Structural Sentence Types

Author(s): Sheena F. Gardner and Sándor G. J. Hervey

Source: La Linguistique , 1983, Vol. 19, Fasc. 2 (1983), pp. 3-19
Published by: Presses Universitaires de France

Stable URL: https://www.jstor.org/stable/30248926

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Presses Universitaires de France is collaborating with JSTOR to digitize, preserve and extend
access to La Linguistique

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
 STRUCTURAL SENTENCE TYPES

Sheena F. GARDNER and Sandor G. J. HERVEY

University of St. Andrews, Scotland

Both in past and in current usage, the term "sentence" has

tended to be one of the least well-defined terms of linguistic
theory. While at one end of the scale the term may designate a
concrete speech event produced and/or accepted by a native
speaker, at the other end of the scale it may refer to a highly
abstract and formalised object containing a full elaboration of the
"constant" linguistic (or, for that matter, logical) constituents
and relationships that determine its identity as an idealised type
Owing to the lack of consensus about the usage of the term
"sentence" it would be inappropriate, in seeking a firm starting-
point for the present discussion, to fall back on some "primitive"
concept of sentence. We shall, instead, have recourse to a particular
definition of the notion "sentence", remembering, of course, that
definitions, if they are to be appropriate at all, can only be so
within the framework of particular theories: they must, in other
words, voluntarily abandon the appearance of offering "uni
versal" solutions to globally conceived problems. (Far too much
of current linguistics still suffers from the presentation of highly
theory-bound notions and definitions as though they were appro
priate to general problems of "linguistics", rather than to specifi
problems in particular linguistic theories.)
The definition of "sentence" which we shall take as our fixed
point in the rest of this article is: "Signum such that it is a self-
contained vehicle for conveying messages."' This definition is

i. Jan W. F. MULDER and Sindor G. J. HERVEY, The strategy of linguistics, Scottish

Academic Press, 198o, p. 56.
La Linguistique, vol. Ig, fasc. 2/1983

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
 4 S. F. Gardner and S. G. J. Hervey

frankly and voluntarily theory-bound-though it is

noting that it is motivated by what may be the o
consensus about the way the term "sentence" is g
namely, the "completeness" of sentences as units i
munication. (Whatever sense different linguists may
term "sentence", they conceive of sentences as unit
by a complete communicative autonomy. The tradit
what naive, way of calling attention to this featur
cative autonomy is found in the often re-iterated
"a sentence conveys a complete thought".)
The already rather theory-bound consideration
it theoretically needful to adopt the definition propos
directly concerns the generally recognised commun
omy of sentences (i.e. their property of being rea
that can, without the addition of any further feature
in actual communication). If linguistic sentence
complete blue-prints for speech events-and this is
by the stipulation that they are "self-contained vehicle
messages"-the question of identifying the proper
them their communicative autonomy assumes cent
importance.
All semiotic systems (systems of conventions for communi-
cation)' contain sentences in the above definition of the term: the
traffic sign " " is, by virtue of its communicative autonomy, a
sentence; so is, forthe same reason, the Morse-code signum "I ... /1',
the number-writing expression "124.85", the deaf-and-dumb
finger-spelling sign denoting the alphabetic letter "t", and so
forth.4
In certain systems, all conventionally meaningful elements
(signa) are automatically endowed with communicative autonomy.
In this way, for instance, every well-formed combination of flag
positions in semaphore is the form of a signum, and every one of
these signa is a sentence, capable of being used in isolation to
denote an alphabetical letter. In such systems communicative

2. Ibid., p. 56.
3. Ibid., p. 42.
4. Jan W. F. MULDER, Sets and relations in phonology, Oxford, Clarendon Press, 1968.
In the terminology of French functionalism such non-linguistic "sentences" are referred
to as "shmes"; cf. E. BUYSSENs, La communication et l'articulation linguistique, Paris, Brussels,
PUF/PUB.

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
 Structural sentence types 5

autonomy is a mere matter of memb

signa.
In other systems, all well-formed combinations of a certain
basic set of signa are, eo ipso, autonomous vehicles for communi-
cation. For instance, the traffic sign E consists of the two
component signa "blue square -* permission" and "P -- parking".
These component signa are not by themselves capable of conveying
self-contained messages; they cannot be found in isolation as
sentences. Their combination, and the same goes for any well-
formed combination of analoguous component signa, is, however,
a sentence regularly used to convey messages in isolation. In such
systems, communicative autonomy is an automatic property of
well-formed constructions and is imparted to signa by the mere
fact of their grammatical well-formedness.
If, in written English, we compare
"he is a fool"

with either "He is a fool."

or "He is a fool?"

-noting that the former is not, in itself, a

in conveying a written message, whereas
are led to the conclusion that in this sy
(communicative autonomy) is not a mere m
(in this case: syntactic) well-formedness.
is a fool" is, indeed, syntactically well-for
such, a sentence-for it may well be, as it
stituent of a more complex syntagm of t
(i.e. it lacks the completeness of "sente
additional features of capitalising the first le
either a "full stop" or a "question mark" a
is a fool." and "He is a fool?" as fully-fl
attribution of communicative autonomy in th
the special function of additional "senten
with the syntactic constructions which they,
sentences.

One important lesson to be learnt from this is that ther

semiotic systems in which the "formation" of sentences is
function of syntactic complexity, or of the production of
priate types of morphologically or syntactically well-formed g

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
6 S. F. Gardner and S. G. J. Hervey

matical construction. In such systems, one must consid

and independently of each other
(a) the grammatical subsystem(s) whose role is mere
bining of constituent signa into well-formed com
and

(b) a further "para-syntactic" subsystem whose ro

among others, the "formation" of sentences.

In standard structuralist theories of language su

ation between autonomous sub-systems of gramm
absent-on the contrary, it is usual to treat sentence
output of a single structural hierarchy extending from
through word, phrase and clause to the ultimate and
of the sentence.
Some structuralist theories do, in spite of includ
in the same structural hierarchy with morpheme, wor
give a degree of recognition to the fact that the "f
sentences is not a straight-forward matter of increasing
complexity-whereas this is the case for the formation
words entirely out of morphemes (in a constructi
with one another), and for the formation of phrase
of syntactic constituents, ultimately words (in synt
with one another). Tagmemic theory,5 for instance,
aware of the basic difference between representing the
of, say, a phrase by a combination of lower level const
representing the formation of a sentence by the su
of an Intonation on a suitable grammatical entity. T
have merely omitted to follow out the implications of
ence to their logical conclusion-namely that the p
the level of sentence is not analoguous with that fro
to word, or word to phrase, but that "sentence for
operation of an ontologically different type.
Using the terminology of Tagmemics, one way
out this difference is to note that

(a) in the morphological formation of words, the constituents

involved are all of the same ontological nature (morpheme +
morpheme),

5. W. A. COOK, Introduction to tagmemic analysis, New York, Holt, Rinehart & Winston,
1969.

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
Structural sentence types 7

(b) in the formation of phrases (syn

stituents are, again, all of the s
(word + word),
but that

(c) where a sentence is formed by

Intonation, there is a clearcut distincti
nature of the "base" over which an In
on the one hand, and the superimpose

This formulation of the difference

syntactic construction and "sentence
our adoption of a theoretical norm
municative autonomy in natural langu
construction as such. It also highligh
that the dimension required for the
of sentences is separate from the hierar
in natural languages.
An example may illustrate the p
languages, mere syntactic complexit
tencehood; i.e. that a syntactic con
complex and well-formed, is not a
particular type of conventional spee
be readily agreed that none of the
"tales", "strange" or "huge" has in
autonomy we have required of a sen
elements we can form ultimately the
"huge men tell strange tales". Setting
sideration that this syntagm can b
proposition, and seems, in that sen
thought", it is still clear that this syn
a degree of syntactic autonomy, is no
print for a type of speech act (i.e. n
for conveying a particular type of m
mere syntactic structure of "huge m
not confer sentencehood on this synt
syntagms, which by virtue of the cla
ositions" might, of all syntagms, have t
called "sentences", do not have intrin
them sentences in our definition. On
of extrinsic para-syntactic features, s

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
8 S. F. Gardner and S. G. J. Hervey

intonation, could the syntagm "huge men tell st

converted into a sentence.
The "grafting on" of features extrinsic to syntactic entities,
whether or not the resultant units constitute sentences, cannot be
part of syntax itself, but requires the isolation of a particular
"para-syntactic subsystem" operating on "syntactic bases" and
converting them into "para-syntactic units", each made up of a
"base" and "para-syntactic features". While not all para-syntactic
features are of a type capable of conferring communicative auton-
omy, the latter is, in natural languages, generally a function of
certain para-syntactic features.
The para-syntactic subsystem forms part of grammar, where
it "interlocks" with syntax; broadly speaking, the latter provides
the "bases" onto which para-syntactic features are "grafted".
Both because para-syntax combines entities of a different onto-
logical nature and because of its additional sentence-forming
potential, this subsystem is more than a mere extension of syntax:
it requires descriptive methods and models that are quite unlike
those of syntactic description.
As suggested above, the ultimate stage of sentence-formation
in natural languages should be described in terms of "grafting"
features that have the function of conferring communicative
autonomy (i.e. sentential para-syntactic features) onto signa pro-
vided by the grammar (i.e. onto sentence bases). The resulting
sentences should be complete blue-prints for all that is semiotically
relevant about the speech acts that realise them-i.e. there should
be no semiotically relevant feature of a speech act that is not
represented in the sentence whose realisation that speech act is.
The two basic types of sentential para-syntactic feature are,
on the one hand, intonational properties that by supra-segmental
means confer sentencehood (and, at the same time, characterise
sentences as being of a particular type; e.g. declarative, inter-
rogative), and, on the other hand, sentence markers that resemble
syntactic entities by having the same kind of discrete phonological
realisation, but which, rather than contracting syntactic relations
within the base, confer sentencehood by segmental means (also,
at the same time, characterising sentences as being of a particular
type; e.g. "est-ce que" in French is a sentence marker conferring
sentential autonomy on the sentences of which it is a part, at the
same time characterizing them as interrogative).

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
Structural sentence types 9

The description of sentential intonat

complex matter, and will not be pursued
it to say that one of the major problems
sentence intonation involves distinguish
ingful supra-segmental contours of mo
purely emotive or "naturally" expressi
of voice", "tempo", "emphasis", etc.
to call attention to the problem here,
pursue its solution, which, in any case, w
specific matter.
The status of sentence markers seem
that while their function is similar to the function of sentence
intonations, their form is similar to the form of syntactic constitu-
ents. By virtue of this ambivalence, they have continued to puzzle
structural linguists working with a purely syntactic model of
sentences. A typical method of sidestepping the issue in such
models has been to refer to them as "particles". A paradigm
example of such particles is the so-called "question-particle" ma
in Standard Chinese. When this particle is grafted onto a syntactic
base, say, the syntagm "ta laile" (he/she has come), a sentence is
formed, in the sense that the resultant is a complete blue-print
for an interrogative speech act; no further functional features
need to be added in order to achieve this result. (Any additional
features of supra-segmental modulation would, in this case, be
regarded as non-functional, concomitant properties of realisation.)
We shall assume, in fact, that signa in natural languages are
identified as sentences either by a particular intonation, or by a
sentence marker. This should not, however, be taken to imply
that the category of sentence markers is necessarily represented
in every natural language. On the other hand, even in those
natural languages that do make use of sentence markers there
are sentences that do not contain sentence markers, and whose
sentencehood is conferred on them by sentential intonations.
While we take all sentential features to be para-syntactic, the
converse is not the case. In other words, there are para-syntactic
features that have no sentence-forming potential; for instance,
clause-intonations are of a para-syntactic nature without having
the capacity to confer communicative autonomy on the bases
that they accompany. Whether para-syntactic features have a
sentence-forming potential or not, they all have an equivalent

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
 Io S. F. Gardner and S. G. J. Hervey

function with regard to contributing to the denotation

sentences in which they figure.
It may be thought that certain types of syntactic constru
(possibly with certain characteristic properties of linea
ation) are intrinsically predisposed to constituting sente
particular kind, which would mean that they have i
sentential properties. For instance, one may imagine th
combination "come here" is intrinsically predisposed to
imperative sentence. If this were strictly correct, we sho
to admit that "come here" is, in itself, already a senten
that its concomitant properties of (imperative) supra-seg
modulation are redundant features of realisation, autom
entailed by the intrinsically imperative, sentential n
"come here". This, however, is not strictly correct. Alt
"come here" is frequently associated with the imperative
"Come here!", it can also correspond to the base of an in
tive or a declarative sentence. Whether "come here" is, in
instance, the base of an imperative sentence still depends, th
the grafting on of an appropriate sentence intonation. I
we reject the notion that, in natural languages, syntacti
nations may possess intrinsic sentential properties-such prop
are always grafted onto them extrinsically.
Sentences may be classified either according to the na
sentence-bases, or according to the nature of their sen
features. Theoretically, the latter only produces two ma
egories: that of sentences formed by intonation and that
tences formed by sentence markers. Descriptively, of co
further classification of the sentences of a language acco
type of intonation and/or sentence marker is an import
of the description of that language. This, however, is not
we intend to pursue in the present article. Instead,
concentrate on the other dimension of classification: that of the
nature of sentence-bases, that is of entities onto which sentential
features may be grafted in such a way that the resultant is a
sentence. The classification itself will shed further light on the
way the term "sentence-base" is to be understood.
The constitution of sentence-bases turns out to be a far more
complex matter than it may seem at first sight. For one thing,
sentence-bases are not always directly furnished by syntax. There
are, in addition, also sentence-bases that contain non-sentential

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
Structural sentence types 1 I

para-syntactic features, i.e. sentence-b

para-syntactic units; as well as sentenc
plereme-like entities that, on examinat
syntactic. Furthermore, sentence-bases
single syntactic, para-syntactic or "quas
to the juxtaposition of several such enti
The latter consideration leads us to establish an initial
dichotomy between simple and compound sentence-bases; a si
sentence-base being one that corresponds to a single base-uni
opposed to a compound sentence-base which corresponds to two
more juxtaposed base-units.
We shall clear the ground for dealing with compound sentenc
bases by first establishing a typology of the units capab
occurring (in natural languages) as simple sentence-bases. T
are, as we have indicated above, three main types of si
sentence base:

(a) syntactic (corresponding, broadly speaking, to a syntactic

entity),
(b) para-syntactic (i.e. corresponding to a para-syntactic unit
formed by the grafting of non-sentential para-syntactic fea-
tures onto a syntactic entity),
(c) a-syntactic (i.e. corresponding to a "quasi-syntactic" entity
resembling a plereme, but, unlike pleremes, not capable of
entering into syntactic relations with any syntactic constituent).

An example of a syntactic sentence-base would be "he --

smokes" (i.e. a predicative-based syntagm) as the base of the
sentence "He smokes?".
In the case of the sentence "Can he sing!" (uttered as an
exclamation), the sentence-base is not considered to be merely
the underlying syntagm "he -+ (can -*sing)". It is, rather, a
para-syntactic unit formed by that syntagm together with a
functional para-syntactic feature which consists in the specification
of a particular linear arrangement of the constituents of that
syntagm. Thus, the two sentences "Can he sing!" and "He can
sing?" correspond to the same underlying syntagm "he -* (can
-- sing)" formed into two different para-syntactic units:

(I) "he - (can -- sing) + auxiliary-subject order"

(2) "he - (can -*sing) + subject-auxiliary order"

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
 x2 S. F. Gardner and S. G. J. Hervey

where "auxiliary-subject order" and "subject-auxiliary order"

functionally opposed para-syntactic features of the appropr
sentence-bases.

As for a-syntactic sentence-bases, these can be exemplified in

sentences such as "Yes!", "Please.", "Eh?", "Hello.", "How do
you do.", "Hmm?", "Shh!", etc. In all these cases-though in
some more obviously than in others-we are dealing with base
entities ("yes", "please", "eh" etc.) that, if they were to be
accorded a syntactic status at all, could only be identified as
mono-monematic pleremes. (Note that even "how do you do",
when used as a greeting, is an unanalysable "fossil".6 They
cannot, however, be identified as pleremes without contradiction,
in that they do not answer the definition of "minimum syntactic
entity";7 such entities are not capable of standing in ordering
relations with other entities in grammar. The only remaining
alternative, then, seems to be to recognise these entities as potential
sentence-bases without according them the status of syntactic
entities. The best we can do under the circumstances is to refer
to them as "quasi-syntactic entities" in token of their marginal
similarity to genuine syntactic entities.
In the context of simple sentence-bases, we can further develop
our typology by sub-classifying syntactic sentence-bases according
to the type of syntactic entity they correspond to. Initially we note
from syntax8 that there are only two theoretically distinct types
of syntactic entity: pleremes which are the minimum entities capable
of standing in ordering relations in grammar, and syntagms which
are self-contained complexes of pleremes and/or syntagms in
constructional relations with one another. Sentence-bases may
correspond directly to either of these types of syntactic entity.
Thus, for instance, in the case of the sentence "Good!", the
sentence-base corresponds to a single plereme (which is elsewhere
found in adjectival position, e.g. in "a good girl", or in the
complement position of a "copulative predicative", e.g. in "she
is very good"). On the other hand, the sentence-base of "He

6. Sa'ndor G. J. HERVEY and Jan W. F. MULDER, Pseudo-composites and pseudo-

words, La Linguistique, 9, 1973; also in MULDER and HERVEY, The strategy of
linguistics.
7. MULDER and HERVEY, The strategy of linguistics, p. 47.
8. Ibid., p. 149.

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
Structural sentence types '3

drives to work every day" is the "exten

syntagm"9

(he
(he-- drives) [to * work]
drives) [every -* day]

Although pleremes and (self-contained)

two types of entity recognised as genui
syntactic theory, from the point of view
reckon with two further types of "syntact
in sentences such as "Tom, the piper's so
lying sentence-base corresponds to an entity
properties of a syntagm: i.e. both "Tom
piper's son" contract normal "subject-pre
"ran away" in "Tom ran away" and "the
and compete for "subject position" in "T
away". Yet, the relation between these two
cannot be identified as a genuine syntactic
could stand in the subject position of
together cannot. Mulder calls the relation b
constituents a "quasi-syntactic relation"
sition".10 Thus, by virtue of containing
tactic" (actually non-constructional) rela
son-ran away" is shown to be an entity t
cohesion of a normal syntagm. We shall tre
tactic entity, and refer to it as an "(appo
While an "(appositional) quasi-syntag
superfluity of constituents (it is this that p
normal syntagm), there are also sentenc
to "syntactic entities" that appear to
essential) constituents. For instance, in
girl!", "Read this book!", etc., the senten
to correspond to complex syntactic entit

+- (this -- book)"). Although these are

and normal syntactic relations do, indee
constituents, they cannot be identified w
syntagms, for they do not have the req

9. For representational conventions see ibid., p. 152

io. Ibid., p. 49.

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
14 S. F. Gardner and S. G. J. Hervey

containedness-they are, in other words, "incomplet

of well-formed (and self-contained) syntagms. The
-+ girl" is not ill-formed (a normal adjective-nomi
obtains between its constituents), but is a "truncate
a "nominal syntagm (sing.)" in which either article
position is, for syntactic purposes, obligatorily filled. A
+- (this -* book)", this bundle cannot be considered
fledged "predicative-based" syntagm, since a "predic
syntagm is, for syntactic purposes, not self-contained
without subject position being filled. On the other han
is not syntactically ill-formed, in the sense that nor
relations obtain between its constituents. It is recognisa
as a "truncated" version of a "predicative-based
Although syntax accords no status to such "truncate
from the point of view of sentences these behave n
from fully-fledged syntactic entities. We shall treat
marginal syntactic entities, and refer to them as "
syntagms".
There are three points worth making here on the subject of
quasi-syntagms and incomplete syntagms before we summarise
and exemplify the categories of simple sentence-base. First of all,
we want to make it clear that we are not treating incomplete
syntagms such as "read this book" as necessarily elliptical; this is,
syntactically, merely an incomplete syntagm, which would be
"complete" or self-contained if the subject position were filled.
Although we may be tempted from a statistical point of view to
claim that "you" is actually understood as the subject, the following
example is sufficient to demonstrate that this is not theoretically
sound: "Read this book?" is not an imperative, and does not
contain an elliptical "you".
Secondly, from the fact that quasi-syntagms are, in a sense,
syntagms with "extra" constituents and incomplete syntagms are
syntagms lacking in one or more constituents, it would seem to be
the case that a sentence-base could not be both a quasi-syntagm
and an incomplete syntagm i.e. could not both contain an extra
constituent and be missing a constituent. On further consideration,
of course, one realises that there is no reason why a simple sentence-
base should not correspond to an incomplete syntagm with a
quasi-syntactic relation between two of its parts (e.g. "Been
here-in this room-long?"). It follows, therefore, that there are

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
 Structural sentence types 15

in fact three types of marginally synt

recognised as self-contained syntactic
syntagms, incomplete syntagms, and
incomplete. For the purposes of a sim
three shall be grouped together and r
syntagms".
The very existence of these marginal syntagms, of entities
which are not fully-fledged syntagms because they are not well-
formed and/or not self-contained in syntax, but which correspond
to the bases of perfectly well-formed and self-contained sentences,
provides us with further evidence in support of our hypothesis
that languages can be most adequately described if the syntax is
treated separately from the sentence-forming aspects of language.
From the sentential point of view, therefore, we can recognise
three types of syntactic sentence-base, as shown in the following
diagram which summarises the categories of simple sentence-
base:

simple sentence-bases

syntactic para-syntactic a-syntactic

sentence-bases sentence-bases sentence-bases

sentence-bases sentence-bases sentence-bases

corresponding corresponding corresponding
to single to single to single
pleremes syntagms marginal syntagms

FIo. I

The above diagram presents theoretical types of simple sen-

tence-base, not all of which will be necessarily encountered in all
natural languages (though it is unlikely that there should be a
natural language that does not make use of each of them). Over

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
16 S. F. Gardner and S. G. J. Hervey

and above these theoretically valid types, linguistic

can establish further and more detailed classificatio
sentence-bases (and, therefore, of the sentences con
according to language-specific categories of plerem
para-syntactic unit and quasi-syntactic entity. Thus
one may envisage, in the description of, say, English
language-specific categories of simple sentence-bas
(e.g. "run" as the base of "Run!"), "nominal syn
(e.g. "the house in the woods" as the base of, say,
the woods?") and so forth. At the same time, linguistic
may reveal important categories of grammatica
cannot, in themselves, correspond to simple sente
instance, the pleremes "a" and "the" in English). Th
are language-specific matters and we shall not conc
with them any further here.
The diagram above (fig. I) represents, of course, o
of the typology of sentence-bases; it still remains for
a typology of compound sentence-bases. As we have sta
compound sentence-base corresponds to two or mor
base-units-these base-units being the very types o
correspond to simple sentence-bases. That is to say,
"parts" of a compound sentence-base fall into the sa
of plereme, syntagm, marginal syntagm, para-synt
quasi-syntactic entity. We use the term "juxtaposed
the "parts" of a compound sentence-base are not i
conceived of as being in constructional relation
another-only the eventual grafting of sentential f
them makes compound sentence-bases "units".
The obvious way to classify compound sentence-
looking at the status of the juxtaposed "parts" fou
Since, theoretically at least, compound sentence-bas
an indefinite number of such "parts"-and each of
may correspond to any of the simple sentence-b
exhaustive listing of types is not merely cumbers
principle impossible. We shall, therefore, restrict
illustrating the typology of compound sentence-b
centrating on computing the logical possibilities th
compound sentence-bases consisting of two juxtapos
can be done in the following matrix, in which
correspond to the examples listed further below:

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
 Structural sentence types 17

One of the "parts" of the "two-part" sentence-base correspond

UU

C,,

One of the Plereme I 2 4 7 II

"parts" of the
Syntagm 2 3 5 8 12
"two-part"
sentence-base Marginal
corresponds syntagm 4 5 6 9 13
to a
Para-syntactic
unit 7 8 9 Io 14
Quasi-syntactic
entity 11 12 13 14 15

FIG. 2

(Notice that, due to the fact that the parts of the base are merely juxtaposed, i.e. there
are no functional ordering relations between the parts of the base, "Look, a plane!"
and "A plane, look!" both have the same sentence-base consisting in the juxtaposition of
the same plereme and syntagm, and are an example of type 2 on the above table.)

The numbers (1-15) designating the examples listed below

correspond to the numbers in fig. 2, i.e. to the different structural
sentence-types found in the matrix. The sub-divisions (a, b, c)
refer to subtypes of sentence differing with respect to the subtypes
of marginal syntagm (quasi-syntagms and incomplete syntagms).

x. "Look, Jane!" with the sentence-base: "look / Jane"

2. "Good, he smokes." with the sentence-base: "good / he
-- smokes"
3. "The butler killed him, my dear Watson." with the sentence-

base: "(the -- butler) - killed - him / my -Watson"

4 a. "Doctor, my husband, Sam, looks ill." with the sentence-

base: "doctor / (my -+ husband - Sam) -> (looks <- ill)"
4 b. "Strange, can't see it." with the sentence-base: "strange /

B -+ can-+ see <-it"

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
 18 S. F. Gardner and S. G. J. Hervey

5 a. "That book, the red one, got lost, John told

the sentence-base: "that - book- red -one

<- lost) / John - told +- me"
5 b. "I hate you, go away!" with the sentence-base: "
+- you / u - go -away"
6 a. "You send the army - our young men - to
fight - they kill - for us." with the sentence-ba
our

-+ send +- (the -+ army - yo

(they - fight - they kill) <- (fo
6 b. "I like him, the old boy - who
thel
tence-base: "I -+ like +- (him - old j- boy) / who
(would old
not__

6 c. "Go west, young man!'.' with the sentence-base: "

-9 go *- west art.
I1/ tum.0
y ;5 man"
young

7. "Don't you smoke, John?" with the sentence-base: "you

not-+ smoke + auxiliary-subject order /IJohn"

8. "Others don't swim - they sink." with the sentence-base:

"others ->not -*+ swim) + subject-auxiliary order / they

-+ sink" not
9 a. "You, my best friend, betrayed me, didn't you?" with the

sentence-base: "(you - MeY friend) - betrayed - me/

you -- d not]- auxiliary-subject order"

9 b. "Plants, let me remind you, are like people." with the
sentence-base: "plants -- (are - (like - people)) + subject-
auxiliary order / A -- (let - me) +- (remind - you)"
io. "Do you like him, may I ask?" with the sentence-base:
"you -+ (do -- like) +- him + auxiliary-subject order I I
-- (may -+ ask) + auxiliary-subject order"

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms
 Structural sentence types 19

S1. "Oh, darling!" with the sentence-b

12. "Hello, little girls." with the senten
- girls"
13 a. "Yes, I read the book - the revised edition." with the
the

sentence-base: "yes /I --> read <- (the -> book revised "revised
-* edition)
13 b. "Please, read this." with the sentence-base: "please /
F -+ read <- this"
14. "Well, today is the deadline." with the sentence-base:
"well / today -+ (is <- (the -+ deadline)) + subject-auxiliary
order"

15. "Oh, gosh!" with the sentence-base: "oh / gosh"

On the basis of the above classification and exemplification

of "two-part" sentence-bases any given sentence-base, of whatever
complexity, may be broken down into its "parts". Thus, for
instance, the sentence "Ready, steady, go!" can be seen to have a
sentence-base in three "parts", each of them corresponding to a
single plereme.
In conclusion, we should like to call attention to the fact that
our present study-albeit preliminary and subject to alteration
in its details-does clearly indicate that a structural treatment of
sentences and sentence-types is not only possible, but a matter of
the greatest relevance, within the framework of functional linguis-
tics. In this respect, structural-functional linguistics has an
important contribution to make to the currently fashionable
field of "pragmatics".

This content downloaded from

193.194.76.5 on Mon, 03 Mar 2025 09:43:43 UTC
All use subject to https://about.jstor.org/terms