No label defined (Q3220566)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 3885878
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 3885878 |
Statements
1985
0 references
natural language semantics
0 references
syntax
0 references
extensional logic
0 references
The book is a new and quite remarkable step towards a rigorous mathematical (model theoretic) approach of the natural language (NL) semantics. Originating in an authors' earlier monograph [''Logical Types for Natural Language'' (UCLA, 1978)], the present work has to be considered among the daring and successful attempts, alongside the Montague's works, to model the semantic behaviour of the NL syntactic structures. Instead of interpreting these structures in a formal logic language, the authors' solution is to model an adequate and general formalization of a NL significant (English) sublanguage in the (Boolean) algebraic frame. This frame, by the well-known (Stone) representation theorems of the Boolean algebras, permits to use properties instead of individuals as semantic primitives (the ontology). Even adopting sometimes unavoidable Montague's basic solutions, the present book is essentially an original one. Using the mathematical tools with deepest competence, it offers a much more workable linguistic modelling, still quite rigorous, situated in a purely algebraic frame. Many solutions are original ones, suitable to the new techniques, with valuable alternatives. The generality and elegance of the treatment are the outstanding features of the present approach, due, mainly, to the adopted mathematical modelling style. In this way, the book represents a theoretical basis, a real ''model'' for the fundamental studies in NL semantics. Part I presents the extensional logic for a core language L, intended to represent the basic predicate-argument structure of the (first-order) English. The syntactic categories of L (section A) are analysed: common noun phrase (N), full noun phrase (\(\bar N\), standing for the classical notation NP) having as subcategory the proper noun phrase \((N_{prop})\), determiner (Det), one place predicate \((P_ 1)\), two place predicate \((P_ 2)\), and sentence (formula or zero place predicate, denoted \(P_ 0\) or simply P). Using ''slash'' (/), ''bar'' \(\bar(.)\) and subcategorization features, the syntactic categories are formally constructed (taking as primitive categories N and P) and the core language L defined on a restricted but significant (English) vocabulary. The semantics for L is defined associating to each category C of L a set of denotations for expression of that category, called the type for C, denoted \(T_ C\). The types for the primitive categories N and P constitute the semantic primitives (the ontology) for L, and are defined directly, while the types for complex categories will be given in terms of the types of those categories from which they are formed. Considering the (extensional) ontology as being the pair \(<{\mathcal P},\{0,1\}>\) (where \({\mathcal P}\) is an arbitrary complete and atomic Boolean algebra associated to \(T_ n\)- called the properties -, and \(\{\) 0,1\(\}\) is the Boolean algebra of the truth values associated to \(T_ P)\), the types for other categories are defined or proved to be: \(T_{\bar N}={\mathcal P}^*\) (the power set Boolean algebra of \({\mathcal P})\), \(T_{\bar N_{prop}}=I_{{\mathcal P}}\) (the set of individuals on \({\mathcal P})\), \(T_{Det}=T_{\bar N/N}\), etc. Denoting the set of complete homomorphisms from an algebra B into an algebra D by \(H_{D/B}\), the types for the n-place predicates can be inductively defined by the formulas: \(T_ P=\{0,1\}\), \(T_{P_{n+1}}=H_{T_{P_ n}}{}_{/{\mathcal P}^*}\). Since the predicative hierarchy of a category C was inductively defined as \(C_ 0=C\) and \(C_{n+1}=C_ n/\bar N\), now the types for the predicative hierarchy of C can be defined as: \(T_{C_ 0}=T_ C\), \(T_{C_{n+1}}=H_{T_{C_ n}/{\mathcal P}^*}\), \(T_{C_{n+1}}\) proved to be a complete and atomic (Boolean) algebra whose operations are defined pointwise on the individuals. The entailment (logical implication) relation is established between structures whose types are in the relation of inclusion. The entailment is also generalized, corresponding to a greater variety of syntactic structures than simply the sentences. In contrast with Montague's approach, the authors' system has the property called ontological perfection (a property being thinked as a set of individuals). Thus the sets of properties can be generated from individuals by set operations. For instance, the type for \(N_{prop}\) is proved to be the minimal set of complete generators for \(T_{\bar N}.\) Section B of Part I extends the core category language L, giving syntactic and semantic (algebraic) characterizations for modifiers (adjective phrases, predicate adjectives), predicate modifiers, predicatives (case marking, subcategories of n-place common noun phrases, transitive adjective phrases, ditransitive verb phrases, prepositions and predicate modifier hierarchies, prepositions as verbal extensors), valency affecting operators (passives and generalized passive operator, agent phrases, extending passive to nominals and other predicative hierarchies), extending and subcategorizing the class of determiners (numerical, non-logical, ''logical'', vague determiners compared with other categories, generalizing determiners), variables and variable binding (lambda, reflexive and relative clause) operators, non-homomorphic predicates. Part II is concerned with intensional logic. The basic change in comparison with the extensional system is that the ontology is the triplet \(<{\mathcal P},\{0,1\},J>\), where J is an arbitrary set indexing the possible worlds which are involved in a particular intensional interpretation. The relations existing between the types in the two systems could be suggested by the following fact: given the category C, the intensional type for C, \(T^ i_ C\) is given by the (classical) formula \(T^ i_ C=F_{T^ e_ C/J}\), i.e. the set of all the functions with domain J and range \(T^ e_ C\). Local, global and functional properties of the set of functions defined on J give the opportunity to formally (algebraically) construct the intensional notions (and system) and to relate them on the extensional ones. Their linguistic and logic ''transparency'' offer fruitful solutions for mathematical modelling. Also from this points of view the book is an exciting invitation to creative reflection. Being of capital interest for (mathematical and computational) linguists, logicians and algebrists, we believe that the book is, at least for NL semantics, one of the most valuable of the present decade and will strongly influence the forthcoming researches in the field.
0 references