Underspecification and phonological activity
Questions to start with
The key idea behind viewing size as a driver of phonological behaviour is that phonological activity corresponds to the presence of structure. Can we push this further?
- If presence of structure is an explanatory factor and phonological activity is epiphenomenal, what justifies presence or absence of structure?
- Where does phonological structure come from? What considerations, other than the patterns we are trying to explain, can support or militate against a representational proposal?
- The story of western phonology as usually told (cf. Anderson 2021) begins with the realization that phonological structure is built because of lexical contrast. It is contrast that plays this external grounding role on which the edifice of phonology is later built: distinctions that are contrastive matter, those that are not are redundant, and phonologists should not care. Seen this way, was it a wise move to abandon this view, as generative phonology tried to do?
Language background
We consider two examples here. First, we look at from Written (Classical) Manchu, a Tungusic language attested in writing from around the 16th-17th century. and from Xibe, a variety spoken by around 30,000 speakers (2000 census), currently in the Xinjiang but historically primarily in north-eastern China, which is closely related to the varieties that underlay written Manchu.
PDF version of the dataset.
Written Manchu
| Front | Central | Back |
|---|---|---|
| i | u | |
| ʊ | ||
| ə | ||
| ɔ | ||
| a |
The vowel [i] exerts a palatalizing influence on at least some preceding consonants.
Vowel harmony in Written Manchu
| Unsuffixed | Suffixed | Gloss | Suffix |
|---|---|---|---|
| xəxə | xəxə-ŋɡə | ‘woman’ | Genitive |
| kumun | kumu-ŋɡə | ‘noise’ | |
| aɢa | aɢa-ŋɢa | ‘rain’ | |
| itʂʰi | itʂʰi-ŋɢa | ‘direction’ | |
| tursun | tursu-ŋɢa | ‘form’ | |
| tʰɔn | tʰɔ-ŋɢa | ‘number’ | |
| susə | susə-tə- | ‘coarse’ | Causative |
| χʊrχa | χʊrχa-ta- | ‘fishing net’ | |
| tulpa | tulpa-ta- | ‘careless’ | |
| tʂili | tʂili-ta- | ‘anger’ | |
| silxi | silxi-ta | ‘envy’ | |
| xətʰu | xətʰu-kən | ‘stocky’ | Attenuative |
| ɡulu | ɡulu-kən | ‘plain’ | |
| farχʊn | farχʊ-kan | ‘dark’ | |
| χʊtun | χʊtu-qan | ‘fast’ | |
| pəki | pəki-lə- | ‘firm’ | Causative |
| paqtʂʰin | paqtʂʰi-la- | ‘opponent’ | |
| xərə- | xərə-ku | ‘ladle out’ | Nominalization |
| paqtʰa- | paqtʰa-qʊ | ‘contain’ | |
| tʂʰili- | tʂʰili-qʊ | ‘choke’ | |
| səxəxun | səxə-xuri | ‘vertical’ | Augmentative |
| laqtaχʊn | laqta-χʊri | ‘drooping’ | |
| əmtʰə | əmtʰə-li | ‘one each’ | Adjectivization |
| taχa | taχa-li | ‘follow’ | |
| uli- | uli-xə | ‘string’ | Past |
| ana- | ana-χa | ‘push’ | |
| pu- | pu-xə | ‘give’ |
Questions:
- What is the distribution of velars and uvulars? How do they relate to vowels?
- What classes do the vowels fall into with respect to ATR alternations?
- Are any vowels exempt from patterns involving ATR?
| Unsuffixed | Suffixed | Gloss |
|---|---|---|
| kumun | kumu-ŋɡə | ‘noise’ |
| tursun | tursu-ŋɢa | ‘form’ |
| pɔtʂʰɔ | pɔtʂʰɔ-ŋɢɔ | ‘colour’ |
| χʊtun | χʊtu-qan | ‘fast’ |
| fɔχɔlɔn | fɔχɔlɔ-qɔn | ‘short’ |
| ɡulu | ɡulu-kən | ‘plain’ |
| tɔ- | tɔ-na- | ‘alight’ |
| tɔː- | tɔː-na- | ‘cross’ |
| pɔtɔ- | pɔtɔ-χɔ | ‘think’ |
| pu- | pu-χə | ‘give’ |
Questions:
- The data is a bit sparse, but can you make a generalization about when rounding harmony occurs?
- Which vowels trigger rounding? Which ones don’t?
Analysis
Given the above generalizations, what featural specifications can we justify?
| Feature | /i/ | /u/ | /ʊ/ | /ɔ/ | /ə/ | /a/ |
|---|---|---|---|---|---|---|
| [low] | ||||||
| [coronal] | ||||||
| [ATR] | ||||||
| [labial] |
Xibe
Xibe has no /ʊ/, but has acquired /ɛ y œ/, primarily by assimilation of original /a u ɔ/ by a following /i/.
| Front | Central | Back |
|---|---|---|
| i y | u | |
| ɛ œ | ə | ɔ |
| a |
Vowel harmony in Xibe
| Written Manchu | Xibe | Gloss | |
|---|---|---|---|
| a. | ɡətʰə-xə | ɡətʰə-xə | ‘awoke’ |
| uli-xə | uli-xə | ‘stringed’ | |
| ana-χa | anə-χə | ‘pushed’ | |
| ɢɔtʂʰi-χa | ɢɔçi-xə | ‘cherished’ | |
| b. | pu-xə | pu-xu | ‘gave’ |
| pɔtɔ-χɔ | pɔtu-χu | ‘thought’ | |
| c. | nətʂi-kən | nətçi-kən | ‘flat’ |
| ərtə-kən | ərtə-kən | ‘early’ | |
| ampa-qan | am-qən | ‘big’ | |
| χantʂi-qan | χantçi-qən | ‘near’ | |
| d. | dʐuʂuxu-kən | dʑyɕxu-kun | ‘sour’ |
| xətʰu-kən | xətʰu-kun | ‘stocky’ | |
| laptu-qan | lavtu-qun | ‘many’ | |
| farχʊ-qan | farχu-qun | ‘dark’ | |
| fɔχɔlɔ-qɔn | fœχulu-qun | ‘short’ | |
| ɔsɔχo-qɔn | ɔsχɔ-qun | ‘small’ |
Questions:
- What happened to [ATR] harmony in Xibe? What about the consonants?
- How does rounding harmony work? What are the relevant classes?
Analysis
Given the above generalizations, what featural specifications can we justify?
| Feature | /i/ | /u/ | /y/ | /ɛ/ | /œ/ | /ɔ/ | /ə/ | /a/ |
|---|---|---|---|---|---|---|---|---|
| [low] | ||||||||
| [coronal] | ||||||||
| [ATR] | ||||||||
| [labial] |
Considerations for discussion
- Having established what featural specifications can be supported from the alternations, can we justify from other considerations such as contrast?
- Compare the representational solutions to the Manchu problems to the representational analysis of Nivkh on Monday. What are the alternative approaches to both?
- Does ‘representational reasoning’ carry all the explanatory burden in the Manchu case? What is missing? Do we need anything else to give a complete account? How do we decide?
Sources
The phonological patterns of Written Manchu and Xibe are discussed extensively in the study by Zhang (1996). The analyses presented here come from Dresher & Zhang (2005).