LINGUIST List 8.1060

Thu Jul 17 1997

Review: Kronning: Modalite, cognition ..

Editor for this issue: Andrew Carnie <carnielinguistlist.org>

Hans Kronning (1996) Modalite, cognition et polysemie: semantique du verbe
modal _devoir_. Uppsala: Acta Universitatis Upsaliensis (= Studia Romanica
Upsaliensia. 54), 200 pages.

Reviewed by Ulrich Detges <ulrich.detgesuni-tuebingen.de>

The last fifteen years have seen a considerable rise in influence of
cognitive approaches in linguistics. It is symptomatic that many
linguistic theories - which by themselves have nothing to do with
psycholinguistics - claim a cognitive status for their theoretical
description of language. It would be unfair, however, to raise this
objection against Hans Kronning's book on modality, cognition and
polysemy, which deals with a semantic and syntactic description of the
French modal devoir must.

	Basically, Kronning's cognitive approach merges three theoretical
components, namely

	(a) Cognitive Grammar as set forth in Ronald Langacker's Foundations of
	Cognitive Grammar (1987),

	(b) Prototype theory (E. Rosch) and

	(c) the theory of polyphony formulated by Oswald Ducrot.

Ducrot's theory of polyphony is certainly not genuinely
cognitive, even though it surely is one of the most interesting and
far-reaching theoretical approaches of the past fifteen years.
Unfortunately, Hans Kronning's book does not really manage to fit these
three appraoches together in a consistent way.

	Kronning's analysis of devoir ("must") has two major parts:

(1) a semantic and syntactic analysis of the different meanings of devoir
(chapters 3 - 6),

(2) a polysemic account of the relationship between different basic and
less basic senses of devoir(chapters 7 - 11).

Traditional interpretations of devoir oscillate between four basic
positions:

(a) The homonymic hypothesis claims that there are two distinct verbs
devoir, one with a deontic meaning (obligation), one with an epistemic
meaning (probability).

(b) The polysemic hypothesis postulates one verb devoir with two
semantically related meanings (obligation and probability).

(c) The monosemic approach reduces all different uses of devoir to a
common semantic invariant.

(d) The traditional point of view simply distinguishes a great number of
different (somehow interrelated) acceptations of devoir (Kronning quotes
37 of such different acceptations).

	In the first step of his argumentation (chap. 3), Kronning
postulates not only two different meanings (deontic obligation and
epistemic probability), but three, the third being what he calls alethic
modality (abstract necessity) as in

(1) 	Tout ce a quoi on se refere doit exister.
	Everything one can refer to must exist.

Among the numerous senses that can be attributed to the modal devoir, the
deontic, alethic and epistemic meaning have a fundamental status. Chapter
4 sets forth a semantic analysis of the three basic meanings of
 devoir. The structure of alethic modality functions, in a sense, as a
link between epistemic and deontic modality:

Deontic: 	Necessity to make s.th. exist (which can be either true or
		false)

Alethic:	Necessity to exist (which can be either true or false)

Epistemic:	Necessity to exist (which is neither true nor false)

The necessity-element which is basic to the three semantic structures is
an apodictic necessity. Something is necessary in an apodictic sense, if
it is always true/true in all possible worlds of a given universe.

	Apodictic necessity as well as the deontic, alethic and epistemic
meaning are the result of inferential operations; for deontic devoir
(2), this operations proceeds from (2.a) to (2.c) via (2.b):

(2) 	Louis XIV doit etre honore.
	Louis XIV must be honoured.
 	In all possible worlds of a certain modal universe the state of
 	affairs [Louis XIV be honoured] must be the case.
(2.a) 	Les Rois doivent etre honores.
 	Kings must be honoured.
(2.b)	Louis XIV est Roi.
 	Louis XIV is the king.
(2.c)	Donc Louis XIV doit etre honore.
 	So Louis XIV must be honoured.

Very often, the linguistic context contains traces of the apodictic
premises of the inferential operation. For (3), the inferential
operation works as pointed out in (3.a) - (3.e)

(3)	Les juifs lui repliqurent: Nous avons une loi, et selon cette loi
	il doit mourir parce quil s'est fait Fils de Dieu (Jean 19:7,
	Bible, 1982, cf. Kronning: 33)
	The Jews replied to him: we have a law, and according to that law
	he must die, because he made himself Gods son.
(3.a)	Quiconque insulte le nom du Seigneur doit etre mis a mort.
 	(Levitique 24: 16, Bible 1983, cf. Kronning: 33).
	Whoever insults the Lord's name must be put to death.
	(3.b) Celui qui se fait Fils de Dieu se
	fait Dieu. (Cf. Jean 5: 18, cf. Kronning: 33)
	He who makes himself God's son makes himself God.
(3.c)	Celui qui se fait Dieu blaspheme (Cf. Jean 10: 33, cf.
	Kronning:33)
	He who makes himself God commits blasphemy.
(3.d)	Jesus s'est fait Fils de Dieu (Cf. Jean 5: 18, cf.
	Kronning: 33)
	Jesus made himself God's son.
(3.e) 	Jesus doit mourir (etre mis a mort) (Jean 19: 7, cf.
	Kronning:33)
	Jesus must die (be put to death).

The starting points of such inferential operations, (2.a) and (3.a),
are nomic rules. Modal universes are structured by such nomic rules, which
are subjected to inter- and intra-individual variation. (In a
force-oriented description of modality like the approach proposed by Talmy
(1988), these nomic rules would have the status of modal forces.)

	Note that in (2.a) and (3.a) the nomic rules already contain the
modal devoir, so that there is a danger of circular argumentation. As
Kronning points out, however, nomic regularities do not necessarily
contain modal predicates; it is possible to formulate (2.a) and (3.a)
in the present indicative ("Kings are honoured") or the future ("Whoever
insults the Lord's name will be put to death") of the simple verb.
In Chap. 5., Kronning shows - within the framework of Oswald Ducrot's
"theorie de la polyphonie" - that deontic, alethic and epistemic
modality belong to different levels of the utterance stratification.

	Deontic and alethic devoir (the meanings which can either be false
or true, as pointed out in chap. 3), belong to the "dictum" (the
proposition).

	Epistemic devoir, on the other hand, does not belong to the
proposition; rather, it is in itself a statement about the truth of
the proposition. In order to account for these properties in a cognitive
perspective, Kronning develops a theory of the utterance stratification.
On its most basic level, the utterance-structure of the sentence is
composed of (a) the substrate and (b) the focus ("foyer"). In
cognitive terms, the substrate is the background of the utterance; its
truth is normally presupposed by speaker and hearer. The focus is the
figure of the utterance; it is what is relevant in the utterance or what
makes the utterance relevant. For sentence (4), (4.a) is the substrate,
(4.b) the focus:

(4) 	Pierre est parti hier.
 	Peter left yesterday.
(4.a) 	Substrate:	Peter left [at a certain moment X]
(4.b)	Focus:		X = yesterday

Both focus and substrate can be either false or true. Of the different
meanings of devoir, only deontic devoir can appear in focus-position:

(5) 	Est-ce vrai, Roger, que les jeunes gens avant d'etre
	soldats, doivent se mettre nus... - ..Naturellement, ils le
	doivent. (Apollinaire 1911: 152, cf. Kronning: 63)
	Is it true, Roger, that the young, before becoming soldiers, must
	get undressed... - .. Of course they must.

The assertion of the focus is regulated by the adfocus. In (6) the
focus is "au debut de 23" ('in the beginning of 1923'), adfocus is "a du"
('must have ...'). Here, it is out of question that Dad met Mom; what
the speaker is not sure about is the precise moment of their first
meeting:

(6) 	Papa a du connaitre Maman au debut de 23.
	Dads must have met Mom in the beginning of 1923.

Epistemic devoir is, as this example shows, part of the adfocus.

	Another layer of the utterance-stratification is made up of the
division into topic and comment. Within the comment, we find the comment
operator specifying tense and aspect of the comment (and hence of the
sentence as a whole). Within the sentence, there is a growing degree of
cognitive salience from the topic on the one hand (low salience) to the
comment on the other (high salience). Within the comment, we find a
growing degree of salience from the comment operator on the one hand (low
salience) to the focus (high salience), which forms the centre of the
comment. According to Kronning, one of the more specific uses of alethic
devoir is its use as a future-tense-auxiliary. So, alethic devoir is
clearly linked to the function of a comment-operator:

(7) 	Le typhon Arthur doit(op) atteindre la Reunion dans les
	heures qui viennent.
	The hurricane Arthur is going to/will reach La Reunion in the next
	couple of hours.

The hypothesis that deontic, alethic and epistemic devoir are linked
to different functions in the utterance stratification is confirmed by
independent syntactic evidence.

	Kronning comes to the following conclusion:

Deontic devoir 	can be part of the substrate; it can be in focus
		position. It can never be adfocus.

Alethic devoir can be part of the substrate; it can never be focus nor
		adfocus of the utterance.

Epistemic devoir can never be part of the substrate nor in the focus
		of the utterance; it can, however, assume the role of
		adfocus.

As Kronning claims, this analysis provides the basis for the solution
of a problem inherent in the semantic analysis of epistemic devoir as
shown in chap. 4. In this section of the book, the epistemic meaning was
analysed as a necessity to exist (i.e. something 'that must be the case'
is necessarily true). On the one hand, this analysis is somewhat
problematic, since epistemic devoir (like must in English) expresses not a
necessity, but (only) a strong probability. On the other hand, the
assumption of a semantic element "necessity" is crucial, since it is this
very element which is - according to Kronnings analysis - responsible for
the polysemous coherence between epistemic, deontic and alethic modality.
According to Kronning, the meaning "strong probability" is only a
secondary "effet de sens" due to an inferential interpretation. As we
have seen, epistemic devoir, as an adfocus, can in itself never be false
or true. Since the truth it expresses cannot be debated by speaker and
hearer, its modal force is rather weak. So, an application of the
maxim of quality ("use the strongest modality appropriate to the
situation!") gives rise to an interpretation of epistemic necessity as a
strong probability or a mere possibility.

	This argument, which is crucial, since it links the semantic
analysis of devoir to a description of its polysemous structure, seems
to me of doubtful validity: the semantic particularity of epistemic
devoir is explained by its function within the structure of the utterance,
whereas a cognitive description of modality should explain which semantic
and conceptual properties of epistemic devoir cause the rather restricted
range of its syntactic functions. This, in turn, raises the question
what empirical evidence - apart from introspection and philosophical
speculation - the semantic description shown in chap. 4 is grounded
on.

	As I have already pointed out, the chapters 7 - 11 deal with
questions concerning the polysemous structure of the different meanings
and acceptations of devoir.

	Homonymy is normally an accidental phenomenon. So, homonymy is
ruled out as an appropriate model to capture the relationship between
epistemic, deontic and alethic devoir by the fact that (deontic)
obligation and (epistemic) probability are in many historically
unrelated languages designated by the same lexical item.

	Monosemic approaches on the other hand reduce different
acceptations of a word to a single semantic invariant "core meaning"
or nuclear meaning, very often an extremely abstract superscheme. All
acceptations of the word are derived from this core meaning by
inference. A candidate for the status of such a nuclear superscheme in
Kronning's analysis certainly is apodictic necessity (see chap. 4). What
distinguishes Kronnings approach from a monosemic analysis is the idea
that the different meanings of devoir are not ad hoc derived from the
nuclear meaning "apodictic necessity", but have attained conventional
status - the speaker must learn, which uses are permitted in a given
language. So, Kronning's concept of polysemy is quite close to a
monosemic model.

	One of the conventionalised uses of devoir is its categorial
prototype. The categorial prototype guarantees the unity of the
category and the coherence between the different uses.

	According to Kronning, it is the deontic meaning of devoir that
must be considered as the categorial prototype.

	In chapters 8 and 9, Kronning gives a description of the
polysemous structure of devoir within the framework of Langacker's
Cognitive Grammar. Deontic modality is the prototypical modality,
alethic modality is, on the one hand, an extension of the prototype. On
the other hand, it serves as a local prototype for the epistemic meaning.

	Chap. 9 shows that, within the polysemous structure of devoir,
deontic, alethic and epistemic modality function as basic level (in
the sense of prototype-theory), which in turn give rise to a great number
of more specific acceptations.

	Chap. 10 gives diachronic, cognitive, ontogenetical and other
evidence for the hypothesis that, among the different meanings of
French devoir, deontic obligation is the prototype meaning. This evidence
is, in itself, absolutely convincing. What seems problematic about
Kronning's argumentation is the fact that this evidence has absolutely
nothing to do with the analysis given in the earlier sections of the
books. In other words, we learn that it is the deontic meaning which is
the prototype, but we do not learn why this may be so. In chap. 10.,
Kronning states, for example, that diachronically and universally, it is
always the deontic meaning which gives rise to the epistemic - a cognitive
description should be able to explain, why this is the case. According to
the analysis developed in the earlier chapters of the book, one would
rather expect the alethic meaning to be the prototype, because it is
closest to the underlying invariant "apodictic necessity" and because in
Kronning's analysis it serves as an intermediary between deontic and
epistemic meaning.

	In chap. 8, Kronning invokes Traugott/Koenig's (1991) model of
metonymic, context-induced semantic change, without noticing that
Traugott/Koenig's analysis is in deep contradiction to his own.
Traugott/Koenig explain the diachronic transition from deontic to
epistemic modality as follows: "If I say 'She must be married in the
obligative sense', I invite the inference that she will get married.
This inference is of course epistemic, pertaining to a state of affairs
that is anticipated to be true at some later time". Traugott/Koenig
postulate a direct metonymic link from deontic to epistemic modality,
whereas in Kronning's analysis, there is no such link. Furthermore,
Traugott/Koenig's analysis explains the unidirectional character of the
transition from deontic to epistemic modality: it is possible to infer a
strong probability from an obligation, but there is no way one could
possibly infer an obligation from a strong probability. If one considers
polysemy as a panchronic phenomenon (an idea already formulated by Ullmann
1962), then it follows that the relationships between the different
meanings of a lexical item on a synchronic level are identical with the
conceptual bridges along which, on a diachronic level, the transition from
one meaning to the other is achieved. In other words: the metonymic bridge
indicated by Traugott/Koenig may well be the conceptual link that
connects obligation to probability in the polysemic structure of devoir.
In Hans Kronning's analysis, this link is not metonymic. Rather it is made
up by a common semantic element "apodictic necessity" - recall that in
Kronning's analysis, it proved to be difficult to explain, why epistemic
devoir did not express a necessity, but a strong probability (end of chap.
5).

	A last question I would like to ask is how useful the notion of
prototype is to the description of polysemy. In footnotes 295 and 296
on pages 92 to 93, Kronning discusses languages in which there is a
difference in the syntactic structure of the deontic and the epistemic
variant of a verb (Spanish) and a case where the deontic meaning has
disappeared or is less common than the epistemic meaning (Catalan)-
cases which Kronning, for obvious reasons, wants to discuss away. As we can
see in modern English, it makes little sense to argue that the loss of the
"old" prototype (deontic meaning) is caused by a shift in the
polysemous structure of the verb, by which the epistemic meaning becomes
the "new" prototype. In English and in modern Spanish, we have a
situation, where there is an old lexical item which has both deontic and
epistemic meaning ("must" in English, "deber" in Spanish). Now, as far as
their deontic meaning is concerned, must and deber tend to be replaced by
"younger" rivals (English "have to", Spanish "tener que"), which, in turn
have not (yet?) given rise to epistemic uses. So, the weakening of the
prototypical deontic meanings of must/deber (which may eventually
result in the complete loss of these meanings) does not have anything to
do with the relationships that those deontic meanings entertain with the
epistemic meanings of the verbs in question.

	Hans Kronning's book is tough work to read. It only has some 150
pages, but it took me about a week to work my way through it. It is
very densely written and the argumentation is on an extremely high level.
The ambition of this book is not only to give a description of the French
modal devoir; it is clearly meant to be a contribution to the theory
of syntactic and pragmatic description of the sentence and the theory of
polysemy. Unfortunately, it comes up to this ambition only in part.