Markedness: diagnostics and structure
Plan for today
- Markedness as a structural property
- Universalizing markedness
- Markedness in SPE
- Diagnosing markedness
- How universal is markedness?
Markedness: the beginning
Classification of oppositions
Trubetzkoy (1939): three kinds of phonological oppositions
- Privative
-
Presence of a property vs. its absence. Example: nasal vs. non-nasal
- Equipollent
-
Presence of a property vs. the presence of a mutually exclusive property. Example: labial vs. dorsal
- Gradual
-
What it says on the tin. Example: vowel height
The ‘mark’
In a privative opposition, one member is characterized by the presence of a ‘mark’ (Merkmal) and the other is characterized by its absence.
This is a statement about the ‘logical structure of the opposition’, not the phonetics of it
This is an application of the ultimately Saussurean idea of ‘meaningful absence’: the absence of the mark has distinctive value because of the contrast with its presence.
This general structure is pervasive in language, most obviously in morphology.
In general, the presence of structure corresponds to more information relative to its absence.
Diagnosing markedness
- How would we know that a phoneme bears a mark? By its behaviour
- Some criteria:
- Implicational relationships in inventories
- Marked items are dispreferred in neutralization positions
- Greater token frequency of the unmarked1
1 With a link to Zipf’s Law!
Universalizing markedness
Markedness beyond the mark
- Jakobson (1941):
- Marked units are the last to be acquired…
- …and the first to go in language disorder…
- …and this is all universal
- General principles of markedness apply far beyond phonology
- Also influential was Greenberg (1966), linking markedness to the new field of linguistic typology
Does it even mean anything?
Markedness: An abstract measure of how unusual a particular linguistic structure is (Samuels 2011:208)
OK, that was bit mean
Hume (2011):
- Descriptive markedness:
-
An abstract relation holding over members of a set of observations displaying asymmetry, such that one subset is unmarked and the other is marked
- Theoretical markedness
-
A universal principle or law that guides language acquisition, loss, inventory structure, processes, rules, etc. toward the ‘unmarked’ form
- Markedness constraints
-
A technical term in Optimality Theory referring to a category of constraints that evaluate the well-formedness of output structures
Could you be more specific?
List from Rice (2007)
| Marked | Unmarked |
|---|---|
| less natural | more natural |
| more complex | simpler |
| more specific | more general |
| less common | more common |
| unexpected | expected |
| not basic | basic |
| less stable | stable |
| appear in few grammars | appear in more grammars |
| later in acquisition | earlier in acquisition |
| early loss in language deficit | late loss in language deficit |
| implies unmarked feature | implied by marked feature |
| harder to articulate | easier to articulate |
| perceptually more salient | perceptually less salient |
| smaller phonetic space | larger phonetic space |
Markedness in phonology: further developments
Markedness in SPE
Chomsky and Halle frame the problem as one of overgeneration: attested languages show markedness asymmetries, but their system does not provide for them. This ed them to introduce the machinery of markedness conventions, they unified with the redundancy rules we discussed yesterday. The reasoning is generally typological, although not based on very rigorous enquiry by modern standards.
Although markedness conventions, like redundancy rules and MSCs, mostly resided in the pre-phonological (lexical) component, they sometimes had to intervene in the operation of phonological grammar per se to make sure its output conformed to markedness theories (Chomsky & Halle used the device of ‘linking’ to achieve this).
Architecturally, markedness conventions added nothing new to representations: they were essentially rewrite rules of the form ‘if X then Y’, manipulating all the same feature values.
Markedness diagnostics
Or, what is it that are we trying to explain again? (Rice 2007)
- Emergence of the unmarked:
-
Neutralization
- Submergence of the unmarked:
-
Unmarked elements are preferred targets, dispreferred triggers
- Preservation of the marked
-
Marked elements are dispreferred targets, preferred triggers
- Transparency and blocking
-
In long-distance processes, unmarked elements are transparent, marked elements are blockers
A closer look at the diagnostics
Typical reasoning: unmarked coronals
- English assimilation: coronals undergo, non-coronals resist
- ba[ŋ] cuts but *alar[ŋ] clock
- Korean: coronals undergo and do not trigger
- /kot-palo/ → [koppalo]
- /pap-to/ → [papto], *[patto]
- /ip-ko/ → [ikko]
- Northern Sámi: place neutralization word-finally
| GEN.SG | NOM.SG | Gloss |
|---|---|---|
| rusttega | rusttet | ‘building’ |
| ustiba | ustit | ‘friend’ |
| rievssaha | rievssat | ‘ptarmigan’ |
| goaskima | goaskin | ‘eagle’ |
| čálána | čálán | ‘writer’ |
Emergence of the unmarked
This sense of ‘emergence of the unmarked’ is quite different from TETU as understood in OT
This is our old friend: the neutralization criterion.
In non-assimilatory neutralization, outcomes tend to be unmarked
- Final obstruent devoicing
- Vowel reduction:
- Centripetal to schwa (= minimally marked vowel)
- Centrifugal to [i u (a/ə)] (= relatively unmarked2 inventory)
- Coda weakening
2 Because well-dispersed and containing the most common vowels
Other criteria
- Default vowel epenthesis
- Default consonant epenthesis
So, what is the unmarked place?
This list relies heavily on work by Keren Rice, see Rice (2009), but some of this I got from various handouts
| Inventory | Examples | |
|---|---|---|
| Stops | /p/ | Godoberi, Lhasa Tibetan, Nimburan |
| /t/ | Finnish, Eastern Enontekiö Northern Sámi | |
| /k/ | Karasjok Northern Sámi, Ecuador Quechua (but also /n/) | |
| /ʔ/ | many | |
| Nasals | /m/ | Lhasa Tibetan, Sentani… |
| /n/ | Finnish, Koyukon, Sekani… | |
| /ŋ/ | tricky one! |
| Inventory | Examples | |
|---|---|---|
| Stops | /p t/ | Kiowa |
| /p k/ | German dialects, Korowai… | |
| /t k/ | Nanchang, Badimaya… | |
| /p ʔ/ | Jabêm | |
| /k ʔ/ | Yaw Burmese | |
| Nasals | /m n/ | Trio, Sonora Hiaki… |
| /m ŋ/ | Nganasan, Palauan… | |
| /n ŋ/ | various Sinitic |
Coda neutralization in alternations
| Infinitive | 3SG.PST | Gloss |
|---|---|---|
| lapma | labana | ‘seize’ |
| apma | abana | ‘come’ |
| jokma | joɡana | ‘search’ |
| pʰaʔma | pʰatana | ‘help’ |
| keʔma | ketana | ‘bring up’ |
| liʔma | litana | ‘plant’ |
| tʰuʔma | tʰurana | ‘sew’ |
| poʔma | porana | ‘topple’ |
- Preservation of the marked: /p k/ → [p k]
- Submergence of the unmarked: /t r/ → [ʔ]
- A markedness hierarchy
-
labial » dorsal » coronal » glottal
- The coda inventory
-
labial = dorsal »
coronal» glottal
Preservation of the marked
- Nganasan
- /koðaʔa-t/ \(\rightarrow\) [koðaʔaʔ] ‘they kill’
- /koðaʔa-t-uŋ/ \(\rightarrow\) [koðaʔaðuŋ] ‘they kill (it)’
- /tapkətə/ \(\rightarrow\) [tapkətə] ‘from there’
/t d/ \(\rightarrow\) [ʔ] in codas, but /p b/ are preserved
Submergence of the unmarked
- Assimilation: English and Korean above
- Modern Greek: front vowels are deleted in hiatus irrespective of the order
Transparency
| ATR | RTR | ||
|---|---|---|---|
| òɡùrò | ‘spurtle’ | ɔrúkɔ | ‘name’ |
| eúrò | ‘bitter-leaf’ | ɛ̀lùbɔ́ | ‘yam flour’ |
| oríwo | ‘boil, tumour’ | ɔdídɛ | ‘parrot’ |
| èbúté | ‘harbour’ | ɛúrɛ́ | ‘goat’ |
Blocking
| Place | Gloss | |
|---|---|---|
| Laryngeal | mihiˈla | ‘west’ |
| weˈʔej | ‘yonder’ | |
| waʔali | ‘cane’ | |
| soʔhoj | ‘seal’ | |
| huˈʔul | ‘a while ago’ | |
| Supraralaryngeal | biʔdu | ‘acorn’ |
| hoja | ‘scoring sticks’ | |
| kʼaʔli | ‘between’ | |
| hoˈpʰune | ‘white-footed mouse’ |
Encoding markedness
Back to the ‘mark’
- In the SPE system, the difference between ‘marked’ and ‘unmarked’ is stipulated by markedness conventions, and justified by appeal to substance
- There is no clear link between how markedness is represented and what kinds of behaviour it is associated with
- Recall that in Praguian phonology markedness was defined in terms of size or complexity
- Marked = merkmaltragend
Markedness and size
What traditional markedness diagnostics are picking up is the presence of structure
Segments consist not of feature-value bundles but of unary features
There are many different traditions within the broad unary family, but arguably they are all unified by aiming to reflect this insight.
The coda condition: Japanese
- Option 1: first half of a geminate
- gakkoo ‘school’
- tossa ‘impulsively’
- kappa ‘legendary being’
- NB! Enforced by alternation
- /bet+kaku/ → [bekkakɯ] ‘different style’
- Option 2: the nasal
- Weak nasal prepausally: [hoN] ‘book’
- Place assimilation before a consonant: [hoŋ ka] ‘book-Q’
- Assimilated glide before a vowel: [hoĩ irɯ], [hoõ o], [hoɯ̃ arɯ]
Coda condition: there is either no place (the nasal in absolute-final position) or place borrowed from the following onset (obstruents, nasal before other segment)
More concretely: no place features allowed in coda
‘More marked’ means ‘bigger’
- Coda conditions
- Final/coda devoicing
- Vowel epenthesis
- Vowel reduction
- Centripetal: remove place specifications
- Centrifugal: remove more complex specifications
- Raising: remove |open|
- Implicational universals in inventories
‘Bigger’ means more phonologically active
- Assimilation triggers
- Harmony triggers
- Harmony blockers
‘Smaller’ means less phonologically active
- Assimilation non-triggers
- Assimilation targets
- Transparent segments in harmony
Some challenges
- Equal in markedness = equal in size?
- What about contextual markedness?
- Where does structure come from?