These are my final comments on Peter Constable's "Issues to resolve in
ISO 639."  The languages discussed here were included in sections 7 and
8 of the document.  I apologize for the length of time this has taken.
Many of these issues are very challenging.
Milicent Wewerka, Library of Congress


Akan/Fanti/Twi.  According to my experience with these terms, this is a
case where the ethnic terms do not correspond to the linguistic
situation.  Although there is apparently only one language, there is an
ethnic split between the Fanti and the other groups who recognize Twi as
an acceptable name (primarily the Ashanti and Akwapem).  This appears to
be similar to the Karachay-Balkar situation (or Serbo-Croatian?) where
there are two ethnic groups with different names, but the groups share a
single language that lacks a mutually acceptable name.  I believe the
government of Ghana decided to use Akan as the term for the language and
the groups collectively.
The proposed solution to use [tw/twi] for all varieties of the language
would include the language of the Fanti under a term they do not
recognize as applying to their language.  If we must combine all
varieties under one designation, I think [aka] would be better.  There
is also the possibility that we might consider [aka] as the designation
for the Akan language and recognize [fat] and [tw/twi] as representing
specific dialects of the Akan language.  The ISO standard does recognize
the possibility of including separate entries for dialects in 639-2.

Serbo-Croatian.  If Serbian, Bosnian, and Croatian are recognized as
separate languages, then documents, Web sites, etc. should be identified
as being in one of these languages.  How then would a macro-language
entry be used?  Would it be used in those cases where the text cannot be
identified as a specific language?  Some librarians have had difficulty
in recognizing these languages, but I am told that native speakers
normally have no such difficulty.

Moldavian/Romanian.  Of the various options presented it seems
undesirable to follow option 2 or 4.  Option 2 would allow [mo/mol] as a
synonym for [ro/ron/rum] without deprecating the combination for
Moldavian.  If we agree that there is only one linguistic entity, then
there should be only one set of entries for the language and one set
should be deprecated.  Option 4 recommends allowing a script
distinction.  I think that is a bad precedent.  One possibility that was
not presented was to consider Moldavian a dialect of Romanian.  There is
some support for this treatment.  Voegelin & Voegelin list Moldavian as
equivalent to "Eastern Daco-Rumanian" with standard Romanian being
equivalent to "Southern Daco-Rumanian."  Ethnologue also lists
Moldovan or Moldavian as a dialect of Romanian.  This treatment would
allow the continuation of both Romanian and Moldavian in the ISO code.
However, it is difficult to distinguish the official written forms of
language from Romania and Moldova.  This makes the application of the
code problematic.

Turkish/Ottoman Turkish.  A proposal to eliminate the distinction for
Ottoman Turkish was debated for the MARC list some years ago.
Convincing arguments were made at that time that the 1928 language
revisions were more extensive than script, extending to vocabulary and
other features.  For library usage, option 2 would be more in keeping
with this position.  Option 2 would treat Ottoman Turkish as a distinct
historical variety of Turkish.  This would be similar to other instances
of historical varieties in the code list, although the date of
linguistic change would be more recent than other cases, which have
dates of 1500 or other earlier times.

Central American Indian/North American Indian.  I am reluctant to
redefine MARC usage here because of the disruption to past files in
libraries.  While it seems strange to include some U.S. languages under
"Central American," certain linguistic classifications support the
inclusion of the Tanoan languages with languages in Mexico.  Merritt
Ruhlen treats Tanoan, Uto-Aztecan, and Oto-Manguean as members of
"Central Amerind."

Papuan.  I agree with the proposal to define Papuan as all
non-Austronesian languages of the New Guinea region.  This corresponds
to the intention of the MARC list.

South American Indian.  The intention of the MARC list is that this
group is purely geographical.  Unlike the Central American Indian
grouping, which includes a specific family regardless of location, the
South American Indian list should include only those languages used in
South America.  However, there are a few languages that are spoken in
both Panama and Colombia.  In those cases MARC practice has assigned the
language to [sai].  This situation also arises with [cai] and [nai] for
languages outside "Azteco-Tanoan" that are used both in Mexico and
the United States.  A decision must be made and recorded for each