This is part 1 of a series of blog posts about modelling stems and principal part lists and covers the three sources of Attic Greek principal parts used to expand and test the Morphological Lexicon.
Because Louise Pratt’s The Essentials of Greek Grammar was the basis for testing a lot of paradigms, it made sense to use it as the starting point for Attic Greek principal parts as well. Pratt lists the principal parts (the standard six, i.e. not separating out the so-called “future passive”) for 247 verbs. It is not indicated the reason for her particular choice of verbs other than them being “common Attic Verbs”.
The second source is James Morwood’s Oxford Grammar of Classical Greek. Morwood has two lists, one of “Top 101 irregular verbs” and one of (81) “More principal parts”. The title of the first list suggests common verbs are omitted if regular. I have included both lists (although can treat them separately). Morwood includes a seventh part for the “future passive” (when and why this is useful is worthy of a separate blog post).
For my third source I used Chris Francese’s principal parts in the wonderful DCC Greek Core Vocabulary list. The DCC core vocabulary consists of 500 common words of which 151 are verbs.
All three lists included the occasional form outside the usual six or seven principal parts and a future post in this blog series will address the modelling of that.
The DCC principal parts were in electronic form and so were relatively easy to deal with (although I’ll discuss specifics in a later post). Both the Pratt and Morwood lists I did not have in electronic form and so manually keyed them in over the course of a few weeks (mostly in Vienna earlier this year).
I have also referred at times to Wilfred Major’s 80% list (discussed elsewhere on this blog) but, as it doesn’t contain principal parts, it was more of a reference for lemma choice and additional metadata than an input for testing part generation itself.
Of course many other lists could be included but these three are sufficient to establish most of the modelling issues and ensure the code works correctly. Data from other lists can be incorporated later relatively easily.