Group 2: Sino-Tibetian Languages - E-Meld

Group 2: Sino-Tibetan Languages
Working Group II:
Sino-Tibetan Languages
Session Report
July 2, 2005
Parts of Speech: General Issues
1. Nature of definitions: semantically based
or morphosyntactically-based?
– Need to allow for language-specific
definition of classes based on
morphosyntactic criteria in that language
Parts of Speech: General issues
2. Consider separating out issues of
morphological boundedness and track as
a separate feature of lexical categories
e.g., Problems with classifiers; concepts
of grammatical vs. phonological word
possibly useful
Issues & Suggestions
3. Do we want to build implicational
universals into the ontology?
e.g., existence of dual number implies
existence of plural number
Issues & Suggestions
4. Do you want to put together a working
group that builds a particular COPE
based on common grammaticalization
patterns? In other words:
What is inventory of relations between
grammatical classes as exemplified in
historical change?
Numeral ‘1’ + classifier --> indefinite article
Verb ‘stay’ -> progressive aspect
3rd-person singular pronoun -> distal demonstrative
People will at times be wondering whether a particular element
is a X or Y. Is it an auxiliary verb or is it a lexical verb? Ideally
there would be one-many linking. (Necessary). As with other
cases of polyesemy, the status of a particular instance will be
determined by context (or even will be unresolved in certain
contexts) – how can you come up with a context-dependent
assignment of a particular mapping to the ontology when
multiple mappings are specified?
State that relation between two concepts in ontology are likely
to be expressed together by the same morpheme
Note that certain types of grammaticalization patterns do not
occur. E.g., as far as we know, no attested examples of first- or
second-person pronouns grammaticalizing into distal
Issues & Suggestions
5. Comments on various parts of speech:
Coordinating conjunctions
Serial verb constructions, auxiliary and
– Interjections, Ideophones
Summary of Basic Points from
Saturday (a.m. meeting)
• Nature of definitions: semantically-based or
morphosyntactically based
• Recommendation that morphological
boundedness be treated independently of other
• Discussed building implicational universals into
the GOLD architecture
• Recommended development of COPE based on
grammaticalization patterns
• More detailed recommendations on particular
parts of speech
What to do with “hybrid” classes?
• Manange (T-B language, Nepal)
– Lexical class of “verbal adjectives” or “adjectival
verbs” in addition to well-defined verb and adjective
– Morphology of verbs but syntax of adjectives
– How do we treat this in GOLD
New class?
-- Category “other”?
Intersection of classes?
Map to both (but not a super-class)
Define by relation to other classes (relation: “morphology of
X”, relation: “syntax of Y”
• Need an ontology of constructions
• For example: ontology of constructions of
verb/“VP”/clause sequencing
Serial verb constructions
Verb-Auxiliary constructions
Converb/clause-chain constructions
• All syntactically distinct but cover overlapping
functional domains
• Demonstrates the need for separation of formal
and functional levels, with links between them
• Complex structure
– Has phonological, morphological, syntactic,
and semantic dimensions
– Need to be able to link between these
different levels
– Certain meanings common, e.g., iterativity,
durativity, plurality, distributive
– Do we want to predefine a “reduplicant” with
pre-association to the meanings? Open World
Assumption allows other relations to occur
• Semantic categories look sufficient, but
possibly add viewpoint and situational
• Must independently specify the meanings
and formal devices by which they are
– Serial verbs, auxiliary verbs, affixes, particles,
tones, constructions
– Any could potentially map to any meaning
More on form-function
• Need mechanisms to link between
features and form
• Helpful to have a list of tendencies of
which type of feature links to which type of
• Form-function independence was
repeatedly mentioned, for example,
evidentials can be realized by many formal
Discussion of larger theoretical
To what extent is the goal to produce a description
of what is possible in natural language or to provide
interoperability between different language
Are there tensions between these different goals?
Is GOLD a metalanguage for doing linguistics or a
metalanguage for finding language resources?
How theory-neutral is GOLD?
Didn’t resolve these questions, but feel they are
important to be asked iteratively in the process of
developing GOLD
Languages carve up functional
space in slightly different ways
• How can we allow for this?
• Ergative case in one language is not
necessarily used for the same set of
expressions in another language
• Discussed various ways of handling this
– Prototypes
– Templates
– Following diagram represents one
Prototype or template
Agent of
Ergative Case_Class
What ERG
means in