Markedness, neutralization, and structural size
Questions to start with
- What is the relationship between the environment where a phonological rule applies and what the rule does?
- What do representations need to look like if we want to make this relationship explicit?
- What kinds of patterns are especially informative in this regard?
Language background
Today we will look at a number of Kiranti languages. Kiranti is a group of Trans-Himalayan (‘Sino-Tibetan’) languages spoken in eastern Nepal. It generally (but not universally) considered to be a genetic subgrouping within the family, although its precise place is unclear. We will focus especially on the two Eastern Kiranti languages Yakkha [yakk1236] and Yamphu [yamp1244]. Both are spoken by several thousand people (around or under 10,000), and are considered endangered.
We focus on data from verbal morphology. Characteristic of the Kiranti languages show a pattern of allomorphy within verbal paradigms in which a stem has two allomorphs, by and large distributed in phonologically conditioned ways; the forms given here are picked (sometimes constructed) to be representative of this conditioning. We start with a consideration of Yakkha.
PDF version of the datasets.
Yakkha verbs
We consider four forms: the infinitive, the negative converb, the 3SG subject past tense form (with a clitic =na), and the 3SG subject + 3SG object past tense form (also with =na). For reasons that we cannot go into here, we should treat the 3SG>3SG marker as consisting of two suffixes /a+u/ underlyingly. The 3SG>3SG form is for obvious reasons absent for intransitive verbs.
The transcriptions are given in slightly stricter IPA.
| Infinitive | NEG.CVB | 3SG.PST | 3SG>3SG.PST | Gloss |
|---|---|---|---|---|
| waʔma | mẽwaʔle | wajana | wana | ‘wear’ |
| soʔma | mensoʔle | sojana | sona | ‘look at’ |
| piʔma | membiʔle | pjana | pina | ‘give’ |
| cama | menɟale | cajana | cona | ‘eat’ |
| ama | mẽale | ajana | ‘descend’ | |
| sima | mensile | sjana | ‘die’ | |
| lapma | mẽlaple | labana | labuna | ‘seize’ |
| apma | mẽaple | abana | abuna | ‘come’ |
| jokma | mẽjokle | joɡana | joɡuna | ‘search’ |
| pʰaʔma | mempʰatle | pʰatana | pʰatuna | ‘help’ |
| keʔma | meŋɡetle | ketana | ketuna | ‘bring up’ |
| liʔma | mẽlitle | litana | lituna | ‘plant’ |
| tʰuʔma | mentʰuʔle | tʰurana | tʰuruna | ‘sew’ |
| poʔma | memboʔle | porana | poruna | ‘topple’ |
| cʰuma | mencʰule | cʰusana | cʰusuna | ‘shrink’ |
| jama | mẽjale | jasana | jasuna | ‘be able (to do)’ |
Reconstruct, as much as you can, the paradigm for the following infinitives:
- cima ‘cool down’
- cokma ‘do’
- hoʔma ‘crumble, fall apart’
Yamphu verbs
This is a very artificially constructed dataset from the related language Yamphu. The patterns are similar to those of Yakkha, and will perhaps help you make more sense of the latter. The forms given are again the infinitive, as well as two negative non-past forms: the 1PL inclusive subject and the 1PL exclusive subject. You will need to make an additional generalization about the distribution of the suffixes.
| Infinitive | 1PL.INCL | 1PL.EXCL | Gloss |
|---|---|---|---|
| apma | abini | abiŋmani | ‘come’ |
| kepma | kebini | kebiŋmani | ‘stick’ |
| kʰaːkma | kʰaːɡini | kʰaːɡiŋmani | ‘scrape throat’ |
| tʰeʔma | tʰeʔni | tʰeduŋmani | ‘lift’ |
| hæːʔma | hæːʔni | hæːduŋmani | ‘bite’ |
| leʔma | ledini | lediŋmani | ‘be brief’ |
| siʔma | siʔni | sitːuŋmani | ‘hit’ |
Considering what you have been able to figure out about Yamphu, reconstruct as much as you can of the the paradigms of the following verbs:
- ʔokma ‘find’
- tɾiʔma ‘be contrary’
The data also attests patterns exemplified by asiʔ ‘previously’ ~ asiʔ-em-ba ‘before’. How do they fit in with the previous data?
Sources
The Yakkha data is from Schackow (2015). The Yamphu data is extensively discussed by Lacy (2006), who draws on the description by Rutgers (1998).
Considerations for discussion
- What is the precise nature of the effects observable in this data? How can we use representational machinery to capture them?
- Are non-representational solutions viable in such cases? What would make us choose between representational and non-representational approaches?
- Herce (2020) argues that despite the phonological predictability of the patterns, they should be considered morphological in nature rather than phonological. How should we weigh this alternative against the phonological analysis?