Using Dictionaries (was Re: Greek question (night?))

[Robert Whiting]

On Sat, 13 Mar 1999 X99Lynx at aol.com wrote:

>In a message dated 3/11/99 3:34:25 PM, you wrote:

><<[ Moderator's response:
>>  Greek _nuks, nuktos_, Latin _nox, noctis_, Sanskrit _nak
>>  (IIRC), naktam_, Hittite _nekuz = nek{^w}t+s_, ...
>>  --rma ]>>

>Does it matter that /t/ does not appear in most of the nom.
>singulars?  I also saw nocz in Polish, nos in Welsh, but was not
>given declension.  It seems the /t/ only makes its appearance in
>the nom sing in German, French and Sanskrit. Please forgive my
>ignorance, but sometimes in analysis on this list, this base
>morphology makes a difference.  Why is the /t/ not regarded as
>just a fairly common stem that sometimes emerges in root and
>sometimes doesn't?  I hope this doesn't sound terribly stupid.

>[ Moderator's response:
>  In those languages in which it does not appear, other processes
>  are at work, usually word structure constraints.
<snip>
>  --rma ]

What has been pointed out here is another pitfall of using
dictionaries to do comparative linguistics without knowing
anything about the languages that are being compared.
Dictionaries traditionally give the nominative singular as the
lexical entry.  In many instances the nominative singular is the
linguistically least-marked form.  Many times features of the
root that are important for comparative work are suppressed in
the nominative singular because of the phonotactics of the
language, but will still be found in the stem (the form to which
other case endings are added).  Thus using a dictionary without
being aware of the phonotactical rules of the language is a
recipe for disaster when doing comparative work.  Not only can
you get false positives by comparing two similar looking forms
that may have resulted from the suppression of entirely different
elements from the stem, but you can also get false negatives from
words where one language has suppressed or mutated a stem element
and another hasn't (e.g., Lat. <nox> - Ger. <Nacht>).  Similarly,
if you don't know that the stem of Fin. vesi (the dictionary
entry) 'water' is vete- you are likely to miss the connection
with IE *wat-/*w at t-.

So doing comparative work by looking words up in a dictionary is
just a matter of collecting "lookalikes" that have similar
meanings.  Some may be valid comparisons and some not; but if
they are, it is more a matter of luck rather than skill since
most dictionaries simply don't tell you everything that you need
to know to do comparative linguistics beyond this simplistic
level.

Bob Whiting
whiting at cc.helsinki.fi