Madras Tamil Lexicon

Two separate walls

The licence. CC BY-NC-ND has two clauses that both bite. NonCommercial bites because our code is Apache-2.0 and anybody may run this server commercially, which we cannot prevent and would not want to. NoDerivatives bites because a restructured, indexed copy inside a knowledge store is plausibly a derivative work, and "plausibly" is not a standard we are willing to ship on.

The robots file. The only query endpoint is disallowed to automated clients. So even the narrow reading, look a word up at query time and store nothing but the fact and the citation, is not available to us without asking.

We noticed the rule and did not scrape. That is worth stating plainly, because it is the difference between a request and a fait accompli, and because the alternative is available to anyone with an afternoon and no scruples.

Why this source is the reason for our whole grading model

This lexicon is grade A and consult-and-cite at the same time. It is the most authoritative source we could possibly have and the most restrictively licensed one we hold.

A system with a single field for "how good is this source" has to lie about one of those. So ours has two: evidential grade, and what we are legally allowed to do with the bytes. They move independently. That design came directly out of running into this entry, and it later caught a more embarrassing case in the other direction, a word list we depended on whose licence turned out to be nothing at all.

What we asked, and what we already know

A letter to the Digital South Asia Library is drafted, with one question and one request:

1. Does the digitised edition carry machine-readable etymology fields? We cannot check, because those pages are the ones we are not fetching. This question matters more than the permission does.
2. Permission for rate-limited programmatic queries, storing derived claims and citations only, never the lexicon's own wording.

The first question is first for a reason. We assessed the other digitisation of the same lexicon and found glosses without etymology. If this edition is the same, permission buys us very little, and it would be a waste of an institution's goodwill to ask for something we do not need.

Meanwhile the honest position holds: where a borrowed word carries no orthographic marker, this project returns unknown rather than a plausible etymology. Some of those unknowns would close immediately if this door opened. thamizh@ief-global.org