama in the Poetry of Robert Frost," PMLA 107 (1992): 131-144), and one article of mine (Daniel Kies, "Fourteen Types of Passivity: The Suppression of Agency in Nineteen Eighty-Four," The Revised Orwell. Ed. Jonathan Rose, East Lansing, MI: Michigan State University Press, 1992, 47-60).

                                TABLE 2:

                  Frequency of but in scholarly essays

   Position                    Occurrences       Percentages

   sentence-initial but:             10           7.8% of all buts
   other clause-initial but:         58          45.0% of all buts
   total tokens of but:             128           0.2% of total words

   total words in scholarly corpus: 56,000

Table 3 illustrates the use of the word but in the Brown University corpus of written American English (Francis and Kucera 1982). The Brown corpus of more than a million words selected examples of prose from a variety of genres and registers. Notice that in terms of overall frequency of use, the 101 student and the scholar are not significantly different from what might be considered a norm for written American English; indeed, it seems that 'academic' writers of all kinds are a bit more conservative in their frequency of usage.

                                TABLE 3:

            Frequency of but in the Brown University corpus
   


	total tokens of but: 4226     0.4% of the total words



	total words in Brown corpus 1,013,466

Table 4 presents some astonishing results. Journalists seem to prefer the sentence-initial but. The pieces of newspaper journalism in the corpus were chosen randomly from a mixture of eight front page stories, six editorials, and four columns collected between November 10 and December 13, 1990. The newsmagazine pieces were four randomly chosen Time and Newsweek cover stories during the same period.

                                TABLE 4:

                 Frequency of but in journalism samples

   Position                    Occurrences       Percentages

   sentence-initial but:             48          61.5% of all buts
   other clause-initial but:         24          31.0% of all buts
   total tokens of but:              78           0.4% of total words

   total words in scholarly corpus: 17,600

The statistics in Table 4 would seem to suggest journalism as a source of the composition students' relatively more frequent use of the adverbial but. The students may be analogously generalizing the use of but, despite the practice and the prescriptivism of their teachers. But that explanation, as simple and intuitively right as it seems, can not be the whole story. First, the demographics of newspaper and magazine readerships weigh against that explanation: news readers are largely an older group. Secondly, the statistics in Table 2 do not represent college instructors' practice quite accurately.

So consider Table 5, where the statistics represent the (informal) prose of several faculty members (mostly English department members) at the College of DuPage.

                                TABLE 5:

                    Frequency of but in faculty memos

   Position                    Occurrences       Percentages

   sentence-initial but:              8          14.0% of all buts
   other clause-initial but:         30          52.0% of all buts
   total tokens of but:              58           0.4% of total words

   total words in scholarly corpus: 16,150

The numbers in Table 5 are very close to the corresponding numbers of 101 students in Table 1. This result suggests that faculty members exhibit a classic case of divided usage: formally they write conservatively, prescriptively, but informally they accept — and use — sentence-initial conjunctions.

To explore the range of this divided usage further, I prepared a fifty page booklet as part of an acceptability study. Each page of the booklet contained a short discourse (two to four sentences) that illustrated the use of coordinators, subordinators, and conjunctive adverbs in sentence-initial, -medial, and -final position. The respondents were students and faculty at the College of DuPage, and they were fully aware of the intent of the survey. Each page asked the respondents to evaluate the short discourse twice, answering 'yes' or 'no' to the same two questions each time: "I have seen this use of __________" and "I have used __________ this way myself." The blanks were filled with the appropriate conjunction or adverb illustrated on the page.

Tables 6 and 7 present some of the result of the survey. Note in Table 6 that faculty both recognize and use the sentence-initial but frequently. This result, I suggest, more closely conforms to the practice of faculty revealed in Table 5, and accounts for the similar numbers in the students' prose (Table 1). Interestingly, Table 7 indicates that students claim to recognize and to use comma punctuation after sentence-initial buts more frequently than faculty. Those numbers are not borne out by the data in the corpus study of 101 prose. Of the sixty-eight sentence-initial buts in the 101 corpus, only two co-occurred with following comma punctuation. However, the students' claim itself, I suggest, argues in favor of their perception of sentence-initial buts as adverbials and argues in favor of a shift of but₄ along a gradient away from CONJUNCTION and toward ADVERB.

                                TABLE 6:

             Acceptability judgments of sentence-initial but
                        without comma punctuation
                 (e.g. But I am not so sure anymore.)


       DuPage faculty:               I have seen:     27     90%
                                     I have used:     18     60%

       Total faculty in survey: 30



       Composition 101 students:     I have seen:     59     56%
                                     I have used:     17     16%

       Total students in survey: 105

                                TABLE 7:

             Acceptability judgments of sentence-initial but
                           with comma punctuation
                  (e.g. But, one type of student is almost
                    always liked by the other dormmates.)


       DuPage faculty:               I have seen:     11     37%
                                     I have used:      1      3%

       Total faculty in survey: 30



       Composition 101 students:     I have seen:     28     27%
                                     I have used:      9      9%

       Total students in survey: 105

The data in Tables 8 and 9 represent another shift in contemporary written American English: certain subordinators (such as when, which, although, though) and conjunctive adverbs (such as however, therefore, plus, then) are also on the move along gradients.

                                TABLE 8:

             Acceptability judgments of sentence-final although
                         without comma punctuation
                    (e.g. I don't like pizza although.)


       DuPage faculty:               I have seen:      9     30%
                                     I have used:      0      0%

       Total faculty in survey: 30



       Composition 101 students:     I have seen:     23     22%
                                     I have used:      5      4%

       Total students in survey: 105

                                TABLE 9:

             Acceptability judgments of sentence-final although
                            with comma punctuation
               (e.g. I felt he was being honest, although.)


       DuPage faculty:               I have seen:      9     30%
                                     I have used:      0      0%

       Total faculty in survey: 30



       Composition 101 students:     I have seen:     38     36%
                                     I have used:     12     11%

       Total students in survey: 105

Although several faculty and students alike recognize 'sentence fragment' errors like those of Tables 8 and 9, no faculty member reports using such a structure. However, the students report using the structure, presumably finding the structure acceptable. Corpus studies do not support the students' usage claims again; only four sentence-final althoughs exist in the corpus, three preceded by comma punctuation. Nevertheless, their report of using it is suggestive about their attitudes towards its adverbial function and its acceptability in that function.

Subordinators as coordinators

Subordinate clauses generally present backgrounded (i.e., presupposed, given, or old) information, cf., Quirk, Greenbaum, Leech, and Svartvik (1985); Dillon (1981); and van Dijk (1977). Participial and verbless adverbial clauses, for example, possess no explicit marker of subordination, thereby creating semantic indeterminacy, which is usually resolved through context; e.g., the participial clause in (39a) is semantically equivalent to either (39b) with a relative clause or (39c) with coordination.

39. a. Jason, told of his son's accident, immediately phoned the hospital.
    b. Jason, who was told of his son's accident, immediately phoned the hospital.
    c. Jason was told of his son's accident and immediately phoned the hospital.

Such indeterminacy is not uncommon in subordination: "In their indeterminacy, adverbial participle and verbless clauses resemble the versatile relationships expressed by nonrestrictive relative clauses and the connective function of the coordinator and" (Quirk et al. 1985:1123). The indeterminacy of several conjunctions between coordinators and subordinators explains, I think, the increasing use of restrictive relative clauses punctuated as full sentences, cf. (40-41).

40. There was no marker to their memory, no record of their graves' precise location.
    Which is why Hermes thought at first that the tangle of bones sticking out from the recently graded hillside were actually tree roots. (David Silverman, "Man's Walk with dog may be stroll into history")
41. The composer who from 1969 to 1976 declared a worldwide moratorium on performances of his music to protest what he called "all the rottenness in the world" has actually become sort of benign.
    Which is not to say he has greatly curbed his tongue or his penchant for recycling his best anecdotes and one-liners. (John von Rhein, "Ralph Shapey turns mellow – even toward the CSO")

Consider also these examples in (42-47) below as further evidence to support the semantic indeterminacy of the apparently subordinate clauses in (40-41). Some adverbial clauses exhibit a strong temporal restriction, similar to the temporal restriction on the coordinator and, suggesting a gradient of taxis between coordination and subordination.

42. He was lecturing to his class when suddenly a door flew open.
43. *When suddenly a door flew open, he was lecturing to his class.
44. He was lecturing to his class and suddenly a door flew open.
45. Suddenly a door flew open and he was lecturing to his class. [not the same meaning as (42)]
46. The car stopped when it hit the pole.
47. When it hit the pole, the car stopped.

As (42) and (43) demonstrate, some clauses with when are restricted to natural time order just as the clauses (44) and (45) with and are temporally restricted. But only the time sense of when in (42) and (43) is restricted like and. (Let when₁ represent the time sense of the word.) Consider the cause sense of when in (46) and (47). (Let when₂ represent the cause sense of the word.)

The examples in (40-47) suggest a gradient of clause connectors between coordination and subordination: and appears to be the best coordinator; relative conjunctions like which seem ambiguous between subordination and coordination (medial linkers); and the two when conjunctions are closer to subordination, although when₁ is closer to coordination than when₂ is, cf. Quirk et al. (1985:1258-59) and Kies (1990:231-58).

             and                 which     when1    when2
              |                    |         |        |
COORDINATION|-X--------------------X---------X--------X-|SUBORDINATION

Adverbials as conjunctions

Finally, it appears that conjunctive adverbs are on the move as well, along gradients between ADVERB and SUBORDINATOR and between ADVERB and COORDINATOR. Consider the examples in (48-56), which are given as informal, but acceptable, examples of conjunction in Curme's grammar of English.

48. There is not only concision in these lines, there is also elegance. (Curme 1931,2:164)
49. I do not suggest that he is negligent, still less (or much less) that he is dishonest. (OED)
50. You never fought with any, lesse slew any. (Ben Jonson, Magnetic Lady, III, iii)
51. First think, then act. (Curme 2:165)
52. My interests are two fold: on the one hand my flowers claim me in the morning, on the other (hand) I am absorbed in language studies the rest of the day. (Curme 2:165)
53. He is very poor, at least he has not the wherewithal to by proper clothes for his wife and family. (Curme 2:166)
54. Seize the chance, else you will regret it. (Curme 2:166)
55. He wanted to take precedence of all the Lowland gentlemen then present, only my father would not suffer it. (Scott, Waverly, Ch. XV)
56. He makes good resolutions, only he never keeps them. (Curme 2:167)
57. I saw the smoke behind them, then I realized it real fast.... (101 essay)
58. You don't really have enough money to stay in a hotel, plus the car rental is draining your wallet. (101 essay)

The examples of conjunctive adverbs as conjuncts testify further that word class categories like SUBORDINATOR, COORDINATOR, and CONJUNCTIVE ADVERB do not have rigid boundaries. Needless to say, one need not read many college compositions before discovering that those categories do not have rigid boundaries in the language of 101 students either, as the many 'comma splice' faults readily testify, cf. (57-58).

Explanations and conclusions

I realize that the claims made here are large in scope while supported by only a small data set. Nonetheless, I believe that the data presented above support the conclusion that we have a classic example of divided usage in the case of COORDINATORS, SUBORDINATORS, and ADVERBS in sentence-initial position. Moreover, I see enough consistency in the data to make the following hypotheses, allowing us to test these claims about the existence of the adverbial but.

HYPOTHESIS 1: Thematic Distribution of Adverbial But

But₄ occurs as a strongly thematic element in complex themes.

Based on the work of Halliday (1967 and 1985) and Fries (1983), the first hypothesis stipulates that the adverbial but occurs in texts as a strongly thematic element in complex themes. Halliday and Fries both define a thematic cline between clause-initial elements that are weakly thematic and those that are strongly thematic. As Fries (1983) and Eiler (1986) point out, for any element to be strongly thematic, the speaker or writer must have a choice, and the data I have seem to allow adverbial but only as one choice of several. Further, Halliday and Fries both characterize theme in English so as to allow for both simple and complex themes, and I suspect that the adverbial but occurs in theme complexes, as is often the case with other sentence- and clause-initial adverbials.

HYPOTHESIS 2: Clause Relation Type of Adverbial But

But₄ occurs only in clause relations of the PROBLEM-SOLUTION type.

The research of Winter (1982), Jordan (1984), Hoey (1986), and Hoey and Winter (1986) into clause relations, the way in which clauses and sentences cohere within the larger text, suggests that well-formed texts maintain a metastructure, often consisting of four elements, Situation-Problem-Solution-Evaluation. The second hypothesis stipulates that adverbial but, with its ability to signal concession, is an overt marker of the transition from Problem to Solution. On the other hand, the conjunct, contrary to expectation but, signals the transition from Situation to Problem.

HYPOTHESIS 3: Reanalysis of Conjuncts and Adverbs

But is not alone in undergoing semantic reanalysis.

The few examples presented here seem to suggest a large scale reanalysis of several groups of word classes in modern English. The motivation for the usage problem seems to be the gradient nature of the items in those word classes. The gradients based on form classes might look something like (59) below.

Yet the gradients in (59) conceal a problem, indicated by the asterisks after the words however and therefore. It is not certain that however is moving toward the COORDINATOR pole or that therefore is moving toward the SUBORDINATOR pole. I place them there only because of their semantic affinities to coordinators and subordinators respectively and because of a few syntactic similarities. Further research is needed to explicate fully the dimensions of those gradients.

One way to overcome the shortcoming of the gradient based on form classes is to look at the same phenomena from a functional point of view, as in (60). If we define PARATAXIS as the juxtaposition of clauses of equal information value and HYPOTAXIS as the juxtaposition of clauses where one clause is foregrounded informationally while the other is informationally backgrounded, then we can present the same data on a unified, function-oriented gradient.

59. Gradients based on form classes:

                      ADVERBS
             (first, tomorrow, quickly)
                       /    \
                      /      \
     then, however*, plus     therefore*
                    /          \
                   /            \
               but4              although, though
                 /                \
   COORDINATORS ------|---|---|-----SUBORDINATORS
(and, or, but3)    which  |   |      (that, as, who)
                      when1   |
                          when2

60. A gradient based on function:

          then,                 that                    as,
          plus,             (rel pro),                  who,
          and   but3           which    when1   when2   that
           |    |                  |        |       |   (conj)
           |    |                  |        |       |    |
PARATAXIS|-X----X------X---X---X---X--------X-------X----X-|HYPOTAXIS
                       |   |   |
                    but4   |   |
                     although, |
                       though  therefore,
                               thus,
                               so,
                               however

This is an on-going change in the history of English. As Curme (1931,1:94) points out "Nothing in English grammar has changed so much within the Modern English period as our conjunctions." And historical hypotheses (like all hypotheses) need careful corpus research to justify any postulated explanation. I admit that much of what has been said here is suggestive and programmatic; however, earlier studies have neither adequately described nor explained these data. Blakemore, for example, completely misanalyzes cases of adverbial but. In an effort to bolster her hypothesis, Blakemore is forced to explain away counterexamples like A: My parents vote Labour B: But my parents vote Tory by ignoring them simply because they are "odd" (1989:31). To his credit, however, Martin (1983) does recognize the concessive interpretation conveyed by some sentence-initial buts. Yet, despite the scope and the very thoughtful analysis of his study, Martin fails to recognize the gradient nature of these items. Gradience is the central concept, I believe, for resolving two problems Martin (1983:54-60) acknowledges in his study: explaining the process of grammaticalization assumed to be at work in English conjunction system and explaining the process of conjunction in English in addition to describing the system of English conjunctions.

Finally, I have tired to present the case for adverbial but in such a manner that the case is testable through the collection of a larger corpus. After all, as Yngve (1986) has pointed out, if linguistics is to be a science, that is the least I could do.

References

Blakemore, Diane. 1989. Denial and contrast: A relevance theoretic analysis of but. Linguistics and Philosophy 12.15-37.

Curme, George. 1931. A grammar of the English language. 2 volumes. New York: D. C. Heath and Company.

Dascal, M. and T. Katriel. 1977. Between semantics and pragmatics: The two types of but – Hebrew aval and ela. Theoretical Linguistics 4.143-72.

Dillon, George. 1981. Constructing texts. Bloomington, IN: Indiana University Press.

Eiler, Mary Ann. 1986. Thematic distribution as a heuristic for written discourse function. Functional approaches to writing: Research perspectives, ed. by Barbara Couture, 49-68. Norwood, N.J.: Ablex.

Francis, W. Nelson, and Henry Kucera. 1982. Frequency analysis of English usage: Lexicon and grammar. Boston: Houghton Mifflin Company.

Fries, Peter H. 1983. On the status of theme in English: Arguments from discourse. Micro and macro connexity of texts, ed. by J. S. Petofi & E. Sozer, 116-52. Hamburg: Helmut Buske.

Halliday, Michael A. K. 1967. Notes on transitivity and theme in English, part 2. Journal of Linguistics 3.199-244.

___________________. 1985. Introduction to functional grammar. London: Edward Arnold.

Hoey, Michael. 1986. Overlapping patterns of discourse organization and their implications for clause relational analysis of problem-solution texts. Studying writing: Linguistic approaches, ed. by Charles Cooper & Sidney Greenbaum, 187-214. Beverly Hills, CA: Sage.

____________ and Eugene O. Winter. 1986. Clause relations and the writer's communicative task. Functional approaches to writing: Research perspectives, ed. by Barbara Couture, 120-41. Norwood, N.J.: Ablex.

Jordan, Michael. 1984. Rhetoric of everyday English texts. London: George Allen & Unwin.

Kies, Daniel. 1990. Indeterminacy in sentence structure. Linguistics and Education 2.321-58.

Lakoff, Robin. 1971. If's, and's, and but's about conjunctions. Studies in linguistic semantics, ed. by Charles J. Fillmore & D. Terence Langendoen, 114-49. New York: Holt, Rinehart, Winston.

Martin, Jim R. 1983. Conjunction: The logic of English text. Micro and macro connexity of texts, ed. by J. S. Petofi & E. Sozer, 1-72. Hamburg: Helmut Buske.

Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1985. A comprehensive grammar of the English language. London: Longman.

van Dijk, Teun. 1977. Text and context. London: Longman.

Wilson, Deirdre. 1975. Presuppositions and non-truth conditional semantics. New York: Academic Press.

Winter, Eugene O. 1982. Towards a contextual grammar of English. London: George Allen & Unwin.

Yngve, Victor. 1987. To be a scientist. The Thirteenth LACUS Forum 1986, ed. by Ilah Fleming, 5-25. Lake Bluff, IL: LACUS.

^Take Note! | Table of Contents | Syllabus | eForum | Search

The HyperTextBooks | The English Main Page | The Composition Links