Ground
Rules For Text Linguistics
Robert
de Beaugrande
In Memory of Peter Hartmann
Its True Father and Founder
— Jimmy Carter
1.
The rationale for text linguistics
1.1
It is now forty years since the publication of Peter Hartmann’s (1963a, 1963b)
monumental works Theorie der Grammatik1 and Theorie der
Sprachwissenschaft. They
were a quiet landmarks of a rare and visionary kind. Their majestically abstract
and theoretical nature and their intricately textured style have kept them from
attaining the wide readership in linguistics that they merited. Yet they
anticipated, elaborated, and justified the developments that led to the
linguistics of text and discourse. I at least learned from them how to grasp
established ‘mainstream linguistics’ as just one possible ‘science of
language’, and how to think my way forward into other less well-mapped
territories (cf. Beaugrande 1997, 2004)
1.2
Hartmann’s influence was naturally most direct on those inaugural milestones
in text linguistics contributed by his own pupils, such as Harweg (1968), Koch
(1971), Wienold (1972), and Schmidt (1973), though the family resemblances among
these works, or between these works and Hartmann’s inspiring volumes, may seem
less pronounced than their diversity.2 Moreover, many works with
similar concerns come from sources which, as far as I can discover, betray no
direct or overt influence of Hartmann either as a teacher or as an author whose
‘theory of grammar’ was in effect a meta-theory
of theories of linguistic form: its
scope encompasses the formalization of any ‘theory of grammar’,
whatever its historical and individual inception.
1.3
The ‘family resemblance’ among diverse works in text linguistics lies
chiefly in their search for broad and high-level perspectives on the formation
and investigation of texts, witness these passages from influential researchers
during that emancipatory stage:3
[1]
Structural linguistics did us the service of showing the error in trying to
build a grammar by going from the smallest units to bigger and bigger ones. We
must start with the biggest units, and we recognize the smaller ones only from
the structure of the whole, [of] texts. (Weinrich 1964: 109f)
[2]
Linguistics needs to clarify how language can be extracted from its social
manifestations enough to be a delimitable and abstract object of inquiry. [...]
A consciously phenomenological step in constructing linguistic theories can
force us to make provisional, but extremely important decisions about
fundamentals. (Schmidt 1973: 13f)
[3]
The fact that a natural language forms a complexly structured organic system and
links together the most diverse combinations of manifestations is the best proof
that an integrated theory for the most diverse phenomena must be possible. [...]
Utterances with the status of texts or discourses are the central linguistic
units to be investigated with such a theory, because only in discourse do all
manifestations of a communicative action appear; and all speakers of a language
also know the conventions of discourse. (Petöfi 1978: 7, 9)
[4]
We will be concerned with a linguistic approach in which grammar and text
structures can equally be described in a subsuming, shared model. [...] The
‘valence’ of the formalization of structures results from the explicit
arrangement of the domains in the model. (Koch 1971: 16f)
[5]
If the use of syntax in language is a domain of combining in general, then the
specifically linguistic aspect comes from the elements (language signs) and
motives (speaking, gathering information). [...] The procedures themselves are
much more universally distributed and may ultimately offer a foundation for all
sorts of combinational acts and products. (Hartmann 1964a: 27)
[6]
Text grammars [...] have to provide an explicit basis for the study of all types
of texts as they manifest themselves in processes of verbal communication [...]
Text grammar gives a more adequate account of the systematic phenomena of
natural language: properties of linguistic objects, i.e., structures underlying
utterances, as well as aspects of linguistic communication, e.g.. psychological
and psycho-social settings of systems and manifestations of natural language. (van
Dijk 1972: 1f).
[7]
If two text grammars are said to obey the same theoretical postulates and to
cover the same domain of application, it may be inferred that the text grammars
are equivalent. [...] The test of whether two text grammars are equivalent
relative to a given domain of application may serve as an evaluative procedure,
[...] not only of considerable theoretical interest but also of practical
importance to those linguists who want to apply the findings of text linguistics
to the solution of practical problems. (Rieser 1973: 276f).
[8]
When uttered in standard situations, every well-formed expression points to
particular states of affairs whose existence is independent of the success of
the communicative action deploying the expression, [but] on the other hand it
signals what states of affairs
must obtain for the action to succeed. [...]
A system of rules to
account for the competencies of the language user must assign to linguistic
expressions not only representations of their semantic content, but also
representations of the pragmatic functions of that content (Posner 1972: 15,
18).
[9] A text is a linguistic sign, i.e., an assignment (performed according to
a plan of action with rules for composing texts and developing topics) of
conceptual content as a reflection of states of affairs and manifestations in
reality, to a sequence of sounds or graphemes. Every text is therefore the
‘nomination’ of a particular event, process, state of affairs, condition, or
situation. [...] But since language activity is productive, creative, and
directed to social goals, every text has a communicative function as well as a
‘nominative’ one. (Viehweger 1976: 197)
[10]
The emphasis falls on the
categories of human praxis as conditions for the origin of language. [...] The reconstruction of language origin is necessarily abstract,
but offers the basis for the concrete analysis of the forms of praxis and
communication. (Kummer 1975: 9)
[11]
Texts are consumed by communicative participants. A precondition for this consumption is that
the participants share a common potential for structuration based on a common
language. [...] The esoteric nature of some texts, artefacts, or actions (e.g.
initiation rites) can be measured and determined via their communicative form
compared to other related forms and a presupposed normal form. (Wienold 1971:
9f).
[12]
In view of the fact that language communication
always takes place in the form of texts, domains such as native and foreign
language instruction, translation, and so on, increasingly demand a
text-theoretical foundation. The creation of a linguistic text theory is thus
urgent not only for linguistics, but for society at large
(Isenberg 1976: 47).
1.4
One lesson then would have been that theories of language are necessarily (if
only implicitly) theories of text and discourse, and we may profit by rendering
this factor explicit. The theory can then be one example of its own object
domain, namely: how linguistic knowledge (rules, forms, structures, etc.) is
derived from language manifestations and communicated in language. To manage so
complex a domain of inquiry, we need a unified theory and terminology on all levels,
with due provisions for specification as appropriate to the domain at hand.
1.5
We can start off with a general working definition of a theory as a predisposition
toward making certain uses of certain classes of evidence; no theory can be
fully disinterested or objective, insofar as it influences the conditions of
discovery and interpretation (cf. Kuhn 1970; Stegmüller 1976). However, this
epistemological necessity does not imply that all theories are equally circular
and equally valid or invalid. It implies that the validation of a theory should
be undertaken not just internally, i.e., by testing whether its
causalities and explanations are consistent within
the domain it
addresses; but also externally, i.e., by testing
whether its causalities and explanations are consistent with what is known about
human contexts.
1.6
In the field of psychology, for example, Ulrich Neisser (1982: 44) observes that
such theories as classical psychoanalysis and behaviorism are self-contradictory
in respect to their accounts of human nature: ‘If all ideas are
wish-fulfillment, then psychoanalytic ideas just express the wishes of the
psychoanalyst. If every belief just results from conditioning, so does the
behaviorist’s belief that this is the case’. These theories imply that
humans, driven by internal wishes or external stimuli, are not well equipped to
construct valid theories. Cognitive psychology, in contrast, is concerned with
‘how knowledge is possible’, and thus can try to account for its own
theory-building as one domain of cognition.
1.7 In the field of linguistics, the statement of theories and evidence is undeniably one type of linguistic act. As such, linguistics can be a proper domain for linguistic inquiry (Beaugrande 1991). In normal practice, a statement made by linguists is a direct or indirect consequence of their own experience with language events. This experience has led to the gradual (though not necessarily conscious) development of general hypotheses about what language is, and what it is for. These hypotheses in turn guide the linguist in enacting what Hartmann (1963a) described as the reduction, isolation, and abstraction of language entities in search of what is typical, general, and repeatable. The resulting object — a ‘linguistic theory’ — has the empirical status of a derived effect, that is, of a second-order phenomenon produced via ‘theorization’ from the first-order phenomenon of language experience. The question of how that effect was derived is far from trivial, given the vast range of possible methods for constructing a theory of language.
1.8
The crucial issue is what gets removed and what gets retained from the totality
of available experiences, i.e., what aspects of real contexts are considered
relevant or irrelevant for analysing, describing, and modelling language.
Knowing an element or a rule of language means knowing how and when to use it
(cf. Bolinger 1968; Labov 1970; Yngve 1975; McCawley 1976). Sometimes, the
element or rule
may not establish its identity until the moment of its use; if so, removal of context
entails a removal of essence. Although, at a fine degree of detail, every
context is unique — in respect to time, location, participants, motor actions,
choice and production of words and phrases, etc. — most contexts share some
features or aspects that identify and define the communicative community. This
shared substrate must have a determinable structure in order that people can use
language reliably amid the vast variety of circumstances that can or do occur.
However, we have no motive to assume in advance that this structure can best be
defined or described in the categories of one particular linguistic theory. We
need to know more about how those categories are created from the totality of
aspects involved in linguistic form
and formation (cf. section 2).
1.9
A linguistic theory can be compared to a map with an inner domain of
theoretical exploration surrounded by a much larger outer domain of
empirical contexts. The question of how the relationship between the two domains
should be regulated remains open for each new theory to determine. If the outer
domain is ignored, the theory can be consistent and reliable in its scientific
context only as far as the investigators
themselves participate in a common culture and work with reasonably uniform presuppositions and interpretations. The
validity of such a restricted
theory might be established after the fact by showing that the outer domain,
though seldom explored, is at least stable among the community of scientists.
Conversely, the theory might be discredited if no such uniformity emerged. For
example, a theory of linguistic form in isolation could be supported by
discovering that the community of linguists were following unified procedures
for extracting form. The opposite conclusion would be drawn if great diversity
were found: the theory would not be useful because the derived effect would be a
different object for each school of investigators (or in the worst case, for
each investigator): generalizations would be illusory or metaphorical; claims
would have a different meaning for each set of evidence; and so on.
1.12
To be sure, validation after the fact can be an arduous and risky
procedure. It would be safer to establish validity in wider contexts before
the fact. We might work out
at least a provisional map of the outer domain, showing some of the more
prominent features that affect the structure of the inner domain — of the
‘linguistic theory’ at hand. The theory could then be extended to encompass
relevant contextual factors as they become tractable. For example, a theory that
treats language as one mode of human action and interaction (cf. sample [9])
could be strategically situated in the outer domain of a general theory
of action. General concepts of action would be specified for the linguistic
concepts of planning, uttering, asserting, justifying, refuting, informing, and
so on (cf. Beaugrande 1979, 1980a, 1980b, 1984a). This procedure would allow
more powerful explanations wherever discourse is found to be determined by the
structure of interaction, e.g., the dependency of conversational utterances on
the success of personal goals (cf. samples [1] and [8]). If neighbouring
disciplines such as sociology and anthropology also agree to derive their
central concepts by specifying the concepts of action theory, hypotheses and
findings could more easily be traded from one discipline to the next. Thus, even
though contexts would still be limited and selective, the relation between the
inner and the outer domain would not be fragmented or obscure.
1.11
Most of the basic studies of text and discourse have precisely this kind of goal
(cf. surveys in Kallmeyer et al. 1974; Grimes 1975; Coulthard 1977; Grimes ed.
1978; Beaugrande 1980c; Beaugrande & Dressler
1981; Kalverkämper 1981; van Dijk ed. 1984). They assume that the organization
of a text or of a discourse (as a set of mutually relevant texts) is determined
in part by the abstract language system, and in part by the context of
communication. Whereas an abstract analysis could uncover a vast total of
linguistic forms and features, the human processing of discourse probably
focuses on that strategic subset that supplies relevant input or output in a
given communication. The fact that everyday discourse may seem structurally
fuzzy or incomplete when viewed in formal terms alone is offset by using
indicators on other levels, such as knowledge of topics and human goals. A
suitable methodology of discourse analysis should thus help us identify
and incorporate contextual factors that
influence the formal properties
of language samples.
1.12
To guide such an enterprise, explicit ground rules for linguistic inquiry could
be set down, such as:
1.12.1
The investigator defines the basic units as actions, and the domain as a domain
of action.
1.12.2
Linguistic items, features, or forms are interpreted and classified according to
their relevance for the performance of actions.
1.12.3
All structured entities (phrases, sentences, texts. etc.) are studied in terms
of how they are created and used in communicative events. Prospective
‘rules’ should be stated as procedures for using them.
1.12.4
The investigator openly states how and why the inquiry is selective and limited,
and what interests and motives the research is serving.
1.12.5
No single set of limitations and selections can be definitive; other needs can
establish other sets.
1.12.6
Evidence is not invented by the investigator, but gathered from empirically real
discourse.
1.12.7
Interdisciplinarity is recognized as essential for the wider validity of
particular findings.
1.13
In such an inquiry, any theory would be expressly identified as a set of
decisions about how to discover and use evidence (cf. 1.4). Scientific objects
would be openly acknowledged as a derived effect of the motives and
preconditions for linguistic inquiry. And the processes whereby that effect is
derived would be a proper part of the inquiry. Each particular study of
discourse would be both more and less than a description of language as an
abstract system: more because the constitution of the discourse is based on
other factors besides the abstract system; and less because not all discoverable
features in the sample discourse will be equally relevant for its communicative
setting.
1.12
The condition of domain relevance guides and restricts the accumulation
of linguistic data that, under a more abstract perspective, might all look
uniform. Empirical evidence from spontaneous activities would be used to not
only to test individual hypotheses, but also to delimit the extent and
conditions of various domains of discourse, so that further inquiries can be
appropriately designed. The discovery of forms, features, and structures would
be controlled by clearly stated epistemological interests and predispositions
regarding relevant modes of evidence. The
question is then no longer ‘what is the form of language?’, but
rather ‘what language events or
aspects of those events lead people (including linguists) to infer
or enact a form?’
1.13
For linguistic research, the question is then: ‘what context does the
investigator4 create when citing a linguistic example in a
discussion?’ The ‘null hypothesis’ would be: no context: the
example is and remains what it is (a word, sentence, text, etc;) even if
completely removed from all real uses. But this hypothesis is clearly
inadequate, because a linguistic discussion is itself a context influenced by
the investigator’s knowledge, experience, and goals. And the example is
intended (and often invented) to ‘show’ something, to justify or refute some
assertion (cf. section 3).
1.14
The next hypothesis would be: investigators create a
uniform minimal context
for their examples. The
contribution of context would be as small as possible, and would be the same for
all investigators. This hypothesis also seems vulnerable in view of the fact that linguists can disagree so
dramatically about what a given
example ‘is’ or ‘shows’, e.g. the notorious ‘Colorless green ideas
sleep furiously’ made popular by Chomsky (1957). At least some divergence in
context must be a factor in these disagreements.
1.15
After uniform minimal contexts, the next hypothesis would be: investigators
create a limited range of context types. Accepting a given theory would
entail a disposition to construct contexts of a corresponding type (cf. 1.4).
Disagreements would then be due to divergences in type, but the range of
contexts would still be constrained and orderly. Results would be reasonably,
though not totally, reliable within each theory. The problem would rather be how
to integrate findings from dissimilar theories into a common body of linguistic
research. We would need to know which factors decide the type of context at
stake, e.g., the source and kind of examples, the investigator’s knowledge of
the language, the goal of the discussion, and so on.
1.16
The least constraining hypothesis would be: investigators create an uncontrolled
range of individual contexts. Every example might be contextualised as
befits the spur of the moment, and no common procedures might be applied from
one example to the next. Research could not be cumulative, except in the vague
sense that all examples from one language are related within the total language
system; and the ‘ungrammatical’ examples traditionally marked with asterisks
would not even share that commonality, since they are by definition excluded
from the system.
1.17
Determining which of these hypotheses holds is urgent, since the validity,
generality, and cumulativity of the whole enterprise of linguistics hang in the
balance. We must explore how the text-fragment ‘linguistic example’ is
constituted within the text type ‘linguistic investigation’. Surely a
perspicacious method of text studies ought to offer key assistance in clarifying
this issue. Let’s examine what Hartmann’s accounts (via my translations)
might suggest.
2. Hartmann’s theory of linguistic form
2.1
In pursuing his ‘theory of grammar’, Hartmann’s (1963a) first concern was:
in what sense is it true that ‘language is a form’, or that ‘language has
form’? Only by facing these questions can we determine how linguistic
investigators get form out of examples, and eventually construct a ‘grammar’
as a system of linguistic forms. As his deliberations reveal, the origin and
nature of linguistic form are far more complex than is usually made explicit in
language theories, no doubt because the investigators, being themselves
participants in the constitution and use of form, leave a considerable portion
of the domain out of view (cf. 2.7). Admittedly, facing the whole domain is
hardly an inviting enterprise, but in Hartmann’s view (and in my own), an
ultimately inescapable one.
2.2
Conventional linguistic theories result from one mode of ‘formation’ in a
sense that will delimited below (2.6.6): from a process of postulating
structured substrates of linguistic form. The Neo-Grammarians
(Junggrammatiker) of the 19th century looked for ‘laws’ governing the
historical evolution of sounds, words, and grammatical paradigms. The
Structuralists of the 20th century treated language as an abstract system of
relationships (especially equivalences and oppositions) on distinctive
‘levels’. The
Generativists argued for a set of ‘rules’ that describe the
grammaticality of sentences. In each case, the seeming diversity of language
examples was counter-balanced by a continuing emphasis on limited, unifying
formal substrates. Yet in each case too, a different substrate was envisioned,
depending on the investigators’ specialized capacities for extracting forms.
2.3
If the ‘formation’ procedures themselves are not recognized as part of
linguistic theorizing, the outcome is predictably just what we see in the
development of linguistics. Priority is given to the aspects of language where
‘form’ is considered the easiest to constitute and agree about. Then, it
appears safe to assume that form is ‘already there’ and the investigator has
contributed nothing to it. By focusing attention on cataloguing sounds and forms
in the compact, orderly taxonomies of phonology and morphology, linguistics
postponed the elucidation of its own cognitive foundations. Phonemic features
are associated with characteristic articulatory events (e.g., ‘stops’) or
with the physiological locations of those events (e.g., ‘dental’). Morphemes
are linked to a long-standing tradition of listing grammatical ‘paradigms’
(e.g. ‘conjugations’). The discussion shifted dramatically after the
ascension of ‘transformational grammar’ as an enterprise with much wider
scope of formalization. Unfortunately, however, those new grammarians were not
too clear or explicit about the consequences of their own formalizing activities
vis-à-vis the empirical domain of language as a complex object. They postulated
a ‘syntactic deep structure’ as the reciprocal input and output of
‘linguistic rules’ (Chomsky 1965). This construct enabled a direct and
stringent means for relating theoretical form to surface form, since both levels
were defined with the same set of categories. But it was far from obvious that
equally direct means would be able to relate these two levels of form to the content
of languages, e.g., to the potential of language elements to designate things
and facts. The consequence has been to seal off one kind of linguistic form, the
grammatical forms of sentences, from other orders of form; or else to force upon
those orders the precise categories of the original one. Without a map of our
own cognitive foundations, we are not well equipped to search for a less
restrictive theory, or to determine how the ‘standard theory’ or its
revisions (e.g. case grammar, generative semantics, trace theory) relate to the
total picture of accounting for linguistic form — and how far we can preserve
or restate their significant results.
2.4
We ought to expect some difficulty in filtering out precisely the ‘form’ of
language from its ‘content’ (‘meaning’, ‘sense’ etc.) and its
‘substance’ (‘sounds’, ‘letters’, etc.). In empirical reality,
language forms don’t occur without expressing meaning, or without being
manifested in some phonetic or graphic events. Hence, we can only define
‘form’ as an aspect of more complex phenomena that include it in various
ways. By ‘formalizing’ language, we alter it as a phenomenon (Hartmann
1963a: 42ff): we subtract and reduce until it appears as if form had an
existence of its own. Yet if form is not a thing, but at most a relation of
a thing, form remains latent until we make it an object of scrutiny. The
‘formalized’ view of’ language can only be a complement to a view
accounting for content; and we risk falsification if we do not keep this
limitation in mind (1963a: 46). As an activity, formalization is neutral until
it loses sight of its own conditions and contexts; thereafter, it can become
distortive.
2.5
Throughout his work, Hartmann (1963a, 1963b, 1964a, 1964b, 1971, 1975) realized
that the ‘text’ is not just a ‘rank above the sentence’ or a ‘sequence
of sentences’: it is the product and attestation of a linguistic action. As
such, the text is the primary, though often unacknowledged, vehicle of
linguistic inquiry (1.4). When linguists examine a sample of language, the
status of the sample as evidence is directly proportional to its status as a
real or possible text (or stretch of text). Usually, two orders of ‘form’
are extracted from texts: (1) forms that are present as types we can assemble in
a grammar, allowing us to distinguish language types as types of formative
systems; and (2) forms that are present by virtue of their distribution among
sentence constituents, allowing us to distinguish syntactic patterns as formal
combinational types available in the individual language (Hartmann
1963a: 34). Hence, linguistics
has dealt mainly with two orders of form: grammatical
forms (shaping and structuring forms of language as a technique); and forms for
organizing sentences and determining senses. But since both orders of form, and
perhaps others as well, are present in every text, we need a more general and higher-level
theory of linguistic form than just one of these orders can engender.
Hartmann explored possible statements that may be true of ‘form’ on all
levels of abstraction, e.g.:
2.5.1
The essence of form is to be or document a relation (1963a: 49).
2.5.2
The content of forms is their designation of the categorical (1963a: 15).
2.5.3
Forms symbolize the procedures that create them in use (1963a: 18).
2.5.4
Formalization is a typifying removal of the repeatable (1963a: 36).
2.5.5
Forms represent relations, language documents relations, and human reason
constitutes relations, such that language is a documentation of reason in the
activity of constituting relations; it is typical and thematic of language that
it contains formations and can thus be called a form itself (1963a: 9, 53).
2.6
On the highest theoretical level, ‘form’ is found in the relations
that must obtain in order for
language to be language at all. Hartmann (1963a: 55-97) works through six
progressive orders of ‘phenomena’ tied to language, each of them giving a
certain picture of ‘form’ as an abstract relation:5
2.6.1
The VALIDITY (‘Geltung’) of a language element is its ability to be a valid
symbol for more than one thing6 and to be recognized as such by more
than one person (1963a: 55ff). Assigning a sign to a sense is not an automatic
result of natural laws, but an intentional
procedure. Designating is actually two acts of assignment at once: (1) the already
learned and accepted act of assigning a spoken or written sign to a concept, as
performed by the language community; and (2) the speaker’s (or writer’s)
current act of applying the word to something. In this basic aspect, ‘form’
is the bond between two entities from disparate domains, such as sound and
concept, of which only the first is manifestly perceptible.
2.6.2
The AS-STRUCTURE (‘Als-Struktur’)
results from assigning a name to a thing such that the thing then adds the character of being known ‘as’
something, or as one of a type (1963a: 60ff). Here too, ‘form’ is the bond between two entities
(name and thing), but this time, the bond is interpretive as well as
intentional.
2.6.3
The ABSTRACTIVITY (‘Abstraktivität’) is the fact that a designation is more
‘abstract’ than any one thing or fact it designates (1963a: 72ff). The
‘plurality’ of things is made ‘abstract’ in the ‘unity’ of the
designation. The ‘form’ in ‘abstractivity’ adds a dimension of breadth
to the form in the ‘as-structure’, i.e., to the interpretive bond
between two entities. One pole of that bond is a complex unit composed and held
together by ‘abstraction’.
2.6.4
The USE (‘Verwendung’) of a language element can occur in a text as soon as the user identifies a fact as an applicable case (1963a: 77ff).
This act of identifying involves making use of — comparing and differentiating
— diverse features of ‘facts’, such as space, time, properties, and so on.
In this aspect, the ‘form’ is a bond, created via identification, between
the ‘type’ and the ‘fact’.
Now, both poles of the bond are complex units.
2.6.5
COMBINING (‘Kombinieren’) is the normal way of using language by putting two
or more elements together (1963a: 83ff). At this point, the ‘form’ becomes
more complex than the bond between two entities, the ‘form’ as noted in
2.6.1-4. In addition to designating something, combining designates something in
relation to something else; two or more types are made compatible and applied to
a thing or fact, so the bond has at least three poles, often more. Though the
content is enriched by the added relation, it is also restricted, since each
relation rules out some possibilities
the prior act of identification is correspondingly more complex than in simple ‘use’.
2.6.6
FORMATION (‘Formgebung’) is the total result of having created ‘form’ by
making something into something else (1963a: 93). Now, the ‘form’ of the act
itself is to create unity (the ‘form’) from the multiplicity of abstraction
(broadening the type), use (identifying and characterizing facts), and
combination (making the text).
2.7
In the normal practice of linguistics, these aspects of linguistic form are not
sharply differentiated, despite the shared preoccupation with formal issues.
Often, the presence of form in real or invented samples is taken for granted a
priori. In this way, the inquiry commences at an already advanced stage, namely
‘combination’, within Hartmann’s scheme of the development of form, and
leaves the more basic stages unexplored. Hartmann in contrast, wants to place
‘linguistic form’ in its proper context as an aspect and product of human
cognition, e.g., of the activities of the intellect involved in intending,
interpreting, identifying, classifying, and so on. He in fact suspends the old
‘form vs. content dichotomy’ by showing that form is also found in all
relations a language element can enter, including its relation to whatever it
can or does designate. This approach helps to integrate theories of language
form into more general theories of cognition
and communication.
3. Linguistic
citations in the discourse of semantics
3.1
In accordance with the ground rules proposed in 1.12, I should now state my own goals for the following demonstration. I am interested in selecting
and interpreting evidence about the discourse action of CITING, defined as
presenting a sample of another discourse as evidence for some claim advanced in
the current discourse. For motives I have set forth so far, my particular
concern is how a citing indicates form being extracted or abstracted from
context. Though my sample authors were presumably unaware of Hartmann’s
theoretical work, both were implicated in defining relations of form even when
they purported to be defining ‘meaning’ or ‘sense’. As we will see,
problems of both implicit and explicit nature arise because the ‘inner
domain’ addressed by each author is not clearly related to the ‘outer
domain’ of a general theory of linguistic formalizing (cf. 1.7).
3.2 Our sample texts appear to belong to ‘that segment of philosophical opinion known as analytical philosophy’, whose ‘hallmark’ is ‘to approach philosophical problems and theories in terms of the language used to formulate them’ (Caton 1971: 4). The ostensible project was to determine how a theory of ‘meaning’ or ‘sense’ should be constructed: whether we should follow the tradition of Aristotle, Frege (1892), and Carnap (1942), where ‘meaning’ is explained in terms of ‘truth’; or the ‘ordinary language’ philosophy in the tradition of Wittgenstein (1953, 1958), Ryle (1957), and Austin (1961, 1962), where ‘meaning’ is explained in terms of language use. Yet the actual project turns out to be essentially rationalising a treatment of form as meaning.
3.3 Our sampling contains three related texts: a long (20-page) exposition by David Wiggins (1971a), a shorter (12-page) comment by William Alston (1971), and a very short (1-page) rejoinder from Wiggins (1971b). To avoid continual repetition, I shall merely indicate the page number of my own citings.
3.4 The objective stated by Wiggins sounds a note of wistful resignation:
[13] I shall [...] make one final assault on the problem of sentence-sense, word-sense and difference of word-sense, from within the traditional theory of meaning — if only to commemorate the achievements (mainly Aristotle’s and Frege’s) of an activity perhaps even more certainly doomed to extinction than everything else which is familiar to us. (14)
Yet he foresees some hope in applying
[14] the existing assets of this branch of philosophy [...] in a renascent science of linguistics or in a new branch of model theory [by] get them redeployed usefully if they are identified and accurately accounted for. (14)
Thus, the current discourse is justified both as a retrospect (for ‘commemorating’) and as a prospect (for ‘redeploying’).
3.5 The framings indicate a certain tension between advancing large claims for the project versus portraying the author as modestly uncertain of his own authority to carry it out:
[15] so far as I am competent to judge — which I hardly am at all, but this is my tentative opinion (15)
[16] In spite of the inconclusiveness of some of the answers, [...] I now make a modest start (25)
[17] I welcome the opportunity to disown my ‘Identity-Statements’, which (amongst a hundred other faults) failed to look hard enough at the possibility... (17)
The author admits that he has a personal interest, but casts his role as one who merely clarifies:
[18] Of course these definitions have been offered in the pursuit of philosophical enterprises with which I am in general sympathy. For this very reason I am anxious to show that they go too far. (28)
The author’s ‘sympathy’ with ‘the traditional theory of meaning’ emerges more plainly in the opening of his reply to Alston, where the tone is far from one who believes that such activities are ‘doomed’ [13]:
[19] Nothing that has happened since J.L. Austin’s lectures [...] seems to me to have undermined the approach or made obsolete the kind of semantic theory typified by Frege and Russell. (48).
In fact, the author’s project isn’t too modest after all. He wants to appropriate for his favoured theory the entire field of semantics, indeed, to use his theory to define what semantics is. So he just says ‘nothing has happened’ rather than ‘I am aware of nothing that has happened’— and is obviously not aware of Hartmann’s far more deeply reflective explorations.
3.6 In addition to justifying the project, Wiggins self-consciously frames7 his own discourse:
[20] But before I can show anything about words, I must say something about the sense or meaning of sentences. (15)
[21] But before I can get to the philosophical theory of dictionaries, there are some problems to clear away. (17)
[22] We are not yet finished with ‘good’ and shall come back to it when I finally get to vindicate the claims I made at the beginning. (31)
Such
framing can combine with justifying, e.g., by suggesting that the author indeed
‘shows things about words’, ‘clears away the problems’, and
‘vindicates his claims’. The tone is at times petulant if not defensive:
[23] There is a clear point in attempting to make such distinctions as these. (20)
[24] These reflections [...] do however serve as a reminder that the difficulty is less a practical difficulty [...] than a theoretical difficulty about meaning as such. We understand fewer things than we need to understand. (18).
[25] The point may seem a trivial one. Perhaps I can make it more interesting by suggesting that it also solves one small problem about conditionals. (27)
[26] Even with so small a fund of information as we have about the theoretical requirements on a good dictionary, one or two further conclusions follow with something near certainty. (28)
These framings guide readers in the plan of the discourse, while asserting that the plan itself has a firm inner necessity.
3.7 The major assertings of the text accord with the intentions and ideas indicated in the framings. Wiggins purports to ‘identify and accurately account for’ certain philosophical ideas [14], transcending the limits whereby traditional views such as Frege’s are ‘almost always debated’ in a ‘straitened context’, and ‘the point’ has ‘got lost somewhere’ amid ‘half-hearted acceptance or inadequately founded rejection’ (15f.). Wiggins ‘commends an interpretation’ whereby Frege’s theory, seen ‘in its proper universality’, is ‘first and foremost a general theory of language’ (1971a: 16).
3.8 The most tangible goal for Wiggins’ project is to sort out ‘word-sense’ from ‘sentence-sense’ and to show that the classical approach can be used in a theory of the latter, rather than — as is often assumed — being limited to the former alone. In my terms, the old restricted theory of meaning (how words are defined) might endure as the ‘inner domain’ of semantics within the ‘outer domain’ of an emerging science of language (how sentences are understood [28]). But he would have to show that the vulnerable assumptions of the old theory were either not being made, or have been improperly interpreted by its adversaries —such as the ‘ordinary language philosophers’ cited in 3.2, with whom Alston was more ‘in general sympathy’.
3.9 The transition to ‘sentence-sense’ support Wiggins’ already noted expansion of Frege. He avers that the discussions of ‘reference’ are
[27] incompatible with Frege’s insistences (in the Grundlagen and elsewhere) that reference itself is unintelligible outside the context of a complete sentence or thought. [...] The theory’s foundation and basis is not naming at all but the notion of sentence-sense. (16)
Wiggins’ own ‘interpretation’, which
[28] makes sentence-sense, not reference, the point of leverage, explains the production and understanding of familiar and unfamiliar utterances by an account of how the constituents of sentences can systematically contribute to the meaning of the complete sentences within whose structure they figure. (16).
The recourse for knowing what ‘sentences mean’ turns out (no surprise) to be the classical one of ‘truth-conditions’, for which strong claims are asserted over competing theories:
[29] For the arbitrary sentence S, to know the meaning of S is to know under what conditions the sentence S would count as true. Unlike theories constructed in terms of belief and intention after the fashion of H. P. Grice, [...] or theories which place trusting confidence in a notion of use which they have still to delimit in a way that does not reimport all the problems, this theory of meaning offers us not a way into the circle of semantic terms but a connexion between two of them. This may be the best theory we shall have for some time. (17)
To implement this ‘theory’, Wiggins makes an ‘austere requirement’:
[30] The truth-condition must have been produced by the operation of a systematic, general, and uniform procedure competent to analyse any sentence in the language into semantic components drawn from a finite list of such components (i.e., a vocabulary or dictionary). (19)
This ‘procedure’ is evidently thought to function correctly automatically without any involvement on the part of the analyser.
3.10 Wiggins acknowledges that the traditional view of meaning as truth-conditions bas been attacked as ‘intolerably restricted’ (48). He cites ‘the standard objection’ but attributes it to no one in particular and so avoids taking on any respected authorities. The thrust of the ‘objection’ is stated as:
[31] Even in the case where the indicative reigns, the truth-condition account leaves out almost everything of some sentences’ communicative significance. (20)
His rebuttal — a common one in philosophical semantics — is to justify the limitations of the theory by limiting its domain until a fit is obtained, viz.:
[32] The total communicative content of an utterance, construing ‘content’ comprehensively, is too complex to be accounted for by any one-level theory. [...] We must rather unpack the speech-act layer by layer. And at the first layer we must, I think, try to isolate all and only what is strictly said — the one element with a claim to be really fundamental or central to linguistic communication — before we can explain how circumstances, conventions, and whatever else added by implicatures, forces, or illocutions; and how these in their turn secure perlocutionary effects. [...] From which we recover an intelligible and satisfactory-sounding doctrine — that truth must be the central notion of semantics, and that the boundary between what does and what does not bear logically on the truth of what is strictly said must be the boundary between the science of semantics and the science of the further effects obtaining in a speech-exchange. (20-21).
If we accept this peremptory and precarious drawing of ‘boundaries’, the traditional theory of meaning, as formulated by Aristotle, Frege, and Carnap, indeed remains the dominant (and ‘best’) theory of contemporary ‘semantics’. But I fail to see how such a ‘boundary’ could be sustained without considerable waffling and hand-waving, which is just what we get from Wiggins.
3.11 Intriguingly, he sees the truth-conditional approach focuses not as too narrow (the customary objection), but as too broad (20):
[33] This may create a much larger number of theoretical alternative readings of some sentences than it will normally occur to us to suppose they have. For some of the theoretically possible readings will be too absurd for them to occur to us as likely or credible. My own view, for what it is worth, would be that this exclusion and this absurdity almost always result from pragmatic factors, matters of fact, and contextual knowledge rather than from anything lying within the purview of semantics, for which all the absurd alternatives must count as theoretically possible. (25)
The ‘absurdity’ is thus handily shunted across the ‘boundary’ depicted in [32]
3.12 Yet questions persist of who should draw that boundary, and how. The implied answer can only be: a specialist who is properly equipped with the a ‘systematic, general, and uniform procedure’ plus a corresponding ‘dictionary’. Surely, to assume that ‘the production and understanding of utterances’ [28] entails such equipment would be the mother of all absurdities. However, I would see it as a typical of armchair philosophy dressed up as armchair psychology, which Hartmann staunchly rejected. The latest edition calls itself ‘relevance theory’, which, Deidre Wilson arrantly vowed in a conference talk in 1991, ‘tells you exactly how much mental processing is needed’. When I pointed to conclusive experimental evidence from real psychology that processing just doesn’t work that way, she paid no attention, as I should have expected.
3.13 At all events, Wiggins’ treatment of ‘sentence-sense’ (15-25) neither presents nor applies any such ‘procedure’ for ‘analysing sentences into semantic components’. Instead, he trots out the customary philosophers’ examples, such as (16-19,21-22):
[34] Snow is white.
[35] If anything is Hesperus [Evening Star], then Hesperus = Phosphorus [Morning Star].
[36] All mimsy were the borogroves. [US variant of Lewis Carroll’s original ‘borogoves’]
[37] Shut the door.
[38] I promise/I am promising.
[39] I love you/I am loving you.
The first [34] is presented as an ‘easy way of hitting the target’, i.e., ‘The True’, by saying ‘anything at all that is as a matter of fact true’ (16) — a shallow interpretation of Frege’s doctrine that Wiggins repudiates, particularly to escape the quandary of ‘leaving all sentences expressing logically or mathematically necessary truths with the same meaning’ (18). [35] is tendered as a possible ‘informative necessary statement empirically discovered’, rather than a paradoxical ‘identity statement’; here, Wiggins argues that ‘the senses of referring expressions’ are ‘simply special cases of senses in general’ (16-17), such that Frege can indeed be credited with ‘a general theory of language’ (cf. 3.7). [36] is cited to show that ‘the “truth” of the statement is not explained by making it satisfy the logical (“open”) reformulation, “If X is a borogrove, then X is mimsy”, because ‘this explanation’ ‘gives no idea at all of what investigations with what outcome would count for or against the assertion’; (19), i.e., that ‘verification’ is not to be handled this way. (No, you ask Humpty Dumpty, who will tell you that ‘“mimsy” is “flimsy and miserable”, and a “borogove” is a thin shabby-looking bird with its feathers sticking out all round — something like a live mop’, with good reason to feel mimsy. The remaining examples (29 -32) are all cited in regard to ‘performative’ issues in speech-act theory (e.g. commanding. promising, convincing), which Wiggins would consider ‘starkly extra-linguistic matters’ across the boundary from the ‘meaning’ of ‘what is strictly said’ (21) (cf. 3.10). Yet those very matters determine why the verbs in [38-39] very rarely occur in the progressive in my large data banks of authentic discourse (cf. 3.17)
3.14 Instead of ‘analysing sentences into components’, Wiggins takers to interpolating words into otherwise unanalysed sentences like [40-44] (26-29, 31, 33) in quest of some ‘minimal requirements for a dictionary’ (26). 3.21. In this way, emphasis gets shifted away from the form of sentences toward the form of dictionary entries.
[40] Kallias is snub-nosed.
[40a] Kallias is concave nosed-nosed.
[41] She had a charming nose, almost snub.
[42] The stone hit the horse’s fetlock.
[42a] The stone hit the horse’s horse’s ankle.
[43] She has good legs.
[44] This is a good knife.
Enlisting these sentences to explicate the distinction between ‘word-sense’ versus ‘sentence-sense’ represents a very reduced and sporadic approach to analysis, viz.:
[45] Consider the word ‘snub’ in ‘Kallias is snub-nosed’. If we wrote the entry for ‘snub’ as ‘(adjective) concave-nosed’ then it would look as if when we go back to our original sentence and write its truth condition it must be ‘Kallias is concave nosed-nosed’. Nor can the trouble be cured by defining ‘snub-nosed’ in one piece. It would not cope with ‘She had a charming nose, almost snub’. [...] Aristotle’s example suggests that at least some dictionary entries will have to carry [...] syntactic as well as semantic information [we can call] the ‘analysis’ or ‘citation’ for the word. (26).11
Equally grotesque results follow when Wiggins interpolates ‘horse’s ankle’ for ‘fetlock’ [42-42a] to make an ‘ambiguous’ version and ‘can’t rule out that mad reading of the genitive which suggests that one horse owned another horse that owned an ankle’ (26). And, in the ‘semantics’ proposed here, ‘all the absurd alternatives must count as theoretically possible’ (3.11).
3.15 Wiggins goes on to, erm, analyse the form of dictionary definitions for items like ‘carnivore’, ‘gaggle’, ‘flock’, ‘Englishman’, ‘Limey’, ‘negro’ ‘nigger’, and so on (26-27, 32), without bothering to make up any sentences at all, as if sloughing off his declared interest in ‘sentence-sense’ (3.4, 8f). Perhaps we have been spared more ‘absurdities’.
3.16 Nonetheless, I surmise he was imagining, without presenting, the contexts and ‘circumstances’ he would exclude from semantic theory (3.10). Some of his words are obviously loaded, but that factor too involves the ‘effects obtaining in a speech-exchange’ which he fenced off from ‘semantics’ in [32]. For example, the 'effect' of using‘nigger’ crucially depends on who does it:
We black people talk about each other that way. [...] 'Niggah, you crazy.' Those are our expressions. If you say it, I'll slap you. (Muhammad Ali to Hunter S. Thompson)
3.17 I would make a similar point about the unscrutinised uses of proper names, pronouns, or definite noun phrases that would entail prior knowledge and assumed contexts, if the sentences were actually uttered and understood by ordinary language users. We can, erm, verify these effects when we examine the same expressions in authentic rather than invented data — a move which Wiggins would decidedly not make, e.g.:
[46]
Baron Ritzner von Jung was by no means a handsome man. The contour of his face
was somewhat angular and harsh, [...] his nose a snub; his eyes
large, heavy, glassy, and meaningless. (Edgar Allan Poe, Mystification)
[47]
It's very odd, but I find Margaret Thatcher enormously attractive [...] I think
it's because she's got good legs. (Punch editor Alan Coren)
[48]
A big strong fellow too, he had good legs on him, he did a lot of
cycling (conversation)
[49]
Oswald had come up to get his knife [...] to cut some wire we were making rabbit
snares of. It is a very good
knife, with a
file in it, as well as a corkscrew and other things. (E. Nesbit, Wouldbegoods)
[50]
We soon sat down to a plentiful meal. I played a good
knife and
fork, as my custom is. (Anthony Hope, Prisoner of Zenda)
According to my data (in computer corpora totalling some 201 million words of authentic texts), a ‘snub nose’ is usually pejorative, not ‘charming’ [46]; ‘good legs’ have a different meaning on a woman (though seeing them on Mrs Thatcher, of all people, certainly is ‘very odd) [47] than on a man [48]; a ‘good knife’ may be one with extra gimcracks for mischievous children [49], or one deployed for eating a great deal [50] (‘pigging out’ in modern parlance). And whereas ‘Limey’ is always belittling or offensive (though not common in my data), the ‘Englishman’ is assigned all manner of qualities, witness the range of adjectives I find just before it: ‘tall’, ‘sturdy’, ‘brave’, ‘free-born’, ‘cultivated’, ‘urbane’, and ‘classy’, but also ‘phlegmatic’, ‘flabby-faced’, ‘grubby’, ‘arrogant’, ‘impudent’, ‘stupid’, and ‘crazy’. The term ‘Englishness’ (surprisingly common in my data) is always complex in contexts [51-52], and implies much more than just ‘the quality of being English’.
[51] Get in touch with the true essence of England. Let the village atmosphere seep into your pores. See if you can make contact with it, this magical thing called Englishness. (Adrian Rendle, So You Want to Be an Actor?)
[52] What are the multiple repertoires of identity buried in this word Englishness? What does it mean to be ‘English’ when so many people are disenfranchised in this country? Here where the streets are not paved with gold, but with garbage. Garbage and the people who sleep amongst it. (Deborah Levy in So Very English)
3.18 Whatever meanings may be attributed to all these ‘words’ has little I can see to do with ‘truth’. Technically, every ‘Englishman’ could qualify as a ‘Limey’, but very few are so designated in my data, and just when one wishes to make the Englishness into some manner of reproach, e.g.:
[53] A mean look crept into his eyes. ‘Hey Jitters, you limey bastid, last Thursday I saw me some Argentinian fellers marching down Main Street with GenTech weapons. You still runnin’ away from that there South Atlantic battle?’ [...] Nobody had known how well equipped the bloody buggering Argies would be. (Jack Yeovil, Krokodil Tears)
Being routed by the ‘Argies’ in that supremely silly Falklands War could all too easily be seen as a national stain.
3.19 In his response to Wiggins, Alston (1971) seems uneasy (though he does so nonetheless) about ‘making some large, and recently highly controversial, assumptions about the possibility of drawing a line between what one means by a sentence and what is true of the subject matter’ (38). He would replace the ‘truth-conditions’ envisioned by Wiggins with something called ‘illocutionary act potential’, being conditions that must obtain or else ‘members of language community’ are ‘enjoined’ from ‘uttering the sentence’ (35). And these conditions are ‘a more general notion’ than ‘truth-conditions’, including the latter as well as the ‘conventions’ and ‘circumstances’ Wiggins would ban from ‘semantics’ (3.10). Thus, ‘drawing a line’ seem less pressing — and less risky too. Alston’s approach.
3.20 Alston too wants a procedure for ‘testing’ the sense of a word in a sentence, but warns that ‘dictionary entries cannot be tested in isolation’ because language (including the ‘evaluation of particular semantic hypotheses’) has an ‘ineradicably systemic character’ (36-37). Still, he hoped for a method of analysing without yet having a ‘proposal for a complete semantic description of the language’. His provisional test works, similar to Wiggins, by interpolating components in the place of a given word in a sentence, viz.:
[54] Harold Stassen is still running.
[54a] Harold Stassen is still actively seeking public office.
[54b] Harold Stassen is still locomoting springily.
[55] The boundary ran from this tree to that tree.
[55a] The boundary extended from this tree to that tree.
[56] John ran from this tree to that tree.
[56a] John locomoted springily from this tree to that tree.
[56b] John extended from this tree to that tree.
I notice here too the use of proper names, pronouns, and definite noun phrases that would normally be contextualised (cf. 3.17). Of course ‘Harold Stassen was still running’; in fact he ran for U.S. president no less than nine times between 1948 and 1992, and stopped only because he died at the age of 93 (running must be good for one’s health). I would have no comparable expectations about ‘John’, the classic, colourless protagonist of innumerable grammarians’ examples; for all I care, he could ‘extend from tree to tree’ by lying down between them because it’s not ungrammatical. Besides, a ‘test’ for ‘run’ cannot be done with ‘locomote springily’, because in all my data, nobody (but Alston) has ever used the expression, not even on the Internet.
3.21 What Alston purported to be doing was
[57] making explicit what resources we are allowing our inquirer. I suggest that we regard him as capable of making reliable judgments of (at least approximate) sameness and difference sentence meaning. [...] That is, we are going to trust his ‘intuitions’ about the semantic relations of sentences taken as (semantically) unanalyzed units, even though, lacking a fine grained semantic description of the language, or even a proposal for such, he is unable at this stage to analyse sentence-meanings into the components contributed by the words that make up the sentence. (38)
Lacking such a ‘description’, Alston proposes ‘a technique that looks promising’, namely ‘partial substitution’:
[58] Although we cannot say what ‘run’ contributes to the meaning of [54], [55], and [56], we can try various substitutions for ‘run’ in each of the two sentences, noting which ones do and which ones do not preserve sentence meaning. [...] What this technique gives us is an indirect, symptomatic approach as a substitute for an unavailable direct account of the underlying structure. (39)
3.22 Alston predictably concedes that his ‘test’ wouldn’t always work, e.g., if ‘some other word or words in the sentence frame’ were to ‘shift its meaning in the course of one of these substitutions’ (40). Unless we assume a ‘semantic fixity of the rest of these sentential contexts’, we fall victim to an ‘infinite regress’ of replacing one word after another and not knowing if the others are holding still in the meantime:
[59] at some point we are forced to invoke a principle of simplicity, according to which terms are held to retain the same meaning over two contexts unless we are forced to recognize a difference. (40)
Notice here that this ‘fixity’ is precisely the consistency with which the investigator imagines contexts for the sentence (cf. 1.4; 4.3ff.).
3.23 Also predictably, Alston admits his test wouldn’t work where we cannot ‘carry out substitutions just for the word in question’ (41), as in:
[60] I ran him a close second.
[60a] I placed second close behind him.
[61] He always runs everything together.
[61a] He fails to distinguish things sufficiently.14
Alston mistrusts this demonstration on purely formal grounds:
[62] What we have in [60a] and [61a] are paraphrases of whole sentences in which several constituents are replaced and the structure is changed. Hence this technique fails to pinpoint the semantic contribution of ’run’. [...] Since we have no resources, analogous to the word substitution test, for prying the sentence apart semantically, we are unable to identify any testable assumptions on which our conclusion depends. (41-42).
Whenever sameness or difference in meaning is ‘properly located at the sentence-meaning level’, proof of ambiguity (Alston obscurely calls it ‘multivocality’) is not conclusive, and indeed may turn into ‘flagrant cases’ of ’philosophical folly’, such as insisting that some minor word is ambiguous on the grounds that the ‘different sentences in which it occurs are tested or verified or established differently’ (44-45).
3.24 In the face of these difficulties, Alston argues, in opposition to Wiggins, that
[63] the philosopher, unlike the lexicographer, does not have to settle questions of word-multivocality. [...] We can give interpretation [sic] of the sentence-types with which we are concerned without going into their fine semantic structure. [...] Often when philosophers think they are essentially concerned with word-meaning, they are really concerned with the meaning or interpretation of sentence-types, [...] thereby avoiding the special difficulties one encounters in analysing sentence-meaning into the semantic contributions made by the various meaningful constituents of the sentence’ (46-47).
Still, Alston would admit the issue of word-meaning to ‘linguistics, even if not to the methodology of analytical philosophy’ (47), as if he were unaware that semantics has always been shared out between ‘linguistics’ and ‘philosophy’.
3.25 Despite the details of the dispute, Alston’s demonstrations some major similarities to Wiggins’ ‘demonstrations’. Both investigators propose to test a word’s meaning in terms of form by interpolating something in its place within a sentence (cf. 3.14 and 3.20). Both assumed that the word’s meaning is its contribution to a sentence’s ‘meaning’ or ‘sense’ — no clear distinction is made between these two terms, though Frege (1892) did exactly that for ‘Bedeutung’ versus ‘Sinn’. Both invent examples containing proper names, pronouns, and definite noun phrases that point off toward an implied but unacknowledged context (cf. 3.17; 3.20). And, as we can see from their actual practices, both are in fact less concerned with linguistic meaning than with linguistic form: the detailed form or surface structure of sample sentences or dictionary entries is again made out to be a main issue for theories of ‘meaning’. So both are keen to find a well-structured means of going between word-sense and sentence-sense, on the (false) assumption that the meaning of the whole is the sum of the meanings of its parts.
3.26 Alston and Wiggins also have major divergences in their ‘demonstrations’. Alston is interpolating components which, unlike Wiggins’, are not presented as dictionary entries; they are simply supplied as being, in the investigator’s opinion, equivalents for certain expressions. Alston alone draws the boundary of his own discipline, namely ‘analytical philosophy’, so as to exclude the formation of the lexicon; and he certainly doesn’t make his boundary coincide with the issue of ‘verification’ (or ‘truth-conditions’) as Wiggins did (3.9). Alston suggests that this issue can give rise to ‘philosophical folly’, though his examples don’t clarify whether Wiggins might be one given to such foolery.
3.27 In a short ‘reply to Mr Alston’, Wiggins (1971b) found it ‘pleasant’ when the two ‘concur in the conviction’ that ‘for a certain word to have a certain meaning is for it to make a certain contribution to some appropriate semantic property of sentences in which it occurs’ (48). All the same, Wiggins refuses to amend his own position. He myopically insists that the ‘indicative’ has ‘theoretical primacy’ over counterfactual, commands, etc.; and that ‘truth-conditions’, not ‘the efficacy of the uttering’, are the ‘semantic problem’ (48-49). He denied that ‘the validity of the substitution test’ as such — a major point for Alston (3.20’22) — is important for the refutation of ’unitary analyses’ : his ‘claims’ about how ‘we can understand or “hear” the sentence’ ‘is itself independent of any substitution test’ (50). And he sees his project of ‘giving the dictionary expansions by which some words need to be supplanted in order to display’ the ‘logical form of the sentences in which they occur’ as a ‘separate enterprise’ from ‘marking out and characterizing interpolatable words’ (51).
3.28 Wiggins’ fresh attempt to weaken the speech-act approach (cf. 3.10) entrains him in sample ‘conversations’ he himself calls ‘mad’ (51-52), such as:
(71) Y: I promise I will be there.
X: Go on then. Promise!
Y: I already have.
X: You haven’t. All you’ve done is assert that you do.
Y: All right. I will be there.
X: But I want you to promise to be there, not just say that you will be!
Following Davidson (1969), Wiggins asserts that the performative is a matter of ‘two utterances’ of which the first (e.g., ‘I promise this’) makes a reference to the second which is an ‘autonomous but subordinate utterance’ of the other (e.g., ‘I will be there’). And the analysis of each ‘utterance’ by itself should concern ‘not convention,’ but logical form’. The same solution has understandably been proposed by linguists for ‘deep structure’ (e.g., Ross 1970; Sadock 1970).
3.29 As so often in philosophy and linguistics, a suspiciously artificial special case is enlisted to affirm the generality of an account. In my view, the reason why ordinary language users would not carry on such a conversation is precisely that having heard ‘I promise I will be there’, a normal person assumes that there is not a difference between the promise and the asserting of the promise: for all practical purposes (aside from such odd cases as lawsuits), the two are construed as equivalent in meaning and intention. ‘Speaker X’ is nit-picking in a truly ‘mad’ fashion, and no sane (non-mad) person would go ‘there’ to meet him.
3.30 In sum, Wiggins and Alton remind me of two irritable seamen in the same boat quarrelling whilst they absent-mindedly bail water into each other’s laps as the boat sinks. The notion that ‘the production and understanding of utterances’ can be ‘explained’ [28] by inventing and interpolating handfuls of simple-minded sentences is just a stale reflex of armchair philosophy dressed up as armchair psychology, which Hartmann staunchly rejected. The latest edition calls itself ‘relevance theory’, which, Deidre Wilson arrantly vowed in a conference talk in 1991, ‘tells you exactly how much mental processing is needed’. When I pointed to conclusive experimental evidence from real psychology that processing just doesn’t work that way, she paid no attention, as I should have expected. ‘Irrelevance theory’, I’d say.
4. Back to the ‘text’
4.1 To expropriate Wiggins’ own terms in 3.7, our demonstration texts span a range between ‘inadequately founded rejection’ and a ‘half-hearted acceptance’ of context as a determiner or ‘meaning’ or ‘sense’. Semantics is enjoined to anticipate the meaning of whatever might be ‘strictly said’ by providing a ‘dictionary’ of ‘word-senses’ and a ‘systematic, general, and uniform procedure’ for ‘analysing any sentence in the language’ (3.9). The universal guideline is stated as ‘knowing under what conditions the sentence would count as true’ (ibid.). Everything that might prove inconvenient can be dumped at the door of the (rather absurdly named) ‘science of the further effects obtaining in a speech-exchange’ [32] — which would seem to be some sort of stunted pragmatics without semantics.
4.2 This arcane project entails some patently untenable assumptions about the ‘production and understanding of utterances’ which it arrantly purports to ‘explain’ [28]. What is ‘strictly said’ would make no sense and have no meaning if speakers and hearers were trapped behind a ‘boundary’ isolating them from ‘circumstances, conventions’, and so on, which would in any case be indispensable for determining ‘under what conditions’ it ‘would count as true’ (cf. [29]). In fact, I cannot imagine any ‘conditions’ under which the accounts of either Wiggins or Alston ‘would count as true’ for the human ‘production and understanding of utterances’.
4.3
For all their unmistakable self-consciousness, these two philosophers do not
seem to have reflected with real profundity upon their own proceedings. They are
thus manoeuvring at the opposite end of the spectrum from Hartmann, whose
precisely delimited conditions and orders
of ‘phenomena’ expounded back in section 2 got collapsed, conflated and
confused in their discussion.
Thus, if ‘forms symbolize the procedures that create them in use’ (2.5.3),
then inventing examples creates a special use that in turn gets symbolized in
the forms you find. If ‘the content of forms is their designation of the
categorical’ (2.5.2), then your prior categories must affect your discovery of
forms. If ‘the essence of form is to be or document a relation’ (2.5.1),
then the relation of a sample to its setting can determine the form it assumes.
If ‘formalization is a typifying removal of the repeatable’, (2.5.4), then
you have to consider whether your samples are in fact repeatable, i.e. whether
they are likely to occur anywhere outside your discussion. Finally, if language
acquires ‘form’ by serving as ‘a documentation of human reason in the
activity of constituting relations’ (2.5.5), then whoever formalizes will have
to bear in mind the operations that make
‘form’ possible.
4.4
Though sanctioned by custom in philosophy and linguistics, inventing your own
data to support your theory proves an uncalculated risk once you turn a
searchlight on authentic data instead. What is proffered as ‘truth’ is found
to be a reductive mix of blinkered literalness and naive realism that, if
anything, should not engender absurdities (cf. 3.11, 14). If I am
told that in a certain ‘wabe’, ‘the ‘borogoves
were all mimsy’, I’d feel absurd judging the statement on its ‘truth’. I
can verify it, if at all, only by studying John Tenniel’s famous illustration,
where they do look very mimsy; but I can’t formulate the
‘conditions’ that made them ‘truly’ so.
4.4
Hartmann himself was kind enough first to give me his weightiest book just when
I was setting off on my long-term safari into ‘text linguistics’, and when I
later returned the favour by giving him mine, he commended me as a pioneer
‘passing through stages’ toward a ‘realistic science of language’
(‘realistische Sprachwissenschaft’), which, as I came to understand in time,
indeed might not turn out to be ‘linguistics’ in the established sense. My
acute shock upon his sudden and premature death was transmuted into more of the
fanatical and maverick energy that had carried me through my own weighty study
of ‘text production’ with 1,251 ‘references’ (Beaugrande 1984), which
had not surfaced as a major issue even when ‘understanding’ and
‘comprehension’ were finally becoming ‘realistic’ in Hartmann’s sense.
(I cannot recall without an acrid twinge of poignancy that he travelled to a
symposium in ‘text production’ in some dreary town like Kassel or Hildesheim
because my name had been advertised as a speaker, and I was not invited; those
who were had done precious little work in the area, and when I mentioned this to
the publisher, he dropped the publication of the proceedings. Bad patch all
round.)
4.6
My own research in striving to unify literally dozens of areas made me
increasingly curious about ‘linguistic texts’, which the famous realist J.R.
Firth (1957, 1968) often wryly called ‘language turned back on
itself’. The harder and further I looked, the more I ran up against airy-fairy
accounts like those I reviewed in section 3. So when I was nominated for a
Guggenheim Fellowship by yet another grand realist, Win Lehmann, although my
project was a close-range ‘discourse analysis’ of literary theorists’
(Beaugrande 1988), I felt I owed it to myself — and to Lehmann as well as
Hartmann — to apply the same approach to prominent works of linguists
(Beaugrande 1991). As far as I know, my summary of Hartmann’s own discourse
there is the only one published in English, and Lehmann expressly told me he was
‘glad you got him in’.
4.7 One lesson from my extended analysis is that there seems to be no consensual ‘ground rules’ among seminal linguistic theorists, whose ‘language turned back on itself’ I discovered to be replete with strenuous, unreflective swashbuckling and bootstrapping. The only one I found who never flatly contradicted himself or else demanded the impossible was in fact Hartmann, who wrote with a level of consciousness that still strikes me as superhuman. Halliday ran a close second, whereas such ostensible mentors as Saussure, Hjelmslev and Bloomfield, on close inspection, are seriously incoherent, and nowhere more so than when they try to lay down ground rules.
4.8
Text linguistics inherited this fix, but the steady contact with texts weaned us
away from the inheritance. The gallery of quotations in 1.3 shows to my mind a
crucial stage in this evolution, and it is no coincidence that a number of the
voices came from Hartmann’s pupils or colleagues. He himself did not attempt
to put in practice any ‘ground rules’ — which is why his students
affectionately called him ‘meta-Hartmann' — or enforce any upon them.
Moreover, the volumes I have cited are rather epistemologies of
‘grammar’ and ‘language science’, that is, essential cognitive and
linguistic conditions and preconditions for anyone pursuing such a science. His
intent was to spell them out in the manner of a ‘consciousness-raising
exercise’ that liberates the reader from the narrowing boundaries and
dichotomies of what we came to call, for purposes of contrast, ‘sentence
linguistics’. He did not even essay to predict the outcomes, so that the ‘diversity’
I mentioned back in 1.2 was quite compatible with a shared sense of the need to
renew and build up the relation between theory and practice in order to
accommodate the study of texts, rather than, say, declaring them to be mere
‘sequences of sentences’ to ‘analyse’ such hoary methods as those of
Wiggins and Alston.
4.9
If text linguistics can be said to have ‘ground rules’ then they would
consist of looking at large amounts of texts and following their lead; balancing
general with specific, and uniform with diverse; viewing texts as linguistic,
cognitive, and social events; systematically relating our own texts to the texts
we are studying; and explicitly taking on the social responsibility for the
implications and applications of our results. Resourcefully applied, such
‘rules’ encourage us to continually reinvent the role of the author,
lecturer, conference speaker, teacher, and so forth. For me, at least, each new
book, paper, speech and so by the same author is in some ways easily
distinguished from it predecessors. But this principle has never been really
palatable to editors and publishers, and today less than ever as commercial
success has come to dominate all other concerns, whereas during the salad days
of text linguistics and discourse analysis, their more commercial ones would be
used to support the less.
4.10
In recent years, critical discourse analysis (now mostly just called CDA)
has not only appeared at our gates, but surged in a set up an increasing visible
camp. Its products are like hot potatoes — they sell well but they sellers in
publishing houses are loath to handle them. There may be a blessing in disguise
here, because pandering to commercial pressure is utterly inconsistent with the
emancipatory ideology of CDA. If we are committed toward promoting linguistic,
cognitive, and social equality, then our work must not be kept out of general
reach by the pompous, obscure styles and stratospheric prices that attend on
more books than I have room to list here, even if I felt rude and rash enough to
try. So when the publisher of my latest book loosed upon me a gaggle of
‘reviewers’ with arrant demands for savage ‘revisions’ to make it more
commercial, I did what I ought to have done before: I posted it for free on my
website, and went on to scan format, and upload all of my steadily more
expensive books as well, whereas before I had only those papers that were
already in electronic format. And now come a fresh set of papers, including this
one. (The web page counter was around 5,000 visits when the books went up; you
may compare that to where it stands now.)
4.11 Looking to the future, I would foresee a further upward trend in CDA, which can recycle the methods of text linguistics and ‘pre-critical’ discourse analysis whilst our time grows more ‘critical’ in a different and ominous sense — whilst academics are endlessly confronted by savage public attacks from press and politics not merely upon our own enterprises, but upon civil rights, consumer health, worker safety, and the environment (Beaugrande 2004). No middle ground or sheltered preserve remains for dithering on about some cloud-cuckoo ‘systematic, general, and uniform procedure competent to analyse any sentence’ (cf. 3.9). Our lives, both inside and outside academia, are on the line.
4.12
I am under no illusions about the daunting enormity of the tasks ahead or the
strength and power of those who would shout us down. (I feel ever more like
Frodo Baggins looking down on the orc-filled plains of Mordor.) But as of this
writing (August 2005), the Internet is still free, for good and for evil. If
this freedom is onr day widely abolished, as it already has been in some
countries I know, CDA websites will undoubtedly be shut down by hired hackers
— and mine too. Which is why I (sort of) happily endure the time-robbing and
mind-bending toils of getting my works into cyberspace now. And why I appreciate
your coming ‘more than I have power to tell’.
Notes
1.
More precisely, the volume
united a set of formal and theoretical studies composed and published over a
series of preceding years.
2.
Hartmann was prominently cited by Schmidt, Koch, and Harweg, the last of whom
compared his very broad notion of ‘substitution’ to Hartmann’s notion of
‘determination’ (Harweg 1968: 68ff).
3.
Where needed, these passages are my translations from the German, and have
undergone some abridgements for the sake of readability or style.
4.
I use this term for any investigator into language, whether academically
designated a ‘linguist’ or not.
5.
These passages are still my translations, but with slight editing.
6.
‘Thing’ (as German ‘Sache’) doesn’t imply a concrete object
(‘Ding’ in contemporary German, though not for Immanuel Kant), but merely
something with an identity. Proper nouns names are not fundamental to
Hartmann’s account (cf. 1963a: 79f), because they are atypically unconcerned
with the issue of imposing unity’ on ‘plurality’, especially people’s
names.
7.
I apply the term ‘framing’ wherever one discourse quotes or absorbs from
another (Beaugrande 1997, 2004).
References
Alston,
William. 1971. How does one tell if a word has one, several, or many senses? In
Steinberg & Jakobovits (eds.), 35-47.
Austin,
John. 1961. The meaning of a word. Philosophical Papers. Chapter 3.
Oxford: Oxford University Press
Austin, John. 1962. How to Do Things with Words. Cambridge: Harv