Implementation: overview of generating a form
The generate function has multiple methods (see API for details). We'll follow, from the top down, the process of generating a token for a given lexeme (identified by a LexemeUrn) and morphological form (identified by a FormUrn), using a given Kanones dataset.
Following the dispatch trail
Find stem and rule pairs
The generating algorithm first selects from the cross product of stems and rules in the dataset all stem/rule combinations matching the requested lexeme and form. Stems are subtypes of the abstract KanonesStem; rules are subtypes of the abstract KanonesRule.
Kanones datasets store form information for regular formations in the rules tables of the data set. Irregular formations identify forms in the stems tables for irregulars. Thus when the type of the rule is IrregularRule , generate tests whether the form of the stem matches the requested form; otherwise, it tests on the form of the rule. This is straightforward since subtypes of both KanonesStem and of KanonesRule implement the formurn function.
Dispatch to appropriate subtypes of stem and rule
The next step is to invoke generate using the stem and rule pair as parameters. We have advanced the problem from the abstraction of a lexeme and form to the concrete problem of generating a token using a specific stem and specific rule.
This method of generate further dispatches based on the type of the stem and rule. For regular formations, each of Kanones' analytical types ("parts of speech", (described elsewhere in this documentation)) has a method of generate to handle a corresponding pair of rule and stem types:
| Analytical type | Stem type | Rule type |
|---|---|---|
GMFNoun | NounStem | NounRule |
GMFPronoun | PronounStem | PronounRule |
GMFAdjective | AdjectiveStem | AdjectiveRule |
GMFAdverb | AdjectiveStem | AdverbRule |
GMFFiniteVerb | VerbStem | FiniteVerbRule |
GMFInfinitive | VerbStem | InfinitiveRule |
GMFParticiple | VerbStem | ParticipleRule |
GMFVerbalAdjective | VerbStem | VerbalAdjectiveRule |
GMFUninflected | UninflectedStem | UninflectedRule |
For irregular forms, generate dispatches to a single function for the IrregularRule type and any StemType.
Irregular forms
Need to add a note on how irregular forms are returned.
Regular forms: concatenating stem and inflection
Conceptually, we want combine the string data included in stems and rules, and found with the stemstring and ending functions. Essentially, regular forms can now be found by concatenating them:
stemstring(stem) * ending(rule)However, there are numerous specific details that need to be considered, such as phonological changes required by Greek string concatenation, or addition of augment or reduplication for verb forms. These are documented in more detail in the following pages.
Regular forms: applying accent
The final step is to apply the appropriate accent to the concatenated string. Accenting Greek words depends on:
- phonological considerations (such as vowel quantity)
- lexical considerations (such as derivation of verb forms from compounds)
- morphological considerations (such as the basic principle that finite verb forms have recessive accent, while substantives have persistent accent)
The details of this process are documented in the following pages.