BMCR 2009.01.29, Jo Willmott, The Moods of Homeric Greek

Jo Willmott, The Moods of Homeric Greek. Cambridge Classical Studies.
Cambridge/New York:  Cambridge University Press, 2007.  Pp. xii, 264.
ISBN 9780521879880.  $99.00.

Reviewed by D.M. Goldstein, University of California, Berkeley
([email protected])
Word count:  3513 words
-------------------------------
To read a print-formatted version of this review, see
http://ccat.sas.upenn.edu/bmcr/2009/2009-01-29.html
To comment on this review, see
http://www.bmcreview.org/2009/01/20090129.html
-------------------------------

Table of Contents
(http://catdir.loc.gov/catdir/enhancements/fy0827/2008272515-t.html)

Jo Willmott's new book, The Moods of Homeric Greek, critically examines
the meaning and use of the modal forms in Homeric Greek.

Willmott advances the claim that the traditional accounts are often
inadequate, both theoretically and descriptively, and consequently
proposes significant revisions.  Her conclusions bear not just on
modality in Homeric Greek, but also on modality theory in general.
This is an important book, as it has set a new standard in description
and analysis of the Greek moods.

The synchronic analyses of the Greek moods in the book are excellent:
they are more sophisticated and detailed than those found anywhere in
the standard handbooks.  Close study of this book will make one a more
semantically sensitive and thus better reader of Greek.  I have only
two general criticisms.  First, I think more needed to be said on
diachrony, as the explanations offered for how and why certain semantic
changes occurred left me wanting more.  Second, the semantic analyses
could have benefited from more engagement with pragmatic meaning.
Overall, however, I recommend this book earnestly.  A broad Classics
audience will benefit from it.

The Layout

In brief, what Willmott does is to demonstrate how the traditional
doctrine of the moods fails both theoretically and empirically, and to
offer in turn new accounts of their semantics.  This involves not only
wrestling with high-order questions like the semantic maps of the
indicative, subjunctive, and optative, but also more fine-grained and
practical ones about the (semantic and pragmatic) interpretation of
specific lines.  Willmott begins in chapter one by laying out the scope
of the work and discussing her corpus and its concomitant pitfalls.
Chapter two presents the theoretical framework; here she reviews
traditional accounts of the moods and surveys more recent work in
linguistics, in particular from the area of grammaticalization (on
which more below).  Chapters three, four, and five are the heart of the
book, dealing with the indicative, subjunctive, and optative,
respectively.  Chapter six offers a look at moods in subordinate
environments, and chapter seven concludes the work.  There are two
appendices, one on formal markers, the other a catalogue of mood uses
in the two Homeric poems (which will prove extremely useful); as well
as a bibliography, index locorum, and general index.  Below I present
and discuss some of the highlights and interesting discoveries of the
work.  I then look more closely at the book within the framework of
grammaticalization.  This is followed by a consideration of the
semantics-pragmatics interface.  I conclude with minor, sundry
criticisms.

The Greek Moods Revised

So what does one gain from reading this book?  Simply put, new semantic
maps of the indicative, the subjunctive, and the optative.  Willmott
upends the traditional semantic accounts of the moods (as found in,
e.g., Chantraine, Monro, Goodwin, etc.), such as their position on an
"irrealis continuum" with the indicative being the most realis or
factual, the subjunctive less so, and the optative least so.

The Indicative

Willmott replaces the traditional system with an account that is
theoretically more coherent and that captures far more of the data.
For the indicative, Willmott argues that the descriptions of the mood
as one of "reality," or "fact," or "objectivity" are unsatisfactory.
She points out that this claim runs into problems with the appearance
of the indicative in conditional sentences, especially counterfactual.
There the counterfactual nature of the clause flatly belies the
proclaimed factual essence of the indicative.  Willmott resolves this
by arguing that the indicative marks "positive epistemic stance": that
is, even if a speaker is not committed to the truth of a proposition,
he is at least committed to the possibility of its truth (p. 51).  This
requires us to accept that a counterfactual implicates, rather than
entails, the falsity of the protasis and apodosis.

Take Il. 16.617 as an example (from p. 49), where Aeneas says to
Meriones:

<greek>*Mhrio/nh, ta/xa ke/n se kai\ o)rxhsth/n per e)o/nta</greek>
<greek>e)/gxos e)mo\n kate/pause diampere/s, ei)/ s' e)/balon
per.</greek>

'Meriones -- even though you're a dancer, my spear would have
stopped you completely, if I'd hit you!' (trans. Willmott)

...

Here Willmott notes that the events described by the conditional,
Aeneas' hitting Meriones with a spear, did not happen.  Nevertheless,
Aeneas suggests pragmatically that it was a real possibility for him to
have struck Meriones (even though he did not).  If it were not,
Willmott claims, his taunt would hardly be effectual.

Her analysis is attractive and persuasive.  I wondered how "positive
epistemic stance" would line up with an earlier analysis that accounts
for indicative counterfactuality in terms of pragmatic assertion (vs.
the non-assertion of oblique moods).[[1]] Despite the appeal of her
analysis, I thought more should have been said here on the contribution
of the "modal" particle <greek>ke/(n)</greek>, as it was given only a
paragraph's worth of attention.  Willmott suggests a compositional
analysis, whereby the particle and the indicative each contribute to
the meaning of the construction: the indicative can be understood as
marking the positive epistemic stance of the subject, while the
particle together with the protasis could mark the difference between a
counterfactual indicative and its normal use (p. 50).

The Subjunctive

The chapter on the subjunctive is the most complex.  To begin with
diachrony, Willmott claims that the subjunctive was "originally" (in
Proto-Greek? Proto-Indo-European?) some kind of future marker, but one
that was developing different meanings from the future indicative.  The
Homeric subjunctive thus occupies a transitional position between
future tense and true mood, but is developing more and more modal uses
(p. 79).  This trajectory is completed, so to speak, only in Attic,
where the "future" meaning of the subjunctive is restricted to
subordinate environments (and this is just where we expect the
preservation of older morphosyntactic patterns).  According to her
analysis, the subjunctive is the older form, while the future
indicative is "perhaps a 'younger' future marker" (p. 79).

There is a long-standing debate as to whether the Greek future descends
from the short-vowel aorist subjunctive or from a desiderative morpheme
(that is, a verbal form that encodes the meaning 'wish, want' on the
base semantics of the verb; such forms are attested in Sanskrit).
Willmott leans toward the former, but does not press her case too
strongly, as it would require getting mired in complex details of
historical phonology.  Still, I thought Willmott could have argued more
forcefully for the aorist-subjunctive origin given her findings on the
semantics and use of the future.

To move away from diachrony, Willmott's synchronic claim for the
subjunctive is that it is not more "irrealis" than the future
indicative.  She looks at the epistemic subjunctive (which others refer
to as the "quasi-future" use of the subjunctive) in various
morphosyntactic environments, e.g. root clauses, conditionals,
subordinate clauses.  For main clauses she argues against an analysis
whereby the future marks projected reality, while the subjunctive only
potential reality. Under this type of model, the speaker uses a future
indicative when he has considerable confidence, if not certainty, in
the future occurrence of the event, while with the subjunctive it is
only a possibility.  Willmott is careful in reviewing the evidence, and
shows that this distinction is not borne out by the Homeric evidence.

Analysis of the subjunctive and future indicative in Homeric Greek is
no easy task, as these two categories are often difficult to
distinguish both formally and functionally.  One can easily be misled
into thinking that they are in free variation: Willmott rightly
stresses that this is not the case, as the two categories are not
distributionally identical.  Rather, the future is restricted in its
functional load, while the subjunctive has a broader array of uses.  To
illustrate this point, she observes that the future indicative in
conditionals is restricted to resumptive conditionals (conditionals
that reintroduce a piece of information from the linguistic or
non-linguistic context).

The Optative

The optative is conventionally considered the most irrealis of the
moods, and also often considered to be a past-time variant of the
subjunctive.  Willmott replaces this view with an optative that
expresses negative epistemic stance in conditionals, and elsewhere
marks unreal events.  As a marker of unreal events, it is thus
timeless.  Her results on this front are especially interesting when it
comes to the expression of wishes.

Willmott also shatters much conventional wisdom about the subjunctive
and optative in subordinate environments, such as purpose clauses and
non-specific (or "general") relative clauses.  In the handbooks (based
as they are on Attic Greek of the classical period), mood usage is
essentially a grammatical rule: after primary-tense verbs, a
subjunctive is obligatory in the subordinate clause; while after a
secondary-tense verb, an optative.  Despite robust evidence in support
of this generalization, there are many counterexamples; these are
described as "breaking sequence." Willmott shows, on the contrary, that
the appearance of the subjunctive or optative (in Homer, at least) is
not dictated by grammatical rule, but rather by the semantics of the
main clause.  Indeed, one can really see from Willmott's work just how
different Homeric and Attic are in terms of degree of
grammaticalization.  The semantic conditioning that Willmott is
describing for the moods can also, for instance, be found among
agent-expressions with passive verbs,[[2]] as well as purpose-clause
markers.

Mood through the Lens of Grammaticalization

The framework within which Willmott comes to her conclusions is that of
grammaticalization.  This is a subfield of linguistics that was
initiated by Meillet;[[3]] it has experienced a burst of attention in
the last twenty years.  Its focus and scope, as traditionally defined,
are the development of lexical items into grammatical items, as well as
the development of new functions in already grammatical items.  One
example of the first type was mentioned above, namely the change of
English will from a content verb to an auxiliary marking futurity.  A
similar example would be the change of body-part nouns into
prepositions (e.g. Hittite hant- 'forehead' and cognate prepositions
like Lat. ante and Grk. <greek>a)nti/</greek>[[4]]).  What Willmott is
looking at with the moods is how already grammaticalized forms take on
new semantic dimensions.  This includes e.g. looking at a verbal
category like 'subjunctive' and examining how it came to be used in
various embedded clauses, e.g. purpose clauses.  It also includes
looking at individual grammatical markers and how they develop, e.g.
temporal-clause markers also taking on purpose-marking functions.

Previous attempts to understand the semantics of Greek moods have
attempted to set up a core abstract meaning from which other meanings
are derived.  One assumption underlying this method is that language is
structural, and that grammatical domains may be neatly defined and
divided, with each individual marker in the system gaining its meaning
through opposition with other markers (e.g., indicative in opposition
to the "oblique" moods).  By contrast, grammaticalization claims that
such neat descriptions are insufficient, as language use is far
messier.  Rather than derive the various meanings of a grammatical form
from some abstract essential meaning, grammaticalization sees the
different uses as stages along a trajectory of development (p. 26):
grammatical markers, e.g. verbal mood, gradually develop and accumulate
new meanings.  Moreover, these trajectories are constrained by the type
of constructions in which the grammatical marker occurs, as well as its
implicatures (that is, broadly speaking, the non-linguistically encoded
meaning that is conveyed by inference).  Thus language change is of
central importance in understanding the synchronic diversity of meaning
in a grammatical form.

This amphichronic framework is a fruitful one, as it has yielded new
and exciting insights.  In practice, however, it creates some tensions
in Willmott's work.  For while Willmott has not set out to write a
diachronic study of the moods from PIE to Greek, or even from Homeric
Greek to Attic Greek, her framework nevertheless demands diachronic
investigation.  The result is that discussions on this front seemed
cramped.  I often found myself wanting explanations for why and how
certain changes occurred, e.g. the discussion of <greek>o)/fra</greek>
pp. 156-158, of purpose clauses on p. 159, and of iteratives on pp.
182-183.  With forms or constructions undergoing change, discussion of
frequency would also have been helpful, e.g. the number of tokens in
which <greek>o)/fra</greek> has temporal, purpose, or indeterminate
semantic values.[[5]] With <greek>o)/fra</greek> in particular, I
wonder whether there is a connection to make between its frequency
(which, impressionistically, does not seem that high) and its
restrained development: for while <greek>o)/fra</greek> became a
purpose marker, its use as such was constrained (apparently by the
semantics of its matrix verb phrase), and it never exhibited the same
kind of development as e.g. <greek>i(/na</greek>.

The Semantics-Pragmatics Interface

The focus of Willmott's study is the semantics of the moods, and her
results from this angle are impressive.  At times, however, I think
more could have been achieved with greater attention to pragmatic
meaning.  That is, to look not only at the propositional meaning of a
sentence, but how a speaker might manipulate the implications of such
meaning.  For there are cases when it seemed that data that were
inconsistent from a semantic angle were not from a pragmatic one.
Moreover, the semantics-pragmatics interface is in some ways crucial to
grammaticalization.  Frajzyngier[[7]] has recently argued that indirect
expression in interpersonal communication is one of the basic driving
forces of grammaticalization.  To be sure, Willmott is aware of the
importance of pragmatic implicature; my point is only that greater
consideration of the interface between semantic and pragmatic meaning
could have strengthened her analyses.  I present three examples below.


Consider the aorist indicative in the protasis at Il. 1.503, discussed
by Willmott on p. 42:

<greek>*Zeu= pa/ter, ei)/ pote dh/ se met' a)qana/toisin
o)/nhsa</greek>
'Father Zeus, if ever I helped you, among the immortals...'(trans.
Willmott)

Here Thetis is speaking,[[6]] and, as Willmott rightly notes, she did
in fact help Zeus.  As such, the protasis appears to be a factual
statement (closer to a causal clause: 'since I helped you...') rather
than a real conditional.  Willmott claims that Thetis does not want to
appear to be bartering her previous duty for a favor from Zeus, and
therefore uses a conditional.  Accordingly, Willmott labels the usage
here a "rhetorical device" (ibid.) and a "stylistic choice" (ibid.).
Both of these observations are probably correct (even if it is not
entirely clear what these labels are supposed to mean), but I thought
that a better motivated account of Thetis' language could be derived
from a pragmatic angle.  By using a conditional here, she explicitly
disavows recognition of having done favors for Zeus in the past, but
implicitly both speaker and addressee know that she has.  As such, the
indirectness seems more like a face-enhancing or assertion-muting
strategy of a speaker with less power attempting to acquire something
from a listener with more.

Also, I am skeptical of the claim that Thetis wants to avoid the
appearance of bartering with Zeus.  For this is what she is doing, even
if the bartering takes the expression of a conditional and not an
outright causal clause: that is, to say 'if I did X, do Y' is still to
barter, irrespective of whether the conditional is true or not.
Moreover, given the overwhelmingly do ut des nature of Mediterranean
religions, to set Thetis up as wanting to avoid the appearance of
bartering may be anachronistic.

On p. 63, in discussing subjunctives found where we might have expected
futures, Willmott cites the following from Agamemnon (Il.1.184-185):

<greek>th\n me\n e)gw\ su\n nhi+/ t' emh=i kai\ e)moi=s
e(ta/roisin</greek>
<greek>pe/myw. e)gw\ de/ k' a)/gw Brishi+/da kallipa/rhion</greek>

'Her I'll send back in my own ships [sic], and with my own crew. But
I'll take Briseis, in all her beauty.' (trans. Willmott)

Willmott uses the example to argue that subjunctives and future
indicatives cannot be distinguished according to a certainty parameter.
Prima facie Agamemnon's use of <greek>e)gw\ de/ k' a)/gw</greek> seems
to support this claim: the event of transporting Briseis is entirely
under his control and he has every reason to regard the event as
reasonably certain.  However, if we imagine that the future indicative
and subjunctive do differ in terms of speaker-certainty, then it is
possible that the subjunctive, encoding less certainty, might be
exploited for pragmatic purposes, here to mute the assertiveness of the
suggestion by presenting a projected reality as a potential reality.
We might compare the use of English 'could' here to talk about
potential events that are not dependent on the ability of the speaker.

We find another scenario of this type on pp. 125 and 130, in a
discussion of the optative <greek>lu/saite</greek> at Il. 1.20:

<greek>pai=da d' e)moi\ lu/saite fi/lhn, ta\ d' a)/poina
de/xesqai</greek>

'Release my darling child, and accept this ransom' (trans. Willmott)

Willmott draws the following semantic distinctions between wishes
(which we have here) and imperatives.  With imperatives, the agent of
the action is addressed, and the speaker has some control over the
addressee or agent.  In wishes, the agent of the action is not
addressed, and the speaker has no control over the addressee.  This
example does not fit neatly into Willmott's scheme because we have a
wish in which an agent is directly addressed.  Here she rightly points
out, as others have before, that the use of the optative is a
politeness tactic, in that it enables the speaker to avoid directly
asking (or ordering) the addressee to do something.  But the analysis
could, I think, have been stronger if she had shown how this politeness
technique results from a manipulation of the semantic scheme.  That is,
in using the wish construction, Chryses frames the situation as if he
has no control over his addressee, and hope is his only recourse.  This
is presumably a deference-tactic.  Likewise, although his message is
directed toward the agent of the desired event (namely Agamemnon),
Chryses addresses the Greek host at large (Il. 1.15-16).  In doing so,
he further distances himself from a direct request over a particular
agent.  Thus, an example like this seems to me not to be a
counterexample, but rather another example of how semantic resources
can be pragmatically manipulated.

Consider again in this light the use of the indicative in
counterfactuals.  Here the propositional value of a protasis (e.g. 'If
I had gone to the store') suspends truth on a propositional level, but
denies it a pragmatic level.  We observed the opposite effect with our
Thetis-example above: explicitly a proposition was suspended, but
implicitly it was affirmed.  More attention to the pragmatics of modal
constructions will surely enhance our understanding of
grammaticalization patterns, texts, and Greek sociolinguistics more
broadly.

Sundry and Small

Translations of the Greek are usually satisfactory, but sometimes
Willmott takes unexpected turns with the modality, and one is not often
sure why, e.g. future <greek>e)fh/sei</greek> on p. 76 is rendered with
'could.'  On p. 185, <greek>ei(/ponto</greek> is translated 'would
follow.' On p. 186 she translates <greek>*Trw/wn</greek> as 'of the
Trojans.' The addition of the definite article would normally not be
cause for concern, but here in a discussion of antecedent specificity
more care should have been exercised.

On p. 80, I doubt that the point of <greek>tenw=</greek> will be clear
to the typical classicist, and its historical development should have
been spelled out to clarify its relationship to a desiderative *-h1s-
morpheme, i.e. *ten-h1s-w > ten-es-w > tenew > <greek>tenw=</greek>.

On p. 200, in her discussion of the semantic contribution of
<greek>a)/n</greek>, Willmott mentions one possible etymology, namely
that it originates from a reanalysis of the string <greek>ou)
ka/n</greek> into <greek>ou)k a)/n</greek>; and that it descends from
the same form as <greek>ke(n)</greek>, which is derived also from
earlier <greek>ka/n</greek>.  Here Willmott should have at least cited
the competing analysis, which holds that <greek>a)/n</greek> and
<greek>ke(n)</greek> are not historically related, and that the former
is in fact cognate with Latin an and Gothic an.[[8]] This is a
plausible view, and if it is true, then the functional overlap of the
two particles could not be ascribed to the same diachronic source.

In laying out semantic maps of the various moods, it would have been
helpful to provide visual maps of the various meanings and their
relationships, as is done elsewhere in studies of grammaticalization.
This would have been especially helpful with the subjunctive, where the
use and semantics of the category are especially rich.

Lastly, a word about the presentation of the Greek.  If classical
linguists intend for their work to be read by general linguists (as
they should), they will need to adopt the conventions of modern
linguistics and transliterate the Greek as well as offer word-by-word
glosses, which this book does not do.  Subjunctive and optative forms
are underlined, which is helpful, but I doubt that for a Greekless
linguist this is sufficient.

Final Thoughts

In sum, this work has superseded much that has come before it and set a
new benchmark in the analysis and description of Greek modality.  It is
the most theoretically sophisticated and empirically detailed account
of the Greek moods that I am aware of.  This book is thus a welcome
addition to a growing body of work by young classicists revealing just
how inadequate the traditional grammars are, especially when it comes
to semantics and pragmatics.  These areas are ripe for more exciting
discoveries in the coming years.


------------------
Notes:


1.   See F.R. Palmer, Mood and Modality, 2nd ed. (Cambridge 2001): 3-4.

2.   On the diversity of agent-marking with passives in Homer, see C.
George Expressions of Agency in Ancient Greek (Cambridge 2005): 62-65,
68-69.

3.   A. Meillet, "L'e/volution des formes grammaticales," Scientia
(Rivista di Scienze) 12/26 (6): 384-400.  The essential idea behind
grammaticalization had existed well before Meillet, however, in the
work of Bopp, von Humboldt, and others; see e.g. C. Lehmann, Thoughts
on grammaticalization: A programmatic sketch (Cologne 1982) and B.
Heine, U. Claudi, and F. Huennemeyer, Grammaticalization: A conceptual
framework (Chicago 1991).

4.   See J. Puhvel, Hittite Etymological Dictionary: Volume 3, Words
Beginning with H (Berlin/New York 1991): 89-96; A. Kloekhorst,
Etymological Dictionary of the Hittite Inherited Lexicon (Leiden/Boston
2007): 287-289. This is a cross-linguistically common development: see
J. Bybee, R. Perkins, and Willmott Pagliuca, The Evolution of Grammar:
Tense, Aspect, and Modality in the Languages of the World
(Chicago/London 1994): 10-11.

5.   The absence of frequency in Willmott's discussion is surprising
given that it generally plays such a prominent role in
grammaticalization studies. See for instance Frequency Effects and the
Emergence of Linguistic Structure, ed. J. Bybee and P. Hopper
(Amsterdam/Philadelphia 2001), or J. Bybee, "Mechanisms of change in
grammaticization: The role of frequency," in Handbook of historical
linguistics, ed. R. Janda and B. Joseph (Oxford 2003).

6.   Willmott misidentifies the speaker here as Chryses.

7.   Z. Frajzyngier, "Grammaticalization, typology and semantics:
Expanding the agenda," in Rethinking Grammaticalization: New
Perspectives, ed. M.J. Lo/pez-Couso and E. Seoane
(Amsterdam/Philadelphia 2008): 61-102.  On meaning change (both
semantic and pragmatic) and grammaticalization specifically, see also
Regine Eckardt, Meaning Change in Grammaticalization: An Enquiry into
Semantic Reanalysis (Oxford 2008).

8.   See e.g. H. Frisk, Griechisches Etymologisches Woerterbuch
(Heidelberg 1960-1972), s.v. <greek>a)/</greek>.  Nicholas
Sims-Williams informs me (p.c.) that he is preparing an article with
Elizabeth Tucker on <greek>a)/n</greek> from a comparative
Indo-European point-of-view.  This article will include new data from
Middle Iranian that has not been included in past discussions.