Convergent Grammar

[convert into an essay or review article at Linguistic Exploration]

CVG Course at ESSLI 08, Day 1 Slides

(23) convergent grammar (CVG): a Look Ahead

• CVG is closely related to both ACG and HPSG.
• Like ACG—but unlike other frameworks descended from EMG—
CVG uses Curry-Howard proof terms (which we will explain)
to denote NL syntactic entities.
• This makes it easy to connect CVG to mainstream generative
grammar because the proof terms are really just a more precise
version of EST/GB-style labelled bracketings.
• Like HPSG—but unlike other frameworks descended from EMG—
the relation between syntax and semaentics is not a function, but
rather is one-to-many.

Curry-Howard Correspondence

(30)
The basic ideas of CH are that, if you let the atomic formulas be
the types of a TLC, then
1. a formula is the same thing as a type.
2. A formula A has a proof iff there is a combinator (closed
term containing no basic constants) of type A.
• Hence the Curry-Howard slogan:
formulas = types, proofs = terms
(34)
• Variables correspond to hypotheses.
• Basic constants correspond to nonlogical ax-
ioms.
• Derivability of Γ ⊢ a : A corresponds to A being
provable from the hypotheses in Γ.
• Application corresponds to Modus Ponens.
• Abstraction corresponds to Hypothetical Proof.

(31) Notation for ND Proof Theory

• An ND proof theory consists of inference rules,
which have premisses and a conclusion.
• An n-ary rule is one with n premisses, and a
0-ary rule is called an axiom.
• Premisses and conclusions have the format of a
judgment:

                        Γ ⊢ a : A

read ‘a is a proof of A with hypotheses Γ’.

Autonomy of syntax is possible, but not at the granularity of word strings. The syntactic parallel to semantic hypotheses must be both words and constructions.

A deterministic functional interface from syntax to semantics seems to be less of a fit to real language than a non-deterministic relational interface. [find slide to quote]

"variables correspond to hypothesis" — does this mean the granularity is that established by referential indexes? Does it make sense to consider a finer granularity? Does this tie up with DRT?

Perhaps we retain the granularity of referential indexes, but consider implicit relations contributed by the constructions (and also by the individual words). This explains why schema instances in working memory have the granularity they have. There may be some finer granularity less accessible to consciousness, but the level of folk semantics demands an explanatory account.

(38) ND-Style Syntax

• The inference rules are the syntax rules.
• The formulas/types are the syntactic categories.
• The proofs/terms are the syntactic expressions.
• The basic constants are the syntactic words;
• The variables are traces.
• The context of a judgment is the list of traces still
unbound at that point in the proof.

This slide confirms the granularity of variables as referential indexes or traces. Can at least some of the syntax rules (for phrase structure) be considered as types and hypotheses from the lexicon, rather than inference rules for constructing results? You still need inference rules on how to combine the word-level and the construction level. And the elements of a the constrution semantics may be implicit, below the surface.

(40) Basic Categories

• To get started: S, NP, and N. Others will be
added as needed.
• Here we ignore morphosyntactic details such as
case, agreement, and verb inflection.
• In a more detailed CVG, these would be handled
(much as in pregroup grammars) by subtyping.

"pregroup grammars"? Does this subtyping refer to feature structures at a finer granularrity?

(41) Function Categories

• As in many frameworks (RG, HPSG, LFG, DG,
traditional grammar) grammatical functions (gram-
funs) like subject and complement are treated
as theoretical primitives.
• To start we just assume the gramfuns subject
(s) and complement (c). Others will be added
as needed.

Can gramfuns be extended to handle thematic roles? Word-specific participant roles? This could be a finer grained semantics, capturing the insights from lexical semantics and its data.

(56) An Embedded Constituent Question

⊢ [whatfill t(s Kim (likes t c))] : Q
• Here what is an operator of type NP^Q_S : it combines with an S containing an unbound NP trace
to form a Q, while binding the trace.
• Notice that the is not analyzed as a “projection”
of a ‘functional category”: there is no null com-
plementizer with respect to which the operator is
a “specifier”.

Can something like this be used to analyze "ang" in "babae ang bumili ng lasones"

Day 2

(2) Some Examples of Overt Movement

a. John_i, Fido bit t_i. [Topicalization]
b. I wonder [who_i Fido bit t_i]. [Indirect Question]
c. Who_i did Fido bite t_i? [Direct Question]
d. The neighbor [who_i Fido bit t_i] was John. [Relative Clause]
e. Felix bit [who(ever)_i Fido bit t_i]. [Free Relative]
f. It was John [who_i Fido bit t_i]. [Cleft]
g. [Who_i Fido bit t_i] was John. [Plain Pseudocleft]
h. [Who_i Fido bit t_i] was he bit John. [Amalgamated Pseudocleft]
i. [[The more cats]_i Fido bit t_i], [[the more dogs]j Felix scratched tj ].
[Left and right sides of Correlative Comparatives]
In all these examples, the expression on the left periphery that
is coindexed with the trace is called the filler, or extractee, or
dislocated expression.

It seems likely that the dislocated "ang" in the Tagalog construction above can be analyzed as Overt Movement in this sense.

This list seems to capture what Goldber 1995 referred to as nonbasic constructions:

... it is not being claimed that all clause-level constructions encode scenes basic to human experience. Nonbasic clause-level constructions such as cleft constructions, question constructions, and topicalization constructions (and possibly passives) are primarily designed to provide an alternative information structure of the clause by allowing various arguments to be topicalized or focused. Thus children must also be sensivtive to the pragmatic information structure of the clause (Halliday 1967) and must learn additional constructions which can encode the pragmatic information structure in accord with the message to be conveyed. These cases are not discussed further here (cf. Lambrecht 1987, 1994).

This would hint that passives are not to be analyzed the same way. Also, the intentional "design" might be glossed as "statistically selected to fill the social function." What representation might be suitable for this pragmatic information structure?

The dislocated "ang" introduces a trace for the intiator of the event of the specified VP, and the predicative noun characterizes that initiator with a common noun type. It is a bit like the Free Relative construction, but with a nominal-type predicate instead of a subject-specified action-verb predicator. So the pragmatic information might be glossed: "[(it) (was) a woman]_i [whoever_i bought the lanzones fruit]"

Onomasiology

I learned a new word today, 

a branch of linguistics concerned with the question "how do you express X?". ...as a part of lexicology, departs from a concept (i.e. an idea, an object, a quality, an activity etc.) and asks for its names. The opposite approach is known as semasiology: here one departs from a word and asks what it means, or what concepts the word refers to.  

It seems to be a field relevant to the distinctions in lexical field.

English	Cebuano	Tagalog
Seed	Lisu	Butó
Bone (Tetrapod)	Bukóg	Butó
Bone (Fish)	Bukóg	Tiník
Thorn	Tunók	Tiník

I should gather data for these concepts from several Central Philippine languages, and see what it tells me about shared derived characters.

"The coinage of a new designation can be incited by various forces (cf. Grzega 2004):

• difficulties in classifying the thing to be named or attributing the right word to the thing to be named, thus confusing designations
• fuzzy difference between superordinate and subordinate term due to the monopoly of the prototypical member of a category in the real world
• everyday contact situations
• institutionalized and non-institutionalized linguistic pre- and proscriptivism
• flattery
• insult
• disguising things (i.e. euphemistic language, doublespeak)
• taboo
• avoidance of words that are phonetically similar or identical to negatively associated words
• abolition of forms that can be ambiguous in many contexts
• word play/punning
• excessive length of words
• morphological misinterpretation (creation of transparency by changes within a word = folk-etymology)
• deletion of irregularity
• desire for plastic/illustrative/telling names for a thing
• natural prominence of a concept
• cultural-induced prominence of a concept
• changes in the world
• changes in the categorization of the world
• prestige/fashion (based on the prestige of another language or variety, of certain word-formation patterns, or of certain semasiological centers of expansion)

The following alleged motives found in many works have shown to be invalid by Grzega (2004): decrease in salience, reading errors, laziness, excessive phonetic shortness, difficult sound combinations, unclear stress patterns, cacophony."

  - Onomasiology, Wikipedia, citing:

Grzega, Joachim (2004), Bezeichnungswandel: Wie, Warum, Wozu? Ein Beitrag zur englischen und allgemeinen Onomasiologie. Heidelberg: Winter, ISBN 3-8253-5016-9. (reviewed by Bernhard Kelle in Zeitschrift für Dialektologie und Linguistik vol. 73.1 (2006), p. 92-95)

Web journals touching on lexical semantics and computational linguistics

Linguistic Issues in Language Technology (LiLT)

Semantics and Pragmatics

Linguistik Online

Onomasiology Online

 -  Helpful Internet Sources: links to online dictionaries and linguistic atlases

papers on the Web September 2008

Brett Kessler

• Kessler, B. (2007). Word similarity metrics and multilateral comparison. In Proceedings of Ninth Meeting of the ACL Special Interest Group in Computational Morphology and Phonology (6–14). Stroudsburg PA:Association for Computational Linguistics.

• Kessler, B. (2007, July). Better than chance? Randomization models for evaluating whether lexical similarity implies historical connection. Paper presented at the Workshop on Alternative Approaches to Language Classification, Stanford, CA.

Kessler's encyclopedia entry on Language Families

Brett Kessler. (in Press). Language Families. . In Hogan, P. C. (Ed.). The Cambridge Encyclopedia of the Language Sciences.
http://spell.psychology.wustl.edu/~bkessler/CELS/Language_Families.pdf

He points out that it is a model divergence, not similarity that is the basis of family relationships.

"... a particular model of LANGUAGE CHANGE: divergence. When innovations in one part of a language community fail to spread to other parts, differences accumulate until the community can be said to speak different languages."

"there is no requirement that cognates be similar at all (e.g., English two is related to Armenian yerku), and many sources of similarity are disavowed as being irrelevant to the model. These include borrowing (see CONTACT, LANGUAGE), onomatopoeia, universals (ABSOLUTE AND STATISTICAL UNIVERSALS), and chance similarities."

The problem I am interested in related to the Bisayan subfamily is called cladogenesis: "Subgrouping seeks to uncover the history of the divergence (cladogenesis) of a language family." What is important is not to look for cognate sets, but for a sister subfamily that has a base character rather than an member of the cognate set.

"the linguist looks for evidence that some proper subset of those languages may have descended from an intermediate common ancestor. This is done by looking for shared innovations (synapomorphies) – sound changes or new words or grammatical constructions that were not in the ancestor language but are found in two or more of the descendant languages."

"some of Greenberg’s key ideas can be transformed into algorithmic (reproducible) methodologies that introduce to language family research the benefit of statistical significance testing. Oswalt’s procedure (1998) minimized experimenter bias by requiring that a specific concept list be used and that one specify in advance specific criteria for measuring degree of similarity between two languages. Baxter and Manaster Ramer (2000) added reliable significance testing procedures based on randomization tests. Kessler and Lehtonen (2006) adapted the technique to handle multiple languages in a single test, informally confirming Greenberg’s claim that such large-scale comparisons are inherently more powerful than two-language comparisons. Ringe (1992; see Kessler 2001 for extensive discussion and methodological refinements) measured not similarity but the number of recurrent sound correspondences. This has the advantages both of being closer to the traditional comparative method and of generating correspondences useful for subgrouping and reconstruction. Disappointingly, however, none of these neo-Greenbergian techniques found evidence for the deep relations that were advertised for the original, impressionistic, method."

There may be hope for quantifying evidence.

"The recent development of computational cladistic methods similar to those used in biology (e.g., Ringe, Warnow, and Taylor 2002) is a tremendous advance in helping the linguist find optimal trees. In addition, several solutions to the problem of borrowing have emerged in the form of programs that construct networks instead of trees. Shared innovations that cannot be cleanly attributed to a shared ancestor are taken as evidence of contact, obviating somewhat the need to make a priori judgments about whether borrowing was involved (e.g., Bryant, Philimon, and Gray 2005; Nakhleh, Ringe, and Warnow 2005)."

"Recent computer techniques add simplicity, reproducibility, and quantitative rigor to methodologies for proving relationships between languages, but so far there has been no noticeable increase in power over what experts are able to do by hand."

References

• Baxter, William H. and Alexis Manaster Ramer, 2000. “Beyond Lumping and Splitting: Probabilistic Issues in Historical Linguistics.” In Time Depth in Historical
  
• Linguistics, ed. Colin Renfrew, April McMahon and Larry Trask, 167–188. Cambridge, England: McDonald Institute for Archaeological Research.
  
• Blust, Robert. 1999. “Subgrouping, Circularity and Extinction: Some Issues in Austronesian Comparative Linguistics.” In Selected Papers from the Eighth International Conference on Austronesian Linguistics, ed. E. Zeitoun and P. J. K Li, 31–94. Taipei: Academia Sinica.
  
• Bryant, David, Flavia Filimon, and 
  
• Russell D. Gray. 2005. “Untangling Our Past: Languages, Trees, Splits and Networks.” In The Evolution of Cultural Diversity: A Phylogenetic Approach, ed. Ruth Mace, Clare J. Holden, and Stephen Shennan, 69–85. London: UCL Press.
  
• Cavalli-Sforza, Luigi Luca, Paolo Menozzi, and Alberto Piazza, 1994. The History and Geography of Human Genes. Princeton University Press
  
• Gordon, Raymond G., Jr. (ed.). 2005. Ethnologue: Languages of the World. 15th ed. Dallas, TX: SIL International. Content also available online at http://www.ethnologue.com/
  
• Greenberg, Joseph H. 1963. “The Languages of Africa.” International Journal of American Linguistics, supplement 29(1), pt. 2.
  
• Greenberg, Joseph H., 1987. Language in the Americas. Stanford (CA): Stanford University Press.
  
• Greenberg, Joseph H., 2002. Indo-European and its Closest Relatives: the Eurasiatic Language Family: Lexicon. Stanford, CA: Stanford University Press.
  
• Kessler, Brett, 2001. The Significance of Word Lists. Stanford, CA: Center for the Study of Language and Information.
  
• Kessler, Brett and Annukka Lehtonen. 2006. “Multilateral Comparison and Significance Testing of the Indo-Uralic Question.” In Phylogenetic Methods and the Prehistory of Languages, ed. P. Forster and C. Renfrew, 33–42. Cambridge, England: McDonald Institute for Archaeological Research.
  
• Mallory, J. P. 1989. In Search of the Indo-Europeans: Language, Archaeology and 
  
• Myth. London: Thames & Hudson.
  
• Nakhleh, Luay, Don Ringe, and Tandy Warnow. 2005. “Perfect Phylogenetic Networks: A New Methodology for Reconstructing the Evolutionary History of Natural Languages.” Language 81: 382–420.
  
• Oswalt, Robert L., 1998. “A Probabilistic Evaluation of North Eurasiatic Nostratic.” In Nostratic: Sifting the Evidence, ed. J. C. Salmons and B. D. Joseph, 199–216. Amsterdam: Benjamins.
  
• Renfrew, Colin. 1987. Archaeology and Language: The Puzzle of Indo-European Origins. London: Pimlico.
  
• Ringe, Don A., Jr., 1992. On Calculating the Factor of Chance in Language Comparison. Philadelphia, PA: American Philosophical Society.
• Ringe, Don, Tandy Warnow, and A. Taylor. 2002. “Indo-European and Computational Cladistics.” Transactions of the Philological Society 100: 59–129.
  
• Swadesh, Morris, 1955. “Towards Greater Accuracy in Lexicostatistic Dating.” International Journal of American Linguistics 21: 121–37.
  
• Thomason, Sarah Grey and Terrence Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley, CA: University of California Press.

Kessler's home page.

His thesis at Stanford was on:

Thesis title: Estimating the Probability of Historical Connections Between Languages. Available through UMI. However, a significantly revised version is published by CSLI Publications under the titleThe Significance of Word Lists: Statistical Tests for Investigating Historical Connections Between Languages and is distributed by The University of Chicago Press (2001; ISBN cloth 1-575862-99-9, paper 1-575863-00-6). From the preface:

recognizing different levels of regularity in a lexicogrammar model of linguistic attunement

lexical rules may themselves be meaning-bearing "constructions" of several morphemes at a level smaller that a phrase.

1. utterance token
2. construction instance, wordform, morph
3. construction class, lexical item, morpheme
4. schematic construction, lexical rule, derivation pattern of morphemes (abstraction over a class, reducing it to essential elements)
5. relations of constructions (central, peripheral)
6. hierarchical lexicogrammar

2b. referring act or speech act segment of building a shared information state


For Austronesian,

study morpholexical constructions of stems in usage as different parts of speech


Goldberg + "Construction as Categorization of Language"

"Exactly which categories are formed is determined to a great extent by the usefulness of a potential category in predicting how the environment will behave. ... Function plays an obvious and important role in determining what is categorized and affords similar predictive value to the category. ... When it comes to language, clearly the relevant predictive tasks are to predict meaning, given the form (comprehension) and to predict form, given meaning (production). Thus given the fact that constructions relate critical aspects of form and meaning, their acquisition is well-motivated. Accordingly, earlier work has emphasized the role of prediction in language learning, and has shown that constructions are good predictors of overall sentence meaning, useful for conveying “who did what to whom” (Bencini & Goldberg, 2000; Goldberg, Casenhiser, & Sethuraman, 2005)."

Thomas of Erfurt on modes of signifying

Relations and roles in action verbs

Language speakers or signers use a verb to pick out a constraint in a scene (or a less concrete situation, often by metaphor) where something is happening. The verb classifies the scene with as an event type, and the verb-constraint is the relation of the Before subscene-type to an After subscene type. Attunement to the verb-constraint allows a speaker-hearer who recognizes (and perhaps expresses with) the verb to construct a shared information state with a number of expected participants, some of which are obligatory in the syntax. What is happening in the action scene is there is a relevant shift from a Before configuration of participants to an After configuration. There may be many different relations picked out in the scene (to a greater or lesser extent), but the most important one is the verb-constraint linking Before and After. It allows the users of the verb-containing utterance to acquire information about the action scene in time, that the action started presupposing the participants in the Before subscene-type and resulted in a situation of an After subscene-type with the participants in a new configuation. A clausal expression predicates that the verb-constraint obtains in the action scene, and creates conditions of satisfaction for a successful communicative act. The receiver of the utterance, if attuned to the verb-constraint, can understand those conditions of satisfaction which can affect their cognition via a shared information state and subsequently their perception and action in the world.

Other relations in the action scene are between the participants. However, certain relations are made explicit by the form of the utterance, by the relation of the participant argument (typically a referring phrase, an NP) to the head verb. The most important (and thus least oblique) argument role defines a relation to the action mentioned by the head verb called a thematic role of Agent. This is in a world where relevant action is performed by other humans, or animals that behave with intentionality and volition similar to humans. A more general thematic role that does not require intention is Effector. Both Agents and Effectors pick out the Initiator of an action occurring. In the Stanford CSLI Verb Semantic Ontology, the two are sometimes not distinguished, and assigned the thematic Role of EFNT, which I guess means Effector-Agent but could simple be pronounced Initiator.

A prototypical type of action scene involves an Agent manipulating an object in the scene. That object may take the role of Patient or Theme. A Patient is structurally changed by the Action, while a Theme may be moved, said or experienced but does not undergo structural change.

Sowa's IRMPACO ontology in Appendix B of his book Knowledge Representation [Sowa 2000] sets out 18 thematic roles, classified according to four kinds that are reminiscent of Aristotle's four aitiae (singluar aitia), traditionally translated as causes (effective, material, formal, ultimate). In an orthogonal dimension, they are classified as Action, Process, Transfer, Spatial, Temporal and Ambient.

It is worth analyzing those 18 roles further, in relation to Goldberg's analysis of constructions (and FrameNet's semantic frames), but that is the subject of a different posting. This a way of exploring verbs and their arguments, and can be contributed to the Stanford CSLI Semantic Verb Ontology. That resource already considers WordNet, I would want to explore mappings to COBUILD and FrameNet.