A Morris Halle memory

Morris Halle passed away earlier today. Morris was an absolute giant in the field of linguistics. His work in the 1950s and 1960s completely revolutionized phonological theory. He did this, primarily, by rejecting an axiom of the previous century’s work.
The theory of phonology was so utterly transformed by his argument against the principle of biuniqueness that the very concept is rarely even taught in the 21st century.
And this was just one of his earliest scientific contributions.

I could say a lot more about Morris’s work, but instead let me tell a short anecdote. In 2010 or so I happened to be in the Boston area and my advisor kindly arranged for me to meet Morris. After getting coffee we walked to his spare shared office. The only thing of note was a single wall-mounted bookshelf containing three books: Morris’ own Sound Pattern of Russian and Sound Pattern of English—with the dust cover removed so as to exhibit the unique bas-relief cover designed by Morris’s wife, a talented visual artist—and of course, Walker’s rhyming dictionary. For whatever reason, we started to discuss Latin. Working with the legal pad, Morris first showed me a novel analysis of thematic vowels. Ignoring a few irregular (“athematic”) stems, all Latin verb stems have a characteristic final vowel: -ā- in the first conjugation, -ē- in the second, a vowel of varying quality (usually e or i) in the third, and -ī- in the fourth. In the first conjugation and most of the third conjugation, this vowel disappears in the first person singular active indicative verb, which is marked with an -ō suffix. Thus for the second conjugation verb docēre ‘teach’, we have doceō ‘I teach’, with the theme vowel preserved, and similarly for the fourth conjugation. In contrast, for the first conjugation verb amāre ‘love’, we have amō ‘I love’, with the theme vowel omitted, and similarly for the majority of the third conjugation. This much I already knew. To me it was just one of those conjugational quirks one has to memorize when learning Latin but Morris suggested that it was not necessarily so. What if, he argued, the first conjugation -ā- was deleted by a following -ō? (Certainly that rule is surface-true, except for a handful of Greek loanwords like chaos.) But what about the third conjugation? Morris suggested that he had long believed the underlying form of the third conjugation theme vowel was [+back], something like /ɨ/, and he proceeded to lay out the necessary allophonic rules, and finally a rule which deletes the first of two [+back] segments! I was floored.

I then showed him an analysis I was working on at the time. Once again ignoring a few irregulars, Latin masculines and feminine nouns of the third declension are characterized by a nominative singular suffix -s. When the verb stem is athematic and ends in a /t, d/, this consonant is deleted in the nominative singular (e.g., frons, frontis ‘forehead’). I argued that this rule ought to be extended to also target /r/ so as to account for the so-called “rhotic” stems like honōs, honōris ‘honor’ (e.g., /honōr-s/ → [honōs]). To make this work, one must write the rule so that it bleeds its own application (see here for the full analysis), and as one of several opaque rules. This is something which is possible in the rule-application framework proposed by Morris and colleagues, but which cannot be straightforwardly implemented in more recent theoretical frameworks. I must have hesitated for a moment as I was talking through this, because Morris grabbed my hand and said to me: “Young man, remember always to speak clearly and to never apologize for your rule ordering.” And then he bid me adieu.