134
John B. Haviland
speakers, often not fully fluent. My own work on the now defunct Barrow
Point language (see Haviland 1998) is a minor example. In such cases,
wordlists reflect serendipitous opportunity more than systematic planning,
and coverage is spotty, based on happenstance and luck. Nonetheless, even
haphazardly assembled lists of words may be significant when political
processes – for example, “native title” claims to traditional Aboriginal terri-
tory – use linguistic evidence to establish links between land and Aborigi-
nal culture and society (Henderson and Nash 2002). Everything from a
place name to a plant name may turn out to have unsuspected relevance.
Thus the issue of coverage is less a matter of scientific “completeness” than
an ideological issue of clear political import, another matter to which I re-
turn fleetingly at the end of the chapter.
There is also a pedagogical tradition in dictionary making, source of the
most common dictionaries: those used by students to look up unfamiliar
words, or by tourists to translate menus. Here the question of dimension is
telling. Dictionaries of Mexican Spanish (for example, Lara Ramos 1986)
are explicitly graded by size: a small version meant for schoolchildren with
several thousand “basic” words, a larger intermediate version with more,
and so on. All celebrate Mexican Spanish, the most widely spoken variety
of the language, but one relegated to a subsidiary status by the language
academy of the colonial home country. The lexicon chosen and the facts of
usage are drawn from a huge corpus of Mexican textual material, from let-
ters, to newspaper articles, to popular songs. In Chiapas, the government
has similarly commissioned a variety of “diccionarios de bolsa” or pocket
dictionaries for the Indian languages of the state. These, along with a series
of grammatical sketches, are meant as both pedagogical tools and political
trophies, evidence of government concern for Indians in the wake of the
Zapatista uprising of 1994. Of a similar design but with an opposite ideo-
logical thrust are the illustrated school primers, or basic wordlists, designed
as literacy aids by Zapatista community schools which resist all govern-
ment aid and standardized school materials.
2. Referential indeterminacy and other pitfalls of fieldwork
What sorts of creatures are the “meanings” of words we wish to set down in
a lexical database? It is hard to escape the weight of many centuries of
Western philosophizing on the subject (although there are useful antidotes
in J. L. Austin’s early essay “The meaning of a word” in Austin 1961).
Chapter 6 – Documenting lexical knowledge
135
Following Frege (1892) it is customary to begin with the notion
that words
(characteristically nouns) can typically be used by speakers to pick out enti-
ties in the world – the words’ “referents” – by virtue of their “sense” or
“denotation” independent of any instance of their use for referring or predi-
cating about a specific state of affairs. Words, on this view, are a kind of
instruction from speaker to hearer, grounded in some shared understanding
of the “meanings” of expressions, and typically designed to achieve com-
mon reference.
Even with apparently simple cases, of course, the conundrums of refer-
ence as a theory of meaning immediately surface. Suppose someone wants
to refer to me as I am lecturing. Consider the following expressions she
might use:
(1) Expressions referring to the same referent
a.
That guy (with a pointing gesture)
b.
The linguistics professor from Oregon.
c.
The tall guy with a black moustache at the front of the room.
d.
The Mexican with a black moustache at the front of the room.
The speaker’s “instructions” if successful – that is, if they induce the inter-
locutor to pick me out as the person to whom she refers – rely on quite dif-
ferent sorts of relations to the “meanings” of the words she uses. The first
relies on some sort of categorial understanding of what we can use ‘guy’ to
refer to, combined with two direct indexical devices, the deictic that and
the pointing gesture. At the other extreme, (b) picks out a presupposably
identifiable individual from the intersection of sets of denotata generated
compositionally from the constituent words (along perhaps with presuppo-
sitions of existence and uniqueness built into the definite article the). Ex-
pression (c) combines such a compositional strategy with some implied
deixis (calculating which room and where its front is), and (d) paradoxically
is likely to succeed as well as (c) despite the fact that, though I live and
teach
in
Mexico
and
possibly
even
look
Mexican,
I
am
not
a
Mexican
at
all –
therefore, the “meanings” of the constituent words cannot
add up to a true
denotation.
So reference, although it is where we start in field linguistics, cannot be
where we want to end up. Quine’s famous
gavagai example (Quine 1960) –
in which a hypothetical and ontologically challenged linguist, in a parodied
setting of monolingual fieldwork, hears the word gavagai in the presence of
rabbits, but cannot decide whether the word means ‘rabbit’ or ‘rabbit part’