[Lexicog] Forest fires and wood fires

[Ron Moe]

I think we are right to reject introspection as a sufficient basis for
lexicography and to look at documented usage. But I have also been struck by
the usefulness of looking at the real world, 'exospection' if you will.
There are aspects of the real world that limit and define a domain. I would
argue that culture and perspective are sometimes a bit overrated. Fires
exist in the real world and we can identify real world aspects of them. They
need fuel. So it is no surprise that cultures and languages talk about
'fuel' and sometimes incorporate a particular fuel into a lexical item in
the domain 'Fire'. What will be culturally determined is what fuels are
available/used for intentional fires. Cow dung can be used as fuel, but some
cultures may find this abhorrent. Here is my list (partial and tentative) of
real world features of the domain 'Fire':

1. Fires occur in the physical world and have a location: campfire (camp)
2. Fires require fuel: wood fire (wood)
3. Fires require oxygen: smother (cover a fire to extinguish it)
4. Except in rare conditions, fires do not start by themselves. So we can
talk about their cause: arson (person), electrical fire (electricity)
5. Some fires are started by people and some are not. So we can talk about
intentional or accidental fires: bonfire (intentional), blaze (accidental)
6. Fires extend over time and go through stages: light a fire, flare up, die
down, die
7. Fires become larger, so we can talk about their size: flicker (small),
inferno (very big)
8. Fires produce resulting states that all cultures have learned to exploit.
So we can talk about their purpose: arson (criminal intent), cook fire
(cook), bonfire (burn rubbish)
9. Because the resulting states can hurt people or help people, we can talk
about good and bad fires: wood fire (nice, cozy), arson (bad), forest fire
(destructive, dangerous)
10. If we combine the ideas that people use fires and that fires grow, we
get the idea that people need to control the growth of fires. So we talk
about controlling a fire: controlled fire (controlled), forest fire
(uncontrolled), tend a fire (control it)

With this list I am attempting to identify the universals of fire. My claim
is that none of them is culturally determined. What is culturally determined
is the details and what is lexicalized. For instance different cultures will
employ different means to start a fire (match, flint and steel, rubbing two
sticks together). Languages pick and choose what to lexicalize. English has
a word 'forest fire', other languages do not. We say, 'set a match to', but
not '*matchify'. Another language may use their equivalent of 'match' as
both a noun and verb. Cultures can also ascribe features to a domain that
are peculiar to their world view. For instance fire may be seen as a living
thing. (We talk about fires 'growing' and 'dying', although we would claim
that this is metaphorical.) However we would not expect such features to be
universal. So there are both universal and culturally determined features of
a domain.

If we did a corpus analysis of all of the words in the domain 'Fire', we
would find each of these features in the context. However when we look at a
particular word, we will find that, for any given feature, the options may
be limited. Patrick Hanks pointed out that in the BNC 'wood fire' is limited
to three locations (out of all possible locations): house/hearth, outdoor
camp, locomotive. Or a specification of a feature may be included in the
meaning of the word. *It was a small inferno. *It was an accidental bonfire.
We determine the meaning of a word by looking at what features (and
specification of features) occur and which do not. It is easy to determine
what does occur. It may take introspection to determine what cannot. But in
either case it helps to know what features to look for.

When I do corpus work, I look in the context for clues as to what domain(s)
the word belongs to. Once I identify the domain, I can assume that certain
features will be present. I look for these features and list the specific
values (collocates) that occur. Then I analyze these for patterns.

Ron Moe
SIL, Uganda
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lexicography/attachments/20040205/ac95f802/attachment.htm>