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

[Robert Whiting]

[ moderator re-formatted ]

On Sat, 20 Mar 1999 X99Lynx at aol.com wrote:

>In a message dated 3/15/99 5:22:43 AM, Bob Whiting wrote:

<snip>

>But my question was specifically about the way the word was being
>reconstructed in the posts on the list.  I looked back again and
>for the most part the nominative (in those languages where it
>occurs) was not mentioned.  I now attribute that to the knowledge
>of those involved who all knew but did not say that (in the case
>of this "night") the nominative form was irrelevant - based on
>the phototactics of the various languages mentioned.

>I wrote:

><<The answer that both you and our moderator have given is that
>they are I guess "truncated" forms,...>>

>You replied:

><<"Truncated" is not really the right term.  The disappearance
>of /t/ in this position in both Greek /nuks/ and Latin /noks/ is
>simply the result of a phonotactical rule:..>>

>On closer consideration - with regard to "nos"(Welsh),
>"noc"(Pol), "nux"(Greek) and "nox"(Latin) - whatever the process,
>the result are all truncated.  No doubt about it.  I looked up
>"truncated" in the dictionary and it hits the nail on the head.

Truncated generally means "cut off" but usually not "cut out" so
whether these forms are "truncated" or not depends on how one
views the process.  The two possible views more or less
correspond to the difference between synchronic and diachronic
grammar (or generative versus historical).  Given the stem nokt-,
the generativist would say that the nominative ending -s is added
to a truncated stem *nok-; the historical linguist would say that
when the nominative ending -s was added to nokt-, the resulting
cluster was simplified by assimilation followed by degemination.
In both cases the result is the same:  the nominative form is
noks.  But whether the historical reconstruction corresponds to
any kind of reality depends on whether there was ever a stage of
the language where *nokts was an acceptable form.  If this form
was never allowed to exist then the historical reconstruction is
just an abstraction, because the generativist knows that no
speaker of the language creates a form *nokts and then simplifies
it to noks.  Noks is the first form of the nominative that the
speaker calls up even though the same speaker knows that the stem
is nokt-.

><<The phonotactical rules that produce <nox> are not operative
>for <nocturnus>.>>

><nox> came out of */nokt-s/> */ts/ > */ss/ > /s/ "by a normal
>phonological rule" - how is <nocturus> (adj) reconstructed?

Nocturnus is created from the stem nokt- plus a suffix -urnus,
an adjectival formative relating to time.  No cluster reduction
rules apply.

><<It's not a question of "solving the problem"; .... It is a
>matter of a tradeoff between ease of articulation and the level
>of morphological differentiation needed to disambiguate meaning
>that the users of the language resolve with even thinking about
>it.>>

>Sounds like classic problem-solving to me.  "Disambiguating" is
>definitely problem-solving.

This is too big a can of worms to open here.  It is more of a
philosophical problem than a linguistic one.  It revolves around
whether you can call it problem solving if you don't realize that
there is a problem to be solved.

>I wrote:

><<After all, to say that only the "ablative" form survived is to
>suggest that the word now carries the additional grammatical
>baggage of the ablative.>>

>You replied:

><<Sorry, but huh?>>

>In Romance languages like modern French, it is said that "in all
>but a few cases, the oblique ^Ö often the ablative ^Ö form
>survived the loss of Latin inflectional morphology,...while the
>nominative did not..."   I don't think I need to remind you that
>the nominative is "often" the least marked form.  The markings
>you refers to includes those related to the ablative as a
>"grammatical case expressing relations of separation, source,
>cause or instrumentality,... not found in the nominative."

>Voila.  The ablative's extra grammatical baggage.  In letter (and
>verse, if need be.)

I'm beginning to get the idea:  this is some kind of free
association test.  If you say stems are removed I have to
understand that you mean that affixes are removed from the stem.
If you say that the word now carries the additional grammatical
baggage of the ablative you mean that the ablative form now
has the additional grammatical baggage of all of the other
cases plus the ablative.

But you are not looking at the process of development, only the
end point.  And you are not taking all the developments into
consideration.  If we consider Latin with five cases (leaving the
vocative out), then the nominative is likely to be the most
frequently used case.  Most sentences will have at least one
nominative.  They may or may not have genitives, datives,
accusatives or ablatives.  So the simplest case marker (-s) gets
used for the nominative.  Adding this ending to the stem often
creates final consonant clusters that are usually resolved by the
removal of a consonant which often turns out to be an affix to
the root. Now as the language develops, genitives and datives
tend to be replaced by prepositional expressions with the object
of the preposition being in the accusative or ablative.  Finally
the accusative and ablative merge leaving only one oblique case.
Whether the surviving form is the accusative or the ablative is
not really important.  What is important is that there is now
only one oblique case and this is now the most frequently
occurring case (all of the four original oblique cases which were
individually less common than the nominative have now been
combined into a single case which is more common).  Furthermore
the case ending has usually developed into a short vowel so that
the oblique form is usually the full stem plus a short vowel
which is often no more heavily marked than the nominative (or
another way of looking at it is that as the most common form it
becomes the unmarked form by default).  So when case distinctions
are finally dropped, it is the oblique case (both as the most
common form and potentially the least marked) that survives.
WARNING:  This is an extremely simplistic explanation of a large
number of extremely complex developments.  Do not try this at
home. :)

><<Contrary to popular opinion, dictionaries do not define words
>(Academies do that, or try to) but only record usage...>>

>Time to send an angry letter to Noah Webster.  In the absence of
>an Academie, of course, dictionaries can and have "defined words
>according to their proper usage" or their common usage.  Contrary
>to non-popular opinion, dictionaries have had a powerful effect
>on usage and definition - as Mr. Webster's did.

Gee, I hate being quoted out of context.:)  My remark was a
comment to your statement that

     3000 year old words are often defined with a precision that
     - if you know your Homer, so to speak - can be as perilous
     as assuming that the nominative is worth talking about.

With dead languages there are only two clues to meaning: usage
and etymology.  Of these two, usage is the more reliable, but
if the word only occurs in limited contexts, etymology may be
important as well.  But there are no native informants to ask
about the meaning or usage.  Conversely the "dictionary
definition" cannot be considered prescriptive for speakers of
a dead language.  And a philologist is under no obligation to
accept the "dictionary definition" (many often do not) in her
translation.  So the precision with which such a word can be
defined is a function of how widespread and varied its usage
is and how well its etymology is known (and understood).

I will freely admit that Webster's is a prescriptive dictionary
and has arrogated to itself the function of an American Academy.
The prescriptions of Noah Webster are why Americans spell the
verbal suffix -ize rather than -ise and write favor rather than
favour, etc.  I find that Webster's is also prescriptively trying
to block the natural development of "all right" in some usages to
"alright."  It also sometimes simply ignores usage that is
incorrect according to its standards.  As an example, many people
use the adjective "fulsome" to mean 'tending to be complete'
considering the root of the word to be "ful(l)-."  But the
meaning of the word actually is influenced by ME "fu:l", the
ancestor of modern "foul."  So the word "fulsome" actually means
'disgusting', which is the only meaning that will be found in
Webster's despite the widespread folk-etymologized usage.

>There should be NO question that "usage" overwhelmingly says
>that the "definition" of a word is the "dictionary definition."

So why do you not use the "dictionary definition" of words like
"cognate" or "stem"?  I can find nothing in my dictionary that
says that "cognate" means 'synonym' or that "stem" means 'root
affix'.  If you so firmly believe that the "definition" of a
word is the "dictionary definition," why do you ignore it and
make up your own?

>And the dictionary says that the "definition" of a word is the
>"meaning of the word."  If you think "dictionaries do not define
>words..." your usage is so uncommon it has not been recorded in
>Webster's.

There are lots of things that are not recorded in Webster's.  The
f-word is one of them.  This does not mean that things that are
not recorded in Webster's do not exist.

>Being an American, I tend to go by Webster's.

I go by Webster's too, unless I want the answer to something
particularly difficult and then I use Chamber's.  But let me
offer you a quotation from Otto Jespersen, _Language.  Its
Nature, Development and Origin_ (London and New York, 1922) [old,
but still useful], p. 25:

     The vocabulary, too, was treated from the same point of view
     [as grammar].  This is especially evident in the case of the
     dictionaries issued by the French and Italian Academies.
     They differ from dictionaries as now usually compiled in
     being not collections of all and any words their authors
     could get hold of within the limits of the language
     concerned, but in being selections of words deserving the
     recommendations of the best arbiters of taste and therefore
     fit to be used in the highest literature by even the most
     elegant or fastidious writers.  Dictionaries thus understood
     were less descriptions of actual usage than prescriptions
     for the best usage of words.

So I will stick by the intent of my original statement:
Dictionaries should be descriptive of usage.  This of course
leads to a cycle.  If one doesn't know how the word is used
(whether one is a native speaker or not), one looks it up in a
dictionary to find out, and then uses it accordingly and thus the
dictionary becomes prescriptive.  The dictionary becomes an
"authority."  Hence the popular opinion that dictionaries define
words, held by those who haven't thought the process through.
But the word doesn't mean what it does because the dictionary
says so.  It means that because that is the way it is used.
When a new usage becomes widespread enough, it will be recorded
in the dictionary.  Webster's now lists a meaning of "gay" as 'a
homosexual person'.  Is the word used that way because Webster's
defined it so?  No.  Usage came first, then the dictionary entry.
That is the way that words change meaning.  ME "nice" meant
'ignorant, foolish' (comes through French from Latin ne-sci:re
'not to know').  Today the word has almost exactly the opposite
meaning.  The same thing will eventually happen to "fulsome" as
the ful- part (rightly or wrongly) becomes more firmly identified
with "full" rather than "foul."

><<When you get right down to it, nobody really knew his Homer
>except Homer.>>

>And when you get right down to it, nobody really knows your
>posted message but you.

Precisely.  And when you say "... one might naturally go to the
form stripped even of stems to get to the elemental form of the
word," you are the only one who knows that you are using "stems"
to mean "root affixes."  Now being a trained philologist, if you
use it that way in enough different contexts, I can eventually
figure out what you mean.  And this is precisely the way that
one defines 3000 year old words.

>Although I occasionally appreciate Zen and the aloneness of
>oneness and all that, this is obviously a bit too much.  We do
>know what Homer meant most of the time, and his intention was to
>be understood.  Language's #1 function is communication and Homer
>was damn good at.

Now when I compare this statement with your earlier quoted

     3000 year old words are often defined with a precision that
     - if you know your Homer, so to speak - can be as perilous
     as assuming that the nominative is worth talking about.

I have a hard time figuring out whether your point that there is
no way to define 3000 year old words with any precision or that
Homer was so good at communicating that we know exactly what he
meant and his words can be defined with precision.  Which
position are you actually supporting?

><<"I know what it says, but what does it *mean*?">>

>But that is a whole different can of worms, isn't it?

Not at all.  Writers, especially of poetry, often use words in
unusual ways both for expressive effect and because they are
constrained to a certain number of syllables or to a stress
pattern by meter and foot.  What the words say and what they mean
can often be two entirely different things.  Irony and sarcasm
often invert or shift meanings in ways that are difficult to
detect.  To illustrate this, I have a few examples that I have
gleaned from my experience with discussion lists of the
difference between what people say and what they often mean:

<humor>
What the writer says:                What the writer means:

in my humble opinion                 I know more about this than
                                     you do, so listen up and
                                     learn something

everyone's entitled to their         you don't have a clue what
own opinion                          you're talking about

I don't mind constructive            mind your own business and
criticism, but                       what do you know about it,
                                     anyway

I must have gotten it wrong          I'm sure I'm right, but I
                                     don't want to argue

with due respect                     thinking as little of your
                                     argument as I do

you and I both know                  you don't know, but I'm
                                     telling you

you will appreciate that             you won't like this at all

I really hate to say this, but       I am really enjoying
                                     pointing out this simple
                                     fact that you obviously
                                     don't know
</humor>

>Since the changing meaning or function of words can be quite
>independent of their structural linguistics.

Change "can be" to "is" and you have a basic principle of
language.  It is called "duality of patterning" and was listed
by Hockett as one of his design features of language.  In fact
it is probably the most important single one of these features.
Spoken language communicates meaning through sound, but the
individual sounds do not have any meaning themselves.  The sounds
are combined to created morphemes that do have meaning.  This
means that the myriad words of a language (all languages that
we know of) are built up by varying arrangements of a limited
number of sound units, themselves meaningless in isolation.
Thus a language can have tens of thousands of words built from
a remarkably small number of sound units (almost always less than
100, usually less than 50).

>When I read the word "gay" in an old novel, I am reminded that
>phonology cannot tell me how or why it came to mean what it means
>today.

Duality of patterning, especially the fact that the individual
sound units do not have any meaning themselves, is what allows
a word to change its meaning without changing its phonology or
to change its phonology without changing its meaning or change
both without the new meaning or new phonology necessarily being
predictable.

On the other hand, there is something called "sound symbolism"
which does hold that there is meaning associated with individual
sounds (universally), but investigations have yet to turn up
generally applicable principles.  Sound symbolism is of course
connected with onomatopoeia, but it may also go deeper than this.
Investigations are continuing, but far more descriptive data as
well as experimental investigations into speakers' intuitions
about the relationships between sounds and meanings are needed.

>I wrote:

><<...can a stem ever be seen as something like a vestigal case
>ending in reconstructing PIE - not part of the original word, but
>a compounded form that produced a universal "stem" in the
>daughter languages?>>

>You wrote:

><<If by "stem" you mean "root augment" this is a very complex
>question that has been discussed extensively without reaching any
>particular conclusion....The problem is that "root extensions"
>don't behave mathematically.>>

>So the answer is: ...maybe.

Definitely.

Bob Whiting
whiting at cc.helsinki.fi