About Me

A Shaggy Dog Story for Phonologists

Posted: March 14, 2013

In another life I would have been a field linguist. As it stands, I'm just a hobbyist who's taken a handful of linguistics and cognitive science classes, ones that seem totally orthogonal to my stated academic trajectory.

Still, I like to think my linguistics education benefitted me in all sorts of ways. I am infinitely grateful to Professor Mary Paster at Pomona College for her phonology course. One of Prof Paster's biggest motivations was getting her students to write like proper academics. An analytical paper every week. I left that class a far stronger writer than when I entered.

So, the shaggy dog story. Midway through the semester, pushing through the latest data set, I noticed the makings of an awful groaner joke, right there in the assignment! And to properly get the punch line requires not-insubstantial phonology knowledge and a lengthy set-up regarding genitive plurals and diminutive forms in Slovak. Things like that are guaranteed to make bad jokes worse, so I was pretty excited!

Look down to read that bad joke. It's an excerpt from my write-up on Slovak morphology, using what I understand to be a data set so old that no one gives proper attribution anymore.

Slovak, a Slavic language spoken primarily in central Europe, often lengthens or diphthongizes syllabic elements within roots when applying affixes. The Slovak data below exhibit this vowel-altering process in particular noun forms.

nom. sg. gen. pl. gloss
lipa liːp ‘linden tree’
muxa muːx ‘fly’
lopata lopaːt ‘shovel’
sr̩na sr̩ːn ‘deer’
ʒena ʒien ‘woman’
kazeta kaziet ‘box’
hora huor ‘forest’
sirota siruot ‘orphan’
pæta piat ‘heel’
mæta miat ‘mint’
kopito kopiːt ‘hoof’
bruxo bruːx ‘belly’
blato blaːt ‘mud’
salto saːlt ‘somersault’
embargo embaːrg ‘embargo’
yabl̩ko yabl̩ːk ‘apple’
koleso kolies ‘wheel’
lono luon ‘lap’
hovædo hoviad ‘beast’
vlaːda vlaːd ‘government’
bluːza bluːz ‘blouse’
dlaːto dlaːt ‘chisel’
viːno viːn ‘vine’
t͡ʃiara t͡ʃiar ‘line’
hniezdo hniezd ‘nest’
noun diminutive gloss
hrad hraːdok ‘castle’
list liːstok ‘leaf’
xl̩p xl̩ːpok ‘hair’
kvet kvietok ‘flower’
hovædo hoviadok ‘beast’

The roots of the nouns in the dataset are discernible directly from the noun column or from the nominative singular column, appended with the nominative singular morpheme, /-a/. This affix takes the surface forms [-a] and [-o], as in sr̩na ‘deer’ and koleso ‘wheel.’ The genitive plural morpheme does not insert a particular phoneme into the root; rather, it lengthens or diphthongizes the ultimate vowel in the word root, as in sr̩ːn ‘deer’ and kolies ‘wheel.’ Roots are unaffected if the ultimate vowel is already long or a diphthong. The long vowels present in the dataset are [aː, iː, uː, r̩ː, l̩ː], and the diphthongs present are [ie, uo, ia]. Lengthened forms of [æ, e, o] are conspicuously absent.

The diminutive marker affix, like the genitive plural marker, lengthens or diphthongizes the ultimate vowel of the root; however, it also appends to the root the suffix /-ok/. The diminutive and genitive plural noun forms do not share any particular phonetic conditions that would explain why those forms experience lengthening and diphthongizing while others do not. This suggests that the boundaries for the environment in which the lengthening-diphthongizing process applies are morphological in nature. Due to this bounding condition, any rules developed to describe this process must account for how it applies only to the root morpheme, even after affix morphemes have been added, such as in the diminutive noun form. So, the rule must be designated as occurring only within root morphemes and only for certain word forms.

Creating one rule to describe both lengthening and diphthongization is perhaps doable, but elaborate and unnecessary. It is more logical to create one rule by which vowels become lengthened, and another rule by which some long vowels become diphthongs, as below.

Rule (A), lengthening, must be ordered before (B), raising, or else word roots such as koles ‘wheel’ would yield incorrect genitive plural forms, such as *koleːs.

The environments and boundaries described above are only denoted in (A). The vowel raising observed in (B) is context-free. There is no hard evidence for [eː, oː, æː] being part of the underlying phonetic inventory of the Slovak language, but words with surface realizations that include the diphthongs mentioned in (B) can be correctly described with the above rules, presuming the underlying form is one of the long vowels from (B). For example, the genitive plural noun t͡ʃiar ‘line’ does not contain a long vowel, which can be attributed to the diphthong [ia], which would be /æː/ in the underlying representation. If the word’s underlying form were /t͡ʃiar/, rule (A) would incorrectly predict its surface representation to be *t͡ʃiaːr. However, if the underlying form were /t͡ʃæːr/ instead, applying rule (A) would not change the representation, and applying rule (B) would correctly yield the surface form t͡ʃiar. Therefore, I suggest that all instances of [ie, uo, ia] in surface representations are the product of rule (B) being applied quite late in the overall rule ordering, after any rules that depend on or affect vowel length.

This is a very abstract analysis, owing to the fact that the long vowels posited to be represented as diphthongs by (B) appear nowhere in the surface phonetic inventory of the language. The universality of the rule suggests an historical presence of [eː, oː, æː] as surface forms, hence the artful name “Slovak Vowel Shift,” though to base any critical analysis on this assumption would be trifling speculation.

Nevertheless, this analysis simply and accurately predicts the surface representations of Slovak words, and without the Slovak Vowel Shift it becomes very tedious to explain why, say, the genitive plural of the root t͡ʃiar ‘line’ is t͡ʃiar, but the genitive plural of the root blat ‘mud’ is blaːt.

I feel the need at this point to comment on a peculiar similarity between modern life and the very structure of Slovak grammar. The importance of the ordering of these rules is something we are duly familiar with. We must be mindful of when we apply them because one is a Shift, and the other is case-sensitive.