Minutes of the Second Meeting
 
           Of the Text Analysis and Interpretation Committee
 
                    Of the Text Encoding Initiative.
 
 
                         D. Terence Langendoen
 
                       Document Number:  TEI AIM2
 
                             April 23, 1990
         Olympic Room, Hyatt Regency Hotel, Washington, DC, USA
                Friday, 29 December 1989, 09:00 to 16:45
 
Present:   D.   Terence Langendoen  (chair),  Stephen  Anderson,  Robert
Ingria, Mitchell Marcus, William Poser, Beatrice Santorini,  Gary Simons
Committee),   Michael Sperberg-McQueen  (TEI  Steering   Nancy Ide  (TEI
Steering Committee  and Editor),  Sandra  Fulmer (Research  Assistant to
Langendoen, who took notes on the discussion)
 
                         Final, April 23, 1990
 
 
The discussion ranged  over the types of linguistic  analysis and inter-
pretation that should be encoded.   It was agreed that we should be aim-
ing at an encoding scheme that is capable of representing detailed anal-
yses of particular  examples within any theoretical  framework for which
linguistic  analysis can  be  expressed in  the  form of  interconnected
directed acyclic graphs (essentially any  current theory).   Whatever we
come  up must  also reduce  to reasonable  and easily  used schemes  for
encoding simple aspects of linguistic analysis by hand, such as part-of-
speech labeling, marking of syllabic structure, decomposition of complex
words into constituent morphs, etc.
 
Much of the  discussion concerned problems of  morphological markup,  as
"test cases",  such as the interleaving  of consonant root elements with
vocalisms in  a 'binyan'  as in  Semitic forms  such as  'katabti' (root
'k...t...b', vocalism 'a...a' (possibly just 'a', which "spreads" to the
two vowel positions  in the binyan CVCVC),  suffix  'ti'),  ablaut rela-
tions, as in English 'sing' vs.  'sang', and morphological deletion,  as
in the analysis  of the Danish imperative that  Steve Anderson provided.
Steve argued that the full encoding scheme must have the power to repre-
sent entire morphological derivations;  this  idea was endorsed also for
phonological and syntactic  markup,  and some discussion  was devoted to
the question of how to represent derivations in SGML.
 
Gary Simons argued that each subcommittee needs to identify the types of
data structures  that are  needed in their  respective domains,   and he
agreed to circulate a general statement  about data types for linguistic
analysis by mid-January.   The subcommittee  heads (Poser for phonology,
Simons for morphology,   Marcus for syntax and  Amsler for dictionaries)
are also responsible  to complete and circulate by 31  January the first
draft of  the statement of needed  data structures for  their respective
areas.
 
 
                                                   Final, April 23, 1990