Conceptual background
Modelling syntax
Any computational analysis of natural-language syntax must choose a syntactic model suited to its purpose. These pages summarize the goals of arsgrammatica, and how its model compares with other models of natural-language syntax.
- Goals of
arsgrammatica - Comparison with the Universal Dependency scheme
- Comparison with the Agent-Action-Target model
Implementation of the arsgrammatica model
The arsgrammatica package is based on three classes, representing the tokenization of a Latin passage, the syntactic anchor points of verbal expressions, and the graph of syntactic relations.
- Tokenization of a passage (the
Tokenclass) - AAT anchor points: verbal units (the
VerbalExpressionclass) - Graph of syntactic relations (the
TokenAnalysisclass)
Examples of analyses
Four sentences from arsgrammatica’s training set illustrate every recognized syntactic relation.