In my copious free time (hah!) Iāve been studying Pali, which is kind of a step-sibling to Sanskrit. Like Sanskrit and a lot of other Indo-European languages, itās overflowing with inflection. Consquently, grammars have metric boatloads of tables: noun and pronominal declensions, verb conjugations, and so forth.
Being me, I have ended up trying to digitize some of the data, and itās the tables that have really gotten me thinking. First of all, the tables are big:
This particular one encodes five features: noun class, gender, case, number, and frequency.
Whatās interesting to me is to realize that this table is just one of an essentially limitless number of ways that one might go about presenting this information. In the pedagogical blog post below, the same information is presented in quite a different way:
For instance, hereās the table that presents the subset of forms that are have neuter gender and singular number:
(Pali is notorious for variation ā yes, there are five possible inflections for singular ablatives in -u!)
Ultimately, a table represents all possible combinations of all possible values of a set of grammatical categories. The first non-gray cell in the first chart āmeansā Nominative case, singular number, āshort aā noun class, and masculine gender. The first cell in the second chart āmeansā nominative case and āshort aā noun class ā but thatās it! The fact that what āgoes intoā that cell is also neuter gender and singular number is a āgivenā ā you could imagine, for instance, another language that has neither gender nor number marking, and this table structure would be just as useful.
Aaaaand waht are you trying to say Pat
Well, tabulations like these serve particular purposes. The first table ā a real beast to look at ā is good for reference (e.g., ācould this form that ends in -isma be an ablative neuter singularā¦?ā). The second one is perhaps useful for a learner, for instance, if youāre trying to get a grip on any kinds of patterns within this subset of inflection ā locatives seem to have an sm in there somewhere, instrumentals have an n, and so forth.
But I think because theyāre a pain in the neck to construct, we tend to think of tables like this as āpresentationā tools, not as āresearchā tools. But itās fair to say that the tables are arrangments of linguistic forms (suffixes, in this case). The tables are just helpful arrangements of those forms. You could make a flat table, of course, this kind of thing:
| form | stem | case | number | gender | frequency | |||
|---|---|---|---|---|---|---|---|---|
| 1 | o | a | nominative | singular | masculine | normal | ||
| 2 | e | a | nominative | singular | masculine | normal | ||
| 3 | aį¹ | a | nominative | singular | neuter | normal |
(See the bottom of this post for all 840 glorious forms. Donāt leave home without it.) ā¦I tried to include the whole table, but hit the post size limit for this forum! 
It seems to me that a table like this is actually the ārightā way to edit data that will be tabulated. The tables should be generated from the flat table, right?
My questions to you:
- How do you handle this sort of thing?
- Do you think automation is worth the trouble? (I.e., maybe hand-editing a table is fine because that works for you as a canonical reference)
Then thereās a second kind of question. Suppose there were a ātable builderā kind of tool. You feed it some flat data (from a .csv or a .json file, say), and then there is some sort of user interface to build tables from that data.
Would such a thing be useful? Can you imagine what it would look like?
: