ROBERT DE BEAUGRANDE——————————————————————
Sentence first, verdict afterwards:
On the remarkable career of the “sentence”
“Sentence first,
verdict afterwards”. “Stuff and nonsense!” said Alice loudly. “The idea of
having the sentence first!”
— Alice in Wonderland
1. An apparent
paradox
1.1. My title and its indebtedness to the fantasy-world of Lewis Carroll may
invoke, in a light-hearted way, the remarkable eagerness to put the “sentence
first” and to postpone until long “afterwards” the task of passing a sound
“verdict” upon its actual nature. For among all the items or units in
discussions and studies of language, none has occupied a status quite so peculiar
as the sentence. Representatives of the most diverse approaches or theoretical
schools, who may disagree sharply on numerous other points, have agreed almost
unanimously that the sentence merits first place. Yet closer examination of how
the term “sentence” has figured in such discussions and studies will detect a
significant diversity of senses and meanings, whereupon the semblance of
unanimity evaporates.
1.2. Perhaps this paradox has arisen from highly strategic evasions whereby
diverse approaches have obtained a basic organisational unit to provide a
reassuring framework, and have maintained a reassuring impression that those
approaches ultimately constitute a consolidated and cumulative enterprise. In
sections 2-4 I shall briefly illustrate some ways the sentence has dominated
much discourse about language and some consequences that have ensued.
2. The “sentence” in discourses of
linguistics and philosophy
2.1. In the discourses of influential modern linguists, the resolve to put
the sentence first has bordered upon a reverent article of faith. For Saussure
(1966 [orig. 1916]:124), “the sentence is the ideal type of syntagm”. For Sapir
(1921:35), “the sentence” “is the major functional unit of speech”. For
Bloomfield (1933:276, 170), “the power or wealth of a language” lies in its
“sentence-types, constructions, and substitutions”; and the “sentence” is the
site where “linguistic forms” are “united” by “meaningful grammatical
arrangement”. For Firth (1957 [orig. 1951]:170; 1968 [orig. 1956]:148), “most
linguists would agree that the study of the sentence as our primary datum is
the order of the day”; and “the sentence and syntactic analysis naturally find
a central place” in a “general theory”. For Chomsky, (1957:13), a “language” is
“a set (finite or infinite) of sentences, each finite in length and constructed
out of a finite set of elements”.
2.2. If these pronouncements sound portentous, they were outdone by some we
encounter among philosophers of language. For Wittgenstein (1963 [orig.
1922]:32), “the thought is the meaningful sentence; the totality of sentences
is the language” (“der Gedanke ist der sinnvolle Satz; der Gesamtheit der Sätze
ist die Sprache”). For Wegener (1885:181), “all elements of language are
originally sentences” (“alle Sprachelemente sind ursprünglich Sätze”); and
Firth (1968 [orig. 1959]:148) expanded this latter notion into the speculation
that “even the origins of all speech considered biographically in the nurture
of the young and in the history of the race are to be found in sentences”.
2.3. This small gallery of quotations among the host of similar
pronouncements should suffice to indicate the fervent reverence wherewith the
“sentence” has been awarded a centrality and a scope subsuming the entirety of
“language”, “speech”, and even “thought”. But if we examine those same
discourses for precise elaborations or definitions of the “sentence”, we are
likely to find evasions or inconsistencies. Right after pronouncing the
“sentence” “the ideal type of syntagm”, Saussure consigned it “to speaking
[parole], not to language [langue]” (1966:124). His reasoning for this
reservation is worth recalling: “in the syntagm there is no clear-cut boundary between the language fact, which is a sign of
collective usage, and the fact that
belongs to speaking and depends on individual
freedom; in a great number of instances it is hard to classify a combination of units because both forces have combined in producing
it, and they have combined in indeterminate
proportions” (1966:125, my italics).
Saussure already conjectured that the “sentence” is an “indeterminate” unit
positioned along multiple borders between system and instance, collective and
individual, regularity and innovation, and so forth.
2.4. Sapir in his turn conjectured that “underlying the finished sentence is
a living sentence type, of fixed formal characteristics”; we can “substitute”
“words” and yet “feel instinctively, without the slightest attempt at conscious
analysis, that the two sentences fit the same pattern, that they are really the
same fundamental sentence”; and “new sentences are being constantly created”,
whereby “these fixed types or actual sentence-groundworks may be freely
overlaid by such additional matter as the speaker or writer cares to put on”
(1921:37, 85). His terminology sounds more psychological and philosophical than
Saussure’s, but the borderline status of the “sentence” is similar: “fixed
type” and “same pattern” versus “constant creations” and “free overlayings”.
2.5. In a pointed repudiation of such discourse, Firth (1968 [orig.
1956]:102) warned: “I am not going to define the sentence; I will, however,
say, that by sentence I do not intend a judgement expressed in words or a
logical proposition consisting of a predicate, that which is not the predicate
being the subject; all this logical or psychological analysis we have
abandoned”. Yet we saw Firth declaring that “the study of the sentence is the
order of the day”; and he elsewhere commented that “when we speak, we usually
‘spill a mouthful’ or use a whole sentence — that is to say, the unit of actual
speech is the whole of a phrase”. Ironically, spoken language is precisely the
domain where the status of the “sentence” seems most insecure (4.1; 5.15, 5.;
5.15, 5.18, 5.20).
2.6. Firth’s outlook was typical of linguists who mistrusted psychological
and philosophical adumbrations and sought to define the “sentence” by
realistic, observable criteria. Bloomfield proposed to define the “sentence” as
“an independent form not included in any larger” form” and thus able to be “spoken alone” (1933:170,
179). But this definition is not too compatible with his own claim, cited
above, that the “sentence” is the site where “linguistic forms” are “united” by
“meaningful grammatical arrangement”. He now had to allow for a single word
(like the exclamation “John!”) being a sentence (1933:170), just as Firth (1968
[1956]: 102) would later assert that “there may be many one-word sentences”.
How can a single word constitute a “grammatical arrangement”? Bloomfield
(1933:170) hedged by allowing that “a form which in one utterance figures as a
sentence, may in another utterance appear in included position”; and that “an
utterance may consist of more than one sentence”. As a practical problem, he
conceded that “the linguist cannot wait indefinitely for the chance of hearing
a given form used as a sentence”. So we find the “sentence” being defined by
apparently realistic observable criteria, only to be told that we may observe
something different or may not get a chance to observe anything at all.
2.7. Longacre’s (1958) more
inventive approach to the “sentence” grew expressly out of the need for
observation to follow “grammar discovery procedures” whilst doing fieldwork in
a discipline that would only later be officially recognised as “discourse
analysis” (cf. 2.11). The fieldwork data he studied were
taken from spoken language, and yet he (as the first, I believe) saw, in their
formal organisation, units that corresponded to what might be a sentence as a
segment inside a “spoken paragraph”
(e.g. Longacre 1970). He centred his analysis of linguistic forms on the levels
of morpheme, word, and clause. So he was left free to treat the sentence mainly
in terms of what speakers might be doing
with it in the paragraph, in analogy to what the paragraph might be doing
within the discourse. He thus conceived of sentences being used to express such
much-discussed logical relations as “reason”, “succession”, and
“conditionality” but also such rarely-discussed ones as “frustration” (see now
Longacre 1996). His analysis was hugely refined and enriched by his focus upon
spoken narratives (or “storylines”) in traditional cultures, where the
speaker’s intentions to tell a good story and keep the audience engaged would
provide revealing clues about what the “sentences” were and what they were
being used for. Those languages had been spared the self-conscious apparatus
for managing “sentences” that has been inflicted upon English; and I shall
suggest later that if we could work backward from large sets of authentic
English data, we might come upon some corresponding discourse-based
reconception of the “sentence”.
2.8. Longacre’s work was
partially in tune with Pike’s so-called “unified theory,” his too driven by the
exigencies of fieldwork observation but focusing on “human behavior” as a more
general category and phenomenon than language. Pike (1967: 1945-64]: 409, 134,
571) expressly “rejected” the immediate constitutent scheme (so esteemed by
Bloomfield) of “combining morphemes to make words, words to make sentences”;
and he cautioned against “the analysis of words or sentences outside of normal
behavioural contexts” and against “working with ‘cleaned-up text’” and
“sentences” in “practical fieldwork”. For dealing with “the total language
event in a total cultural setting”, “the sentence is a totally inadequate
starting” or “ending point” (1967: 484, 147). In full accord with Longacre, he
realised (though his monumental 1967 “unified theory” volume failed to show it
for any non-English language) that “many important characteristics of sentence
structure can be adequately handled” “only in reference to discourse structure”
(1967: 485f). He clearly (but vainly) warned
of the perils of the then emerging Chomskyan paradigm: in “its broader
setting”, the “sentence” entails “the deep problem of identity of unit against
ground” and “remains immune from attack” only if “it is taken, in a regularized
form, as an axiomatic starting point” (1967: 8n).
2.9. And such was just the starting point taken by Chomsky, who urgently
needed to camouflage the observational vacuity of his own academic project by
summarily dismissing fieldwork linguistics: the “observed use of language”
“surely cannot constitute the subject-matter of linguistics, if this is to be a
serious discipline” (1965: 4). This view dovetailed with defining a “language”
to be an “infinite set of sentences”, since no infinite set could conceivably
be observed. So he evaded all the problems of observing data by “assuming that
the set of sentences is somehow given in advance” (1957:13, 18, 54, 85, 103).
If Bloomfield’s “sentences” had been defined in terms of what we can (but might
not) observe in speech, Chomsky’s “sentences” were defined so as to render
observation irrelevant; they can freely invented by linguists to suit the
occasion, such as 24 artificial sentences which were analysed in his Aspects. Recalling Sapir’s terms, we
could say that Chomsky’s conceptions of the “sentence” emphasised the “fixed
types” and “same patterns”, whilst limiting the potential for “constant
creation” and “free overlaying” to mechanical “transformations” of sentence
structure.
2.10. Chomsky’s evasions regarding “sentences” were
to prove enormously influential for other linguists, whether or not they adhered
to his school. Lyons (1977:31, my italics) proposed to called them
“system-sentences” and frankly asserted that they “never occur as the products of ordinary language behaviour”, yet
their “representations” “are customarily cited in grammatical descriptions of
particular languages”. These “system-sentences” would not even be of the type
“the man hit the ball” or “the cat sat on the mat”; though mindlessly banal,
they could not be said to never
occur.
2.11 The influence even extended to Brown and
Yule’s officially designated “discourse analysis”: they programmatically
opposed Chomskyan linguistics, yet stipulated
that the “grammarian’s ‘data’ is [sic] invariably the single sentence, or a set of single sentences
illustrating a particular feature of the language being studied”; and that “typically” “the grammarian will have constructed the sentence” (1984:20, my
italics). They vowed that “denying the admissibility of a constructed sentence”
would be “a dangerously extreme view”; and they tried to salvage the issue by
arguing that “the sentence must, in some sense, derive from the ‘ordinary
language’ of his [the grammarian’s] daily life”. But how can this “deriving”
operate for sentences that, in Lyons” view, “never occur” in ordinary life?
2.12 The real crux of the problem might be detected
beneath Sampson’s (1980:153, my italics) reservation that the use of
“constructed sentences and personal introspection” “leads to non-testability in principle of any
claims made”. My own detailed study of Chomsky’s reasoning and arguments
(Beaugrande 1998a) has convinced me that this apparent drawback was deliberate.
The tactics were imposed precisely in
order to protect the claims from being tested, just like claims about the
“ideal speaker” who Chomsky (1977:192) has cheerily granted ”does not exist in
the real world”.
2.13 Again, my gallery of quotations about what or
where the “sentences” actually are, or where we can find them, has been small
in proportion to the many similar pronouncements. Yet we may already detect an interesting
discrepancy between the reverent assertions about the supreme importance of the
“sentence” and the curiously evasive manoeuvres which prevented the “sentence”
from being pinned down in sufficiently operational terms to determine whether a
given sample or set of language data might indeed consist of sentences, how we
might determine their internal constituents or their external boundaries, and
so on. Instead, influential linguists have shared a symptomatic eagerness to
put the “sentence first” by assuming that sentences are and must be there,
“given in advance” or “instinctively” prior to any “conscious analysis”,
whether or not we can reliably define or observe them.
2.14
The chief practical consequences of these theoretical
evasions have been that a majority of linguists have tended either to address
only language data that are already formatted as sentences, or else to alter
the language data as a step in the description and analysis; and that a
“grammar” has tended to be considered a device for relating one sentence to
other sentences and not for relating a language to a theory of human
interaction as Pike and Longacre have advocated. Due these tendencies, much
real data from everyday discourse have been ignored; and problems situated
beyond the boundaries of the sentence have been regarded as non-issues. In the
best-known grammars of this type, the “transformational” or “generative”
grammars, altering the data has indeed been the main concern, e.g., by converting a sample sentence into “kernel sentences”
with patterns rather like “the man hit the ball”, or by relating it to “deep
structures” which are purely hypothetical. These tactics purport to explain the original data, whereas in
reality they just get rid of the data and
replace it with idealised data whose tidy “syntactic structures” present no
further difficulties.
2.15 Such tactics have implicitly undercut the most
stoutly proclaimed motivation for the “science of modern linguistics”: to
describe a language the way it really is
and not the way some self-confident authorities or language guardians believe
it ideally ought to be. Choosing the
“sentence” to be the ultimate framework of order and structure in language was
a key step along the path to a linguistics concentrating on data that “never
occur as the products of ordinary language behaviour” (Lyons).
2.16
This same path has led to a pervasive disdain for the
data that do occur. Saussure’s (1966
[orig. 1916]:14, 9, 11) declaration that “language is a well-defined object in
the heterogeneous mass of speech facts” was linked to his grim warning that
“speech cannot be studied”, nor indeed can it be “put in any category of human
facts, for we cannot discover its unity”. The same disdain pervaded Chomsky’s
(1965:201) assertion that “from the point or view of the theory”, “much of the
actual speech observed consists of fragments and deviant expressions of a
variety of sorts”.
2.17
This disdain has entrained linguistics in an
extraordinary dualism that seems to have escaped general notice. In their zeal
to affirm the ideal order of
“language”, some influential linguists have evidently viewed actually occurring
data as a mass of disorder. The
implication would be that “ordinary language behaviour” continually triggers a
bizarre “catastrophe” similar to the technical sense of “catastrophe theory”
(cf. Thom 1989), namely an abrupt transition from stable and integrative order to unstable
and disintegrative disorder, perhaps like a demonic reversal of what
semiotics and aesthetics (e.g. Vukovich 1970) has called the “Birkhoff
transition” from sign-plenitude to sign-order (cf. Birkhoff 1950). Along the
way, the “well-formed system sentences” de-generate into “fragments and deviant
expressions”.
2.18
Made explicit, the dualism is flagrantly absurd.
Evidently, the theoretical linguistics of Saussure or Chomsky has routinely
attributed to “language” an idealised
mode of order which is fully
determined within the abstract system, e.g. by some “collective usage” or
“langue” or “competence” quite independent of the “individual freedom” in
“parole” or “performance”. Saussure situated the sentence in between, whereas
Chomsky situated it at centre of the system. But surely the vastly more
plausible principle would be that the real
order of language elaborately
supports the order of discourse
without fully determining it. This principle, which has been most
thoroughly elaborated in systemic functional linguistics, can be exploited only
if we examine how “langue” or “competence” gets actualised in “parole” or
“performance” instead of “assuming that the set of sentences is given in
advance” or just “constructing” our own sentences as we go along. As Pike and
Longacre foresaw, We will undoubtedly discover many constraints which are only determined upon the plane of the
actual discourse; and a system or “grammar” that abstracts away from these
to attain ideal order is in fact promoting disorder, which prompts the feverish
construction of vast apparatuses of “rules” and “features” (5.25).
3.
The “sentence”
in discourses of pedagogy
3.1. Ironically, the same pattern of reverence plus evasion can be detected
in language pedagogy, even where there has presumably been no direct influence
from linguistics. A typical “grammar book” or “composition” textbook offers
such grand pronouncements as “a sentence is a basic unit of language, a
communication in words” (Willson 1980:29); and “we naturally speak in
sentences” (Glazier 1981:2). If these claims were valid, pedagogy would be spared
its effusive efforts to teach the
sentence.
3.2. The learners who use such textbooks resemble a linguist in confronting
a sequence of words they have themselves constructed and considering if it
should count as a sentence. Yet they differ starkly in having little or no
conceptions or training in the description of linguistic structure. Are the
criteria offered by the usual textbooks operational enough to be applied by
ordinary language learners, who do not “naturally
speak in sentences”?
3.3. Patently inadequate are the traditional criteria that “a sentence is a
combination of words expressing a complete thought and making complete sense”
(Gartside 1981:239). Even for philosophers, the “completeness” of “thought” or
“sense” is an undecidable issue. And if a single sentence really could
“complete your thought”, you could stop and write no more.
3.4. Not much better are the criteria that “the sentence” “nearly always
contains two pieces of information the listener is conditioned to expect from
it: who or what is involved, and what does he, she, or it do or feel” (Willson
1980:29). Here, the potential confusion lies in trying to tell what might or
might not constitute a “piece of information”, and, once more, in implying the
sentence to be a spoken unit, one aimed at an “expectant listener”. Yet a
learner might more reasonably assume that ‘the man hit the ball”, and not just
“the man”, is just one “piece of information”.
3.5. We might feel reminded of Bloomfield’s notion (cited above), though the
resemblance is probably coincidental, by the criteria that “sentences” are
“complete units, capable of standing alone without the support of supplementary
comment” (Willson 1980:29). But learners may still get confused. Their typical
task is to write an expository essay and not a “sentence standing alone”, and
to provide a deal of “supporting comments” for what they say. They know quite
well they can’t get away with writing: “My Essay on Sports. The man hit the
ball. The End.”
3.6. Why should language textbooks, even at university level, persist in
defining the “sentence” with criteria learners cannot reliably apply to their
own samples? The answer lies mainly in the complex mechanisms for producing and
marketing such books, as I have described in an ethnographic report on my own
experiences compiling a basic textbook (Beaugrande 1985a, 1985b). Prospective
publishers send the book draft to practising teachers who expect to find
familiar textbook explanations and who find them meaningful and reassuring by
virtue of already knowing how to recognise a “sentence”. The learners for whom
these explanations are not meaningful and whose language varieties are quite
different from the careful written prose expected in schools tend to get blamed
individually for being “illiterate” or “unintelligent”. These learners have no
self-confidence or leverage to put the true blame on the vague pedagogical
discourse or “meta-language” that communicates to insiders but not to
outsiders. Indeed, I found that my concerted attempts to practice a widely
accessible discourse were misunderstood to be “lowering the standards ” or
“catering to the ignorant”.
3.7. The only reliable criteria I have found for defining a “sentence” are
structural ones: “having as its core at least one independent finite verb with
its subject” (Willson 1980:29). These criteria are plain enough to most English
teachers and guide their relentless search and exposure of major “errors” like
“sentence fragments”. But the criteria may not be at all plain to ordinary
learners, who use terms like “subject” and “independent” in more everyday
senses. And the criteria are biased toward declarative sentences, whose use is
favoured in the writing of school assignments, and against imperative and
exclamatory ones, whose use is discouraged.
3.8. Moreover, the structural criteria shift the problem from defining the
“sentence” over onto defining the “Subject” and “Predicate” (or “finite verb”).1
The natural status of these two constituents is also complacently assumed,
viz.: “whatever its form, and however long or short, the sentence always
consists of two parts, the naming
part” and “the doing part” (Gartside
1981:239), where the “clause” is apparently meant, since a “long sentence”
would easily have more than these two. Or: “if you are asked to divide a
sentence into two parts, you will invariably divide it between Subject and
Predicate”; “we are aware of the noun-like quality of the Subject and the
verb-like quality of the Predicate, whether we can explain them or not” (Moody
1981:310), where the “invariable” outcome” and the tacit “awareness”
conveniently obviate the need to “explain”.
3.9. At all events, the explanations offered to learners are seldom usable.
You are told that “the Subject tells who or what the sentence is about, and the
Predicate tells something about the Subject” (Mills 1979:2), where the
distinction between “being about” versus “telling something about” sounds
rather slippery. Or, you are told that “the Subject is the word or words that
denote what we are talking or writing about”, whereas “the Predicate consists
of the word or words used to say something about the Subject and must contain a
verb, for otherwise no expression of thought is possible” (Gartside 1981:239f)
— where the basic sentence structure is declared, in a manner reminiscent of
Wittgenstein (cited above), indispensable for “thought” itself (but cf. 5.6).
Or again, you are told that “a Subject is a noun or noun equivalent that
performs an action or is in a particular state of being” and “usually appears before
the verb and determines the number (singular or plural) of the verb”, whereas
“the Predicate is the verb and all words related to it” (Corder 1979:463),
where confusion arises because “nouns” do not “perform actions” — agents do.
Once more, we find a mix of psychological and philosophical adumbrations which
are unlikely to be operational for ordinary learners.
4. The “sentence” and the “clause” in a reference grammar
4.1. The Comprehensive Grammar of
English (Quirk, Greenbaum, Leech, & Svartvik 1985) is a well-known
reference grammar seeking to bridge linguistics with pedagogy (cf. Beaugrande
1998b). Their position is circumspect and reserved, witness the hedges (my
italics):
In
the past, grammarians have aimed to define “sentence” as a prerequisite to
defining “grammar”, or to define “grammar” as a means of defining “sentence”.
But both approaches will be avoided here: indeed, neither of these terms can be
given a clear-cut definition. The
sentence is an indeterminate unit in
the sense that it is often difficult to
decide, particularly in spoken
language, where one sentence ends
and another begins. The term
“grammar” is indeterminate in the
sense that “what counts as a grammatical English sentence?” is not always a question which permits a decisive answer; and this is not only
because of the difficulty of segmenting
a discourse into sentences but because questions of grammatical acceptability inevitably become involved with questions of meaning, good or bad
style, lexical acceptability, or acceptability in context. (1985:47)
Here, the “indeterminacy” of the “sentence” is frankly admitted, but
without discarding the notion that “spoken language” does consist of
“sentences”; “the difficulty of segmenting a discourse into sentences” is
treated as a largely practical problem with no significant theoretical
implications. Also, the impression is created that the conception of the
“sentence” would be “clear-cut” if factors like “meaning”, “style”, “lexical”
judgements”, and “context” did not interfere.
4.2. In consequence, this reference grammar occasionally follows the
convention of linguists, noted in 2.14, to alter and tidy up language data as
part of the description and analysis. The most striking examples I found are
not for the “sentence” but for the “clause”, which is allowed to be
“subjectless” (1), “verbless” (2), or even both “subjectless and verbless” (3)
(1985:725, 996, 1120ff). These structures are said to result from “syntactic
compression” in which the verb is “omitted” but “understood” (1985:996).
(1) Susan telephoned before coming over [before Susan came over]
(2) There he stood, a tray in each hand [a tray was in each hand]
(3) Confident of the justice of their cause, they agreed to put their
case before an arbitration panel. [“Since
they were confident of the justice of
their cause…”]
When, along with John Algeo and R A Close, I “undertook the heavy task
of giving a detailed critique of the entire book in an earlier draft”, as the
authors put it (1985:v), I objected that a “clause” being “subjectless” or
“verbless” (or both) contradicts the authors’ own stipulation that “subject”
and “verb” are “indispensable to clause structure” (1985:50). Today I would
lodge a more general objection against the implied claim that altering one type
of structure into another type somehow explains it. We have seen how this claim
was made quite explicit and carried to extremes by the “transformational” or
“generative” grammars, where such alterations constituted the principal mode of
explanation (2.14).
4.3. The Comprehensive Grammar of
English is plainly not like these, but some idealisations are still being
applied to tidy up the grammar by assuming that one pattern which is
“understood as” another requires no place of its own in the grammar. A paradox
persists here between postulating an orderly pattern like the clause to be
“understood” and then bending that same pattern even to the point of allowing
either or both of its required constituents to be “omitted”, in order to avoid
granting a place in the grammar to less orderly patterns. Both linguistics and
language pedagogy should seek to formulate a realistic and “user-friendly
grammar” for describing real language data just as we find them and discovering
order within patterns that might seem disorderly (Beaugrande, in preparation).
5. Applying multiple sets of
criteria
5.1. The key step toward a reliable conception of the “sentence”, I submit,
is to assign it the status of an optional
unit that is not taken as “given
in advance” nor built into the very definition of “language” and “grammar”.
This step accords with Halliday’s conception of “paradigmatic” or “choice
grammars”, “the functional ones, with
their roots in rhetoric and ethnography”, and “organised around the text or discourse”, as opposed to the “syntagmatic”
or “chain grammars”, “the formal ones with their roots in logic
and philosophy”, and “organised around the sentence”
(1985:xxviii, xiii, xix, my italics). He has indeed proposed that we do not
“need to bring in the term ‘sentence’ as a distinct grammatical category”;
instead, we can use the term “to refer to the orthographic unit that is
contained between full stops” and thus to “a constituent of writing, while the clause complex is a constituent of
grammar” and “the only grammatical unit above the clause” (Halliday 1985:193).
5.2. Our prime question would then be: which
sets of criteria might be relevant for making (or not making) a given stretch
of discourse into a sentence, or for recognising it to be a sentence? At
least the following sets of criteria might be considered:
5.2.1. structural: a “sentence”
consists of an array of relations (“structures”) among units, e.g., the
“Subject” and the “Predicate” in an “independent clause”;
5.2.2. formal: a “sentence” matches
an array of formal symbols stipulated by a “formal grammar”;
5.2.3. logical: a “sentence” is an
“expression” derived via “rules” from a “logical system” of basic “axioms”;
5.2.4. philosophical: a “sentence”
is a “true or false statement” about a “state of affairs” in a “real or
possible world”;
5.2.5. cognitive: a “sentence” is a
“proposition” with a “predicate” and one or more “arguments”;
5.2.6. thematic: a sentence is a
pattern for relating the “theme” (or “topic”) conveying known or predictable
information with the “rheme” (or “comment”) conveying new or unpredictable
information;
5.2.7. intonational: a “sentence”
corresponds to a “tone group” uttered as a “prosodic” unit with a
characteristic pitch, e.g., rising or falling;
5.2.8. pragmatic: a “sentence” is
the expression of a “constative” or “performative speech act”;
5.2.9. behavioural: a “sentence” is
a separate segment within the “stream of speech”;
5.2.10. orthographic: a “sentence” is
an orthographic unit of written language whose outer boundaries and at least
some of its inner patterns are indicated in many writing systems by
punctuation;
5.2.11. stylistic: the sentence is
one of the key units for working out the style of an individual or group,
especially in literary discourse;
5.2.12. rhetorical: the sentence is
one of the key units for achieving rhetorical effects such as being expansive
or terse, brisk or relaxed, excited or calm, and so on;
5.2.13. registerial: the sentence is
a unit whose form and organisation adapt to differing “registers”, e.g., in an
official business letter as compared to a casual memo;
5.2.14. social: the sentence is a
unit of higher importance for persons in some social roles or positions than
for others, e.g., for a BBC radio announcer as compared to a barman in a rural
pub.
5.3. Moving from top to bottom, these sets of criteria roughly progress
along such familiar parameters as “grammar - logic - rhetoric” or “syntax -
semantics - pragmatics”, but the progression is not straightforward on either
parameter. Also, the terms have been subject to considerable variation and
inconsistency, e.g., “logic” and “logical” vacillating between form versus
content. Moreover, especially in formalist linguistics and philosophy, syntax
tends to encroach upon semantics, or semantics tends to get built right on top
of syntax, so that (in Halliday’s terms) issues of “choice” appear to be issues
of “chaining”.
5.4. In order to sustain a realistic link with the real data of practical
language experience, my discussion will explore these sets of criteria in the
inverse order, going from the bottom of my list toward the top. The criteria
toward the bottom indicate that people will typically tune their consciousness of sentence structure to their practices
for participating in specific discourse domains, especially ones relating to
their livelihoods.
5.5. Social roles (5.2.14) clearly
influence this process of tuning. A court messenger who formulates of legal
notices like (4) (Botswana Gazette,
10.9.1997) must follow prescribed sentence formats, whereas the street woman in
Cyprian Ekwensi’s (1961:25) Jagua Nana
is represented speaking a Nigerian variety of English with a relaxed and
improvised grammar (5) quite disparate from the sentence patterns of standard
English.
(4) In the matter between Barclay’s Bank of Botswana,
Plaintiff, and Naughty Karabo, Defendant, be pleased to take notice that
pursuant to the judgement granted by the above Honourable Court, the following
property will be sold by auction by Deputy Sheriff Benjamin Motswakhumo to the
highest bidders.
(5) “Freddie I sorry for wat happen. I shame too much. If I tell you I
no shame I telling you lie. So I begin tink, as I lay down in de cell. If to
say ah get me own man!”
In between extremes like these, the ratios between social roles and
consciousness of sentence structure can fluctuate to accommodate differing
types of situation, such as professional versus intimate.
5.6. Registers (5.2.13)
too are constrained chiefly by practical factors, such as the social roles of
the participants but also their training, education, interests, and so on. The
register of traffic signs like (6-7) is constrained to get messages across
quickly and usually does without sentences, whence the wildly implausible
quality of (7a), whereas the register of tourist guides may prefer complete but
unadorned sentence patterns, e.g. (8).
(6) narrow
bridge
(7) dangerous
curves ahead
(7a) be pleased
to take notice that pursuant to the judgement of the national highway
commission the curves ahead are declared dangerous.
(8) From October to March temperatures can soar up to 44°C, the average
being 35°C to 40°C during midday. Night temperatures seldom fall below 26°C.
During winter, daytime temperatures are about 27°C, lowering to 6°C at night. (Shell Tourist Guide to Botswana)
We see in (6-7) how easy it is to have an “expression of thought” with
no verb, which a textbooks cited in 3.9 vowed is “not possible”.
5.7.
Rhetoric (5.2.12) has had an ancient and venerable history, but has not
maintained a place of its own in such institutions as education, science, or
the humanities, having been suspected of encouraging showy or manipulative
language, and having been largely overshadowed by “grammar”. Short, unadorned
sentences can create the effect of telling a sober and unemotional story (9)
and allowing the audience to draw their own conclusions.
(9) Tuki pulled
himself up. He must get on the road. He would have to enter the Free State by
car. He began to walk. (Serote 1981:330)
Other rhetorical effects may be sought by punctuating
like sentences some units that would not be sentences by structural criteria,
e.g., to evoke a restless sense of the “adventure” supposedly provided by a
brand of car.
(10) Adventure. It’s a state of mind. A desire to journey beyond the
familiar and into the unknown. To carry on when the road doesn’t. (Mmegi [“reporter”] 12.9.97).
5.8. Style (5.2.11) is proximate to rhetoric, but has been mainly focused upon the
language use specific to an individual speaker or writer or a group of these,
whereas rhetoric has been more focused upon the audience. Most attention has
been given to literary styles that stand out. The opening of Alex La Guma’s Time of the Butcherbird (11) uses
elaborate sentence patterns and similes to evoke a memorable scene whose full
meaning only becomes evident later on when the story reveals that here is the
new “homeland” apportioned by the “Bantu Commissioner” to African villagers
whose ancestral homelands have been confiscated for a mining company. In
contrast, Jack Kerouac’s (1973:98) Visions
of Cody strings out very long, spontaneous sentences to evoke the
narrator’s feverish stream of thought (12), Or again, Mongane Serote’s To Every
Birth Its Blood consciously dispenses with sentence formats to invoke a
mosaic of images in the graveyard in a South African black township (13).
(11) The dust settled
slowly on the metal of the tank and on the surface of the brackish water it
contained, laboriously pumped up from below the sand; on the rough cubist
mounds of folded and piled tents dumped there by officialdom; on the sullen
faces of the people who had been unloaded like the odds and ends of furniture
they had been allowed to bring with them, powdering them grey and settling on
the unkempt and travel-creased clothes, so that they had the look of scarecrows
left behind
(12) I could keep the most complete record in the
world which in itself could be divided into twenty massive and pretty
interesting volumes of tapes describing activities everywhere and excitements
and thoughts of mad valuable me and it would really have a shape but a crazy
big shape yet just as logical as a novel by Proust
(13) Women, some in fresh black mourning clothes; all of us for some
reason, wearing casual clothes — men trying to walk straight, holding spades
and rakes; children, forever children, now and then playing, now having to
follow the elders, now being scolded; families, holding on to each other by
freshening the graves of their beloved; a hymn, a desperate prayer, whispers,
the wind, the silence of the dead.
5.9.
Orthography (5.2.10) offers numerous resources for organising sentences, but the
criteria for marking these off with punctuation are far from simple and have
not been systematically featured in linguistics or language pedagogy. If we
follow Bloomfield’s stipulation (cited above) that a sentence can be “spoken
alone”, then a written sentence would be marked off at points where major
pauses would occur in speech. But this stipulation is hardly adequate. The
extensive empirical research on speech pauses or hesitations (surveyed in
Beaugrande 1984a: 158-169) has demonstrated numerous contributing factors; and
a unit spoken alone (like Bloomfield’s exclamation “John!”) might well not constitute
a sentence by other criteria.
5.10.
We would do better by stipulating that pauses would
appear at sentence boundaries if a written text gets read aloud. Several
factors for pausing in spontaneous speech, such as planning what to say next,
would not apply there. But sentence boundaries would not be the only places to
pause, nor would the various pauses have the same length or the same role in
the intonation of the total sequence. We might well find longer pauses
appearing after a more definitive falling pitch if the end of sentence is also
the end of a paragraph or of some larger division such as a chapter. Such is
apparently the motive for switching from a long sentence to several short ones,
each in its own paragraph, to highlight the Pirx’s dramatic discovery of the
“man’s” identity (14). The added pause indicated by three dots contributes to
the same effect.
(14) Then, from out of the cabin staggered a writhing
hulk of a man in a brown uniform, his helmetless head bobbing around like a
blurry blotch, his face contorted in a mute shriek.
Pirx’s knees buckled.
It was…Boerst.
He had crashed into the Moon. (Stanis»aw Lem, Pirx the Pilot)
5.11.
The length
of written sentences to accommodate the register or domain of discourse, the
age or educational level of the intended audience, the writer’s style, or the
difficulty of the topic or subject matter, can hardly become a topic for
concerted research if these factors are disregarded by schools of linguists who
customarily construct their own short, simple sentences to keep the grammar
tidy (2.14).
5.12.
We might stipulate that simple or familiar content
tends to be presented in longer sentences, and difficult or unfamiliar content
in shorter sentences — what I have called the loading principle, in my
own account of punctuation (now revised in Beaugrande 1997). If the single
sentence in a sample like (15) from a popular science magazine were broken down
into very short sentences like (15a), the results seem dull or patronising.
(15) The bright light you see in the western sky right
after the sun has set and long before any stars are visible, is Venus — not a
plane, a balloon, or a UFO, as is often thought.
(15a) You can see a bright light in the western sky. The time is right
after sunset. That is long before the appearance of any stars. The light is
Venus. It is not a plane. Nor it is a balloon or a UFO. People have often
thought so, though.
In contrast, a long and complex sentence like (16)
from a draft on “critical discourse analysis” by Norman Fairclough and Lilie Chouliaraki
makes for difficult reading, and could be helpfully broken down, e.g. (16a).
(16)
Reflexivity demands both an appreciation of the specific value of the
intellectual relation to language (in explaining social continuity and
identifying the unrealised potential for social emancipation) and a vigilance
towards its dangers (notably in dressing up the viewpoint of government as the
viewpoint of science), and a sense of how the intellectual agenda is generated
out of and oriented to practical living which requires a continuous
thematisation of the links and differences between intellectual and practical
relations to language.
(16a) We need to reflect upon the how the intellectual
relation to language might explain social continuity and might identify the
unrealised potential for social emancipation. We also need to be vigilant
towards the dangers of that relation, notably how it might dress up the
viewpoint of government as the viewpoint of science. Further, we need to
consider how the intellectual agenda is generated out of and oriented to
practical living; and to thematise the links and differences between
intellectual versus practical relations to language.
The
revision still seems appropriate to the register.
5.13. Because orthography must accommodate other criteria
despite having its own peculiar functions as well, we cannot stipulate that any stretch of written text beginning
with a capital letter and ending in a full stop counts as a sentence. Rather, a
writer has the option of making it coincide with a sentence; and whether it
coincides or not can readily affect our other criteria such as rhetorical,
stylistic, thematic, cognitive, and so forth, just as the choice of type,
length, and complexity of sentences can do.
5.14.
The behavioural
criteria (5.2.9) suggested for the sentence by Bloomfield, Firth, and Pike
among others, have not been pursued very far in later linguistics or in
language pedagogy. Certainly, the contention that language behaviour naturally
consists of sentences is untenable. Still, language pedagogy might be able to
tap that behaviour in order to deal with sentences. For example, to find
Subject and Predicate of a declarative sentence like (17), learners can make up
a “who/what” question like (17a); the Predicate is all the words you used again
in the question (double underline), while the Subject is the rest (single
underline). As we see from (17b-c), having a part of the Predicate ahead of the
Subject is not the obstacle it is when Subject and Predicate are defined as the
first and the second part of the sentence.
(17) The northward march of the Voortrekkers was
a gigantic plundering raid. (Sol Plaatje, Native Life in South Africa )
(17a) What was a gigantic plundering raid? => Predicate: was a gigantic
plundering raid; Subject: The
northward march of the Voortrekkers
(17b) In the following years, the northward
march of the Voortrekkers was a gigantic plundering raid.
(17c) What was a gigantic plundering raid in the following years;?
=> Predicate: was a gigantic plundering
raid; + In the following years; Subject: The northward march of the Voortrekkers
Such a tactic exploits the learners’ already
conditioned language behaviour to make tractable an issue that otherwise
requires a deeper grasp of sentence grammar. I found it eminently effective on
pre-tests and post-tests at the University of Florida, where naive learners
improved their ability to recognize Subjects and Predicates by 500%.
5.15.
Again moving up my list of criteria in 5.2, pragmatics (5.2.8) might offer ways of
relating sentences to speech acts. To
be sure, the term “speech act” clearly implies a unit of spoken language; yet
theorists like Austin (1962) and Searle (1969) have not hesitated to assume
that the standard format of a speech act is still the “sentence”. This
assumption may well have been influenced by the older philosophical
inclination, cited in my list, to regard a “sentence” as a “statement” about a
“state of affairs”; and retaining the sentence-unit may have helped make speech
act theory attractive to the many linguists who were reluctant to venture
beyond the sentence boundary (2.14). Moreover, discussions of speech acts have
routinely been illustrated with single invented sentences. By way of contrast,
consider the threat uttered to a victim of police interrogation, in a format
that hardly qualifies as a sentence at all:
(18) Talk or the window. (Serote 1981:154)
For people who knew about the practice among South
African “security police” of throwing detainees out of high windows and then
claiming suicide, the threat could hardly have been more explicit.
5.16. Extensive research is still needed concerning
the pragmatic criteria for deciding how to make a given stretch of text into a
sentence, or when to end a sentence at a given point. Longacre’s work offers notable soiurces regarding non-English languages.
For a language like English itself, such an inquiry might be
constructively pursued by examining the placement of full stops in large corpora of authentic writing. We
may find that some types of lexicogrammatical choices or patterns tend to be
judged more suitable ends for sentences or to suggest more “finality” than
others do. In the data on the verb “warrant” I took from the “Bank of English”
at Birmingham University in July 1994 (reported in Beaugrande 1996),2
I noticed a high proportion of sentences ending with the pattern “warrant” plus
a direct object that was either a pronoun (12-22) or, more often, a fairly
simple noun phrase (23-28). Considerably
more interesting was the proportion of data where that end of the sentence
coincided with the end of that text or speaking turn, as indicated by quotation
marks and by the codes in diamond brackets: <LTH> for starting a text in
the same source (e.g. when going to the next item in a newspaper); <M +
digit> for starting the turn of male speaker with identification number;
<p>: for starting a paragraph; and <t> for starting a new text and
source.
(19) some of the costs erm just wouldn’t warrant it.<M01> Yeah. I suspect also
(20) action if our national interests warrant it. <t> Credibility with our
allies
(21) age limits for jobs when jobs do not warrant them. <LTH> As local government
(22) will not be long or bloody enough to warrant one.<p> Perhaps. Hitler said that
(23) not enough of its own character to warrant serious consideration. <LTH> The
(24) are large enough really er to warrant an assistant
manager.<M01> Mm.
(25) believed the need was so great as to warrant the expenditure.” A new sea wall
(26) jobs overall is rare enough to warrant no apology.<p> The IIE’s estimate
(27) are large enough really er to warrant an assistant
manager.<M01> Mm.
(28) six of those were serious enough to warrant prosecution.<M03> I mean some
This tendency might plausibly be related to the pragmatic force when
some authority of institution — shown by the data to be chiefly government,
judiciary, military, sports, business, science, and medicine — declares what
does or does not “warrant” what. The declaration is judged final, and nothing
more gets added, either to the sentence or to the text. Obviously, these few
data are merely suggestive, but they indicate some types of pragmatic criteria
we might want to explore on much larger scales.
5.17. Again further up our list, the significance for
English grammar of intonational criteria
(5.2.7) concerning the degrees of pitch, stress, volume, and pacing of sounds
was not fully appreciated until the publication of Halliday’s (1967) pioneering
research. The long delay may well have been encouraged through the tendency of
linguists to work with written language despite the priority officially
consigned to spoken language by Saussure, Bloomfield, and others, plus the
well-known difficulties of representing intonation in graphic formats. Halliday
reported that intonation is controlled partly by thematic criteria, to which I
shall turn in a moment; and partly by degrees of belief: falling pitch usually
means “certain”, rising pitch means “uncertain”, and so on.
5.18. The unit of intonation is not the sentence, but
the “tone group”, which is described as being the “melodic
unit of the language” and also the “realisation” of the “information unit”
(Halliday 1994:295). In the “unmarked or default condition”, “one information
unit will be coextensive with one clause”, but such is far from obligatory, as
we see from his own sample “marking the boundary of a tone group by a double
slash” and pauses with a single slash:
(29) ap/parently he / is // yes
// although I / don’t really / know / why//
So the boundaries of tone groups can at best give some
indications of the boundaries of
clauses rather than sentences.
5.19. Closely related to intonation are the thematic criteria (5.2.6) organising a
sentence in a pattern that relates the “theme” (or “topic”) of known or
predictable information with the “rheme” (or “comment”) of new or unpredictable
information. This “functional perspective” was first described for English by
the “Prague School”, whose perception was enhanced by the strongly thematic
organisation of sentences in Czech and Slovak (cf. Beaugrande 1992, 1998c). As advocated by Pike (2.8) This
“perspective” obliges us to keep looking beyond the boundaries of the single
sentence, as in this opening of a chapter in a novel:
(30) Alexandra is one of the oldest townships in South Africa. It is
closely related to Johannesburg. From the centre of the Golden City to the
centre of the Dark City is a mere nine miles. […] Everything that says anything
about the progress of man, the distance which man has made in terms of
technology, efficiency, and comfort: the Golden City says it well; the Dark
City, by contrast, is dirty and deathly. The Golden City belongs to the white
people of South Africa, and the Dark City to the black people. (Serote 1981:28)
At the start of the opening sentence, “Alexandra” is the prominent
theme, being also the major setting for the novel, whereas being “one of the
oldest townships in South Africa” in the high-informational rheme position
might not be generally known. The next sentence seems to indicate what Daneš
(1974:118) has called “the continuous theme”, i.e.. the same theme in
consecutive sentences, whist the rheme presents new and potentially surprising
content: the “close relation to Johannesburg”. But then the two cities are
consolidated into joint themes. The paired items “Golden City” and “Dark City”
appear three times in thematic positions after the first appearance has put
Johannesburg ahead of Alexandra in the syntax just as it is in wealth and
power; the first rheme invokes proximity (“a mere nine miles”) whilst the
second and third rhemes invoke disparity. The full irony of the “closeness” is
displayed by a series of polar contrasts of “Golden” versus “Dark”, “white”
versus “black”, and “technology, efficiency, and comfort” versus “dirt and
death”, underscored by parallel phrases like “the centre of the ___ City” and
“to the ___ people”. The irony is complete when we recall how much of its
wealth white Johannesburg owes to its exploitation of black labour in the
“townships”, and specifically for the “gold” mines.
5.20. The term “functional sentence perspective”,
though well established for decades, should be regarded as heuristic, since the
relations among theme and rheme may be relevant inside a clause or across as
sequence of sentences. When intonation is also taken into account, we might
even ponder whether a “grammar of spoken language”, which is now finally
beginning to materialise (cf. Brazil 1995; Carter & McCarthy 1995), might
differ from the grammar of standard written language to the extent of not
requiring the “sentence” to be a major unit after all. The resources of tone,
pitch, rhythm, pausing, and so on might prove sufficient to indicate the
organisation and boundaries of spoken utterances, as in the speech of Tolkien’s
(1991 [orig. 1954]:176f) villain Bill Ferny in The Lord of the Rings (is
“Stick-at-naught Strider” a transformed sentence?)
(31) His large mouth curled in a sneer. “Morning, Longshanks!” he said.
“Off early? Found some friends at last?” […] “That’s Stick-at-naught Strider,
that is! Though I’ve heard other names not so pretty. Pah!” He spat again.
Villainous “sneering” fits well with this abbreviated
grammar, and we can readily imagine what it might sound like with its whining
rises and falls. I also imagine a particular British dialect but for the sake
of tact I shall not say which one.
5.21. Still further up on my scale are the “cognitive” criteria focusing upon “propositions” (5.2.5). The
field of experimental psycholinguistics was for a long time dominated by the
complacent assumption that the sentence is the obvious and primary unit of
cognitive processing (cf. surveys in Clark and Clark 1977; Levelt 1978). But
with carefully designed experiments, Walter Kintsch demonstrated that human
processing of a text varies according to its organisation into “propositions”
rather than “sentences” (Kintsch and Keenan 1973; Kintsch 1974; survey in
Kintsch 1979). In free recall experiments at Kintsch’s laboratory at the
University of Colorado, the number of propositions recalled did not vary
between the original text (32) and the revised version (32a) with longer
sentences but the same propositions.
(32) A great black and yellow V-2 rocket 46 feet long
stood in a desert in New Mexico. Empty, it weighed five tons. For fuel it
carried eight tons of alcohol and liquid oxygen. Everything was ready.
Scientists and generals withdrew to some distance and crouched behind earth
mounds. Two red flares rose as a signal to fire the rocket [etc.]
(32a) With eight tons of alcohol and liquid oxygen as fuel to carry its
five-ton frame, a 46-foot black and yellow rocket stood ready in a New Mexico
desert. Upon a signal of two red flares, scientists and generals withdrew to
crouch behind earth mounds. [etc.]
Such findings naturally brought into focus the
problems of how to determine which propositions should be assigned to a text,
and how to express or label them (see Turner and Greene 1977). Kintsch adopted
the practical stance that the main requirement for making proposition lists is
that the methods be consistent among the researches and also among the several
versions of a text. But some of the criteria I have examined here suggest that
propositions should be ranked in importance, e.g., the thematic “rocket
standing ready” ranking well above the “rocket” being “46-foot”, or “black and
yellow”. In other experimental studies, pragmatic criteria were found to be
influential too, such as the perspective of a reader who is interested in
certain information (Anderson and Pichert 1978).
5.22. Yet the sentence might still play a major
cognitive role as one prominent format for expressing propositions. I have
suggested above that the length and complexity of sentences within a text ought
to be regulated to accommodate the levels of difficulty of the content for
potential readers. More training in this area should be provided in schools and
universities, especially for people who expect to be writing pedagogical or
technical materials later on.
5.23. Also, Kintsch and I found an interesting
tendency for the subject’s recall protocols to imitate the sentence style of
the versions they had read. A subject who read the original (32) recalled
(32b), whereas a subject who read the version (32a) with very long sentences
recalled (32c).
(32b) In a New Mexico desert, a V-2 rocket waited to
be launched. It was 60 feet tall and weighed 5 tons empty. The generals and
technicians launched two red flares signalling the launch.
(32c) With 8 tons of alcohol and liquid oxygen for the 5 ton rocket, the
rocket is signaled by 2 red lights and the scientists and generals crouch down
behind an earth mound, the rocket takes off [etc.]
Further work on the relations between sentences and
propositions may yet provide an empirical basis for the traditional notions of
a sentence being (or corresponding to) “a communication in words” (3.1) or “a
combination of words expressing a complete thought” (3.3).
5.24. The philosophical
criteria relating “true” sentences to a real or possible world (5.2.3) reflect
the optimistic aspiration that “truth must be the central notion of semantics”,
so that “to know the meaning of a sentence is to know under what conditions the
sentence would count as true” (Wiggins 1971:21, 17). As I have shown in a previous
analysis of philosophical discourse in this same journal (Beaugrande 1984b),
that aspiration interlocked with the pessimistic view that “the total
communicative content of the utterance” is “too complex to be accounted for”
(Wiggins 1971:20).
5.25. Here we evidently witness another retreat where
discourse was suspected of harbouring massive disorder because the real order
of discourse is not based upon the fully determinate ideal system that
linguists and philosophers have often aspired to find (2.17). The discussants predictably adduce
their own fictitious sentences to make context appear unimportant, such as
“Kallias is snub-nosed” (Wiggins 1971:20). But the real disorder of making
meaning depend on truth conditions soon emerges when we examine authentic data.
In my own samples, how could we reliably determine whether the Nigerian street
woman in (5) (a notorious liar in the story!) is truly “sorry” and “ shame”; or
what makes a “bridge” truly “narrow” (6), or “curves” truly “dangerous” (7) without smashing my car; or
whether “adventure” is truly “a state of mind” (10); or whether “the most
complete record in the world” “would really have” “a crazy big shape yet just
as logical as a novel by Proust” (11); or what truly “freshens the graves”
(12); and so on. Without contexts, such data would indeed become the mere
“heterogeneous mass” Saussure nervously imagined to constitute the “speech
facts” of a language (2.16). Wiggins
even admits as much when, in order to discount contexts, he stipulates that
“the purview of semantics” must include “theoretically possible readings” that
are “too absurd to occur to us” (1971:25).
5.26. Near the top of my listing, the “logical” and “formal” criteria for sentences (5.2.2-3) have been highlighted by
formalist schools in linguistics and language philosophy. In most formalist
work, the term “sentence” has been used in senses quite distinct from those I
have reviewed so far. The “sentence” is routinely defined to be a “string of symbols” whose chief (or even
sole) requirement is to be “well-formed”, i.e., “derived” by means of “formal
rules” from the basic “axioms” that are given a priori. In most such systems,
the operations of this “derivation” do not at all purport to take account of who might have uttered the sentence and
when, where or why, what the speaker or hearer might have meant to communicate
or wanted to achieve, what is known or new, and so forth. All this would again
be viewed as a “heterogeneous mass” of disorder.
5.27. How the criteria proposed in formal or logical
research might relate to all the other criteria I have surveyed so far is a
moot question insofar as that research has expressly excluded them all from
“linguistic theory”, presumably regarding them as sources of the massive
disorder linguists since Saussure have seen in discourse (2.16f). “A consequence of restricting
linguistics to purely formal matters was an extreme narrowness of focus”; “the
results of a linguistic analysis are not taken to be relevant to an
understanding of the capacities and fundamental characteristics of human
beings” (Maclay 1971:163). What they are
“relevant to” is a question for which my own detailed analyses of the
discourses of formalist linguists (Beaugrande 1984b, 1991, 1998a) have yet to
find a convincing answer.
5.28.
Formal-logical schools of linguistics and philosophy
are especially prone to the trade-off I have diagnosed above. The exclusion of
context is expected to produce a deterministic ideal system of perfect order,
but in fact generates disorder by erasing many of the constraints which
determine the organisation of sentences but which are only decided upon the
plane of the discourse (2.18). To
compensate for the erasure, vast apparatuses of arbitrary “rules” and
“features” must be constructed, as in (33) (Lakoff 1971:275),
where the formula states it is true for all members (x’s) — ‘"’ being a ‘universal quantifier’ — that each x loves x’s wife. How this apparatus could be
extended to the sometimes patently absurd invented sentences like (34) (Lakoff
1971:271, disputing with Postal) I
cannot imagine.
(33) John and Harry love their wives. = " (x1,
x2) [x loves x’s wife]
x
e
(34) John reminds me of a gorilla with no teeth though I don’t perceive
any similarity between John and a gorilla with no teeth.
Instead of admitting that the exclusion of context is
a fundamental error which cannot remedied by “revising” or “extending” the
“syntactic theory”, many formalists, including Chomsky himself, have steadily
withdrawn even further from context by shelving the task of describing the
structures of English sentences in favour of describing “the structure of
mental representations”, where “there are no constructions, there are no rules”
(Chomsky 1991:93, 81). Or, they have retreated into a “minimalism” whose name
paradoxically advertises the ability to account for the fewest possible issues
as proof of sound scientific credentials (Abraham et al. eds. 1996).
5.29. The structural
criteria of a sentence having a “Subject” and a “Predicate” in at least one
“independent clause”, placed at the very top of my list (5.2.1), remain the
most widely established. They have been referred to in various ways and labels
by nearly all of the other sets of criteria. I hope to have shown that they
should be regarded not as an account of the “real or essential sentence” but as
tools for a multi-functional theory or method that also integrates the other
criteria. The patterns of “Subject” and “Predicate” present the most convincing
signal of a language user having chosen the option to form a sentence; our
theory should now explore the other criteria in order to determine in steadily
finer detail why and how that choice might have been made and what the
consequences might be.
6. Toward a
multi-functional theory
6.1. What are the prospects for an integrative
multi-functional theory of the sentence? Like many other prospects, these
depend upon the outcome of a general contention between division versus
integration all across linguistics and its neighbouring disciplines today. On
the divisive side are all the
approaches which aspire to describe or explain human language in terms of an
orderly idealisation constructed by means of dichotomies and abstractions to
exclude all aspects or factors that appear to foster disorder by depending on who
speaks to whom, when, where, and why (cf. 2.17). Here, the sentence has readily
been enlisted to bridge the gap between real language and ideal language: as
the prime unit of “utterance” in “language behaviour” and as the prime unit
within a ideal system or “infinite set”. This ambivalence, already detected but
not properly appreciated by Saussure (2.3), has accorded the sentence a
remarkable career at the centre of theories so disparate from each other that
the meaning of the term “sentence” has undergone numerous transmogrifications.
As long as these were not acknowledged, a theory purporting to account for
sentences could also purport to account for many other issues or indeed for all
of a language — even when that same theory bristled with peremptory exclusions
like the “observed use of language” “surely cannot constitute the
subject-matter of linguistics” (Chomsky 1965:4) (2.9).
6.2. Those same theories have been radically
confrontational in their zeal to advance themselves at the expense of their rivals,
precisely because their divisions and exclusions had stymied most other means
of argument, such as testing a theory against the “observed use of language”.
Linguists grew strategically adept at inventing (or ‘constructing’) just
those sentences needed for rebutting opponents in the effusive debates that
crowded the agenda after the observation and analysis of “ordinary language
behaviour” had been pushed to the margins. Such divisive tactics have
collectively prevented the consolidation of a cumulative theory either of
language or of sentences, and are thus patently unsuited for developing the
type of integrative theory I have proposed.
So we can see why these linguist have recently been losing interest in
sentences and turning to “mental representations”.
6.3. On the integrative
side are all the approaches that strive to describe real language on the
basis of real discourse. No aspects or factors are excluded if they are
observed to be relevant to human interaction in or by means of language and
discourse, and they are all assumed to support order far more than disorder
(2.186). They cannot therefore
generate the overwhelming complexity that the linguistic theories on the
divisive side have invoked as a pretext for “abstracting away”; the complexity
of discourse is successfully managed by ordinary participants most of the time,
and our theories must reflect this management.
6.4. Such integrative approaches are obliged to rely
on multiple disciplines, including linguistics along with anthropology,
sociology, ethnography, history, psychology, poetics, semiotics, and so on. In
the past, these have been tapped for interdisciplinary work in such areas as
discourse processing and cognitive science. The next stage could seeking higher
integration by developing approaches that are transdisciplinary in the sense that their theories, concepts,
terms, and methods are consciously designed throughout to be used in multiple
disciplines (Beaugrande 1997a).
6.5. In that framework, the prospects for an
integrative theory of the sentence would be quite auspicious. A framework could
be provided for situating and combining previous investigations of the
sentence, whilst granting it no more importance than can be justified with
authentic data.
6.6. Pending such a framework, I would predict that
each set of criteria I have surveyed will be found relevant for describing the
status of the sentence, but only in rich interaction with other sets. The most
essential requirement will be to examine extremely large corpora of spoken and written
data, which have only recently become manageable on powerful computers and
which clearly mark the various boundaries of sentences or utterances, as we saw
in the COBUILD data (18-28). Corpus data have repeatedly provided insights into
regularities that are not open to the unaided intuition deployed for inventing
sentences (Francis and Sinclair 1994), whether banal like (33) or bizarre like
(34). The reason should be perfectly plain: many regularities are decided not
on the plane of the system (e.g. the “grammar”) but on the plane of the actual
discourse (2.18). The discourse of
inventing sentences is a peculiar one, and cannot be expected to generate data
representative of real contexts where the many sets of criteria I have listed
interact in natural ways to integrate diversity and complexity.
6.7. The status of the sentence is just one among
many issues in the science and pedagogy of language where the deliberate choice
between division versus integration is becoming steadily harder to evade or
postpone. The linguistics of division, of dichotomies, of exclusion and
confrontation, has dominated many conferences, journals, and university
departments for four decades, and has still not even supplied a remotely
complete or accurate “grammar of sentences” — only fragments, proposals, and
notations for handfuls of unrepresentative invented data. Surely the wise
choice should now be for a linguistics of integration.
NOTES
1 I shall capitalise “Subject” and “Predicate” henceforth to distinguish
them from other words of the same appearance.
2 I am deeply indebted to John Sinclair for his guidance in working with
the Bank of English.
REFERENCES
Abraham, Werner et al., Eds.
1996. Minimal ideas: Syntactic studies in
the minimalist framework. Zaragoza: P\rtico LibrerRas.
Anderson, Richard &
James Pichert. 1978. “Recall of previously unrecallable information following a
shift in perspective.” Journal of verbal
learning and verbal behavior 17:1-12.
Austin, John. 1962. How to do things with words. Cambridge:
Harvard UP.
Beaugrande,
Robert de. 1984a. Text production.
Norwood, NJ.: Ablex.
____. 1984b. “Linguistics as
discourse: A case study from semantics.” WORD
35:15-57.
____. 1985a. Writing step by step. New York: Harcourt
Brace Jovanovich.
____. 1985b. “Composition
textbooks: Ethnography and proposal.” Written
communication 2:391-413.
____. 1992. “The heritage of
functional sentence perspective from the standpoint of text linguistics.” Linguistica Pragensa 34/1-2:2-26 and
55-86.
____. 1996. “The
‘pragmatics’ of doing language science: The ‘warrant’ for large-corpus
linguistics.” Journal of pragmatics
25:503-35.
____. 1997a. New foundations for a science of text and
discourse. Greenwood, CT: Ablex.
____. 1997b. “On history and
historicity in modern linguistics: Formalism versus functionalism revisited.” Functions of language.
____. 1998a. “Performative
speech acts in linguistic theory: The rationality of Noam Chomsky.” Journal of pragmatics.
____. 1998b. “Society,
education, linguistics, and language: Inclusion and exclusion in theory and
practice.” Linguistics and education.
____. In preparation. A friendly grammar of English.
Birkhoff,
George David. 1950. Collected
mathematical papers. New York: American Mathematical Society.
Bloomfield,
Leonard . 1933. Language. Chicago: U
of Chicago Press.
Brazil, David. 1995. A grammar of speech. Oxford: Oxford UP.
Brown,
Gillian & George Yule. 1984. Discourse
analysis. Cambridge: Cambridge UP.
Carter,
Ron & Michael McCarthy.
1995. “Grammar and the spoken language.” Applied
linguistics 16: 141-58.
Chomsky,
Noam. 1957. Syntactic structures. The
Hague: Mouton.
____.
1965. Aspects of the theory of syntax.
Cambridge: M.I.T. Press.
____.
1977. Language and responsibility
(with Mitsou Ronat). New York: Pantheon.
____.
1991. “Language, politics, and composition.” Interviews: Cross-disciplinary perspectives on rhetoric and literacy.
Ed. Gary Olsen and Irene Gales. Carbondale: Southern Illinois University Press.
Pp. 61-95.
Clark,
Herbert, & Eve Clark. Psychology and
language. New York: Harcourt, Brace Jovanovich, 1977.
Corder,
Jim. 1979. Contemporary writing.
Glenview: Scott Foresman.
Daneš,
František. 1974. “Functional sentence perspective and the organization of the
text.” Papers on functional sentence
perspective. Ed. František Daneš. Prague: Academia. Pp. 106-28.
Ekwensi, Cyprian. 1961. Jagua Nana. London: Hutchinson.
Firth, J.R. 1957. Papers in linguistics 1934-1951. London:
Oxford UP.
___.
1964 [orig. 1930 and 1937]. Tongues of
men and speech. Ed. Peter Strevens. London: Oxford UP.
______.
1968. Selected papers of J.R. Firth
1952-1959, ed. Frank R. Palmer. London: Longman.
Francis, Gill & John
McHardy Sinclair. 1994. “‘I bet he drinks Carling Black label’: A riposte to
Owen on corpus grammar.” Applied
Linguistics 15:190-200.
Gartside, L. 1981. English for business studies. London:
Pitman.
Glazier, Terersa Ferster.
1981. The least you should know about
English. New York: Holt, Rinehart, & Winston.
Halliday, M.A.K. 1967. Intonation and grammar in British English.
The Hague: Mouton.
____. 1985. Introduction to functional grammar.
London: Arnold.
____. 1994. Introduction to functional grammar.
Second revised edition London: Arnold.
Kerouac, Jack. 1973 [orig.
ca. 1960]. Visions of Cody. London:
Andre Deutsch.
Kintsch, Walter. 1974. The representation of meaning in memory.
Hillsdale, NJ: Lawrence Erlbaum Associates.
____. 1979. “Semantic
memory: A tutorial.” Attention and
performance VIII. Ed. Raymond Nickerson. Hillsdale, NJ: Lawrence Erlbaum
Associates.
____ & Janice Keenan.
1973. “Reading rate and retention as a function of the number of propositions
in the base structure of sentences.” Cognitive
psychology 5, 257-274.
La Guma, Alex. 1979. Time of the
butcherbird. London:
Heinemann.
Lakoff, George. 1971. “On
generative semantics.” Semantics. Ed.
Danny Steinberg and Leon Jakobovits. London: Cambridge UP. Pp. 232-296.
Lem, Stanis»aw. 1981. Tales of Pirx the Pilot.
New York: Bard.
Levelt, Willem. 1978. “A
survey of studies in sentence perception: 1970-1976.” Studies in the perception of language. Ed. Willem Levelt and
Giovanni Flores d’Arcais. New York: Wiley. Pp. 1-74.
Longacre, Robert. 1964 [1958]. Grammar Discovery
Procedures. The Hague: Mouton.
____. 1970. Discourse, Paragraph, and Sentence
Structure in Selected Philippine Languages. Santa Ana: Summer Institute of
Linguistics.
____. 1976. An
Anatomy of Speech Notions. Lisse: de Ridder.
____et al. 1990. Storyline
Concerns and Word Order Typology in East and
West Africa. Los Angeles: UCLA Dept. of Linguistics.
____1996 [1983]. The
Grammar of Discourse. Second Edition. New York: Plenum.
Lyons, John. 1977. Semantics. Cambridge: Cambridge
University Press.
Maclay, Howard. 1971.
“Overview.” Semantics. Ed. Danny
Steinberg and Leon Jakobovits. London: Cambridge UP. Pp. 157-182.
Mills, Helen. 1979. Commanding sentences. Glenview: Scott,
Foresman
Moody, Patricia. 1981. Writing today. Englewood Cliffs:
Prentice-Hall.
Pike, Kenneth. 1967. Language in Relation to a
Unified Theory of the Structure of Human Behavior. The Hague: Mouton, 1967.
Quirk, Randolph, Sidney
Greenbaum, Geoffrey Leech, & Jan Svartvik. 1985. A Comprehensive grammar of English. London: Longman.
Sampson, Geoffrey. 1980. Schools of linguistics. London:
Hutchinson.
Sapir, Edward. 1921. Language. New York: Harcourt, Brace,
& World.
Saussure,
Ferdinand de. 1969 [orig. 1916]. Course
in general linguistics (transl. Wade Baskin). New York: McGraw-Hill.
Searle, John. 1969. Speech acts. Cambridge: Cambridge UP.
Serote, Mongane. 1981. To every birth its blood. Braamfontein:
Ravan Press.
Thom, René. 1989 [orig.
1972]. Structural stability and
morphogenesis (trans. D.H. Fowler). New York: Addison-Wesley.
Tolkien, J.R.R. 1991 [orig.
1954]. The lord of the rings. London: Harper Collins.
Turner, Althea, & Edith
Greene. 1977. The construction and use of
a propositional text base. Boulder: University of Colorado Institute for
the Study of Intellectual Behavior Technical Report 63.
Vukovich, Adolf. 1970. “Zur
Metrisierung der Ästhetik.” Text
Bedeutung Ästhetik. Ed. Siegfried J. Schmidt. Munich: Bayrischer
Schulbuchverlag. Pp. 189-216.
Wegener, Philipp. 1885. Untersuchungen über die Grundlagen des
Sprachlebens. Halle: Niemeyer.
Wiggins, David. 1971. “On
sentence-sense, word-sense, and difference of word-sense.” Semantics. Ed. Danny Steinberg and Leon Jakobovits. London:
Cambridge UP. Pp. 14-34.
Willson, Robert. 1980. Writing: analysis and application. New
York: Macmillan.
Wittgenstein, Ludwig. 1963
[orig. 1922]. Tractatus
Logico-Philosophicus. Frankfurt: Suhrkamp.