SLIDE 1 Trade-offs in the contrastive hierarchy: Voicing versus continuancy in Slavic
Daniel Currie Hall University of Toronto Saint Mary’s University NELS • Concordia University • October
SLIDE 2 Outline
Our approach to phonological representations: The Successive Division Algorithm (SDA) Contrast and phonological activity: What does the SDA actually predict? Applying the SDA to Russian
L Revising Halle’s hierarchy L Consequences of the change
Evidence elsewhere in Slavic
SLIDE 3
Our approach to phonological representations
Two components of a theory of phonemic contrast:
. . The Contrastivist Hypothesis: Only contrastive features are
phonologically active.
. . The Successive Division Algorithm: Contrastive features are
assigned by recursively dividing the underlying inventory.
Dresher (, ); Hall (, forthcoming)
SLIDE 4
Our approach to phonological representations
Two components of a theory of phonemic contrast:
. . The Contrastivist Hypothesis: Only contrastive features are
phonologically active.
. The Successive Division Algorithm: Contrastive features are
assigned by recursively dividing the underlying inventory.
Dresher (, ); Hall (, forthcoming)
SLIDE 5
Our approach to phonological representations
Two components of a theory of phonemic contrast:
. . The Contrastivist Hypothesis: Only contrastive features are
phonologically active.
. . The Successive Division Algorithm: Contrastive features are
assigned by recursively dividing the underlying inventory.
Dresher (, ); Hall (, forthcoming)
SLIDE 6 The Successive Division Algorithm
- a. Begin with no feature specifications: assume all sounds are
allophones of a single undifferentiated phoneme.
- b. If the set is found to consist of more than one contrasting
member, select a feature and divide the set into as many subsets as the feature allows for.
- c. Repeat step (b) in each subset: keep dividing up the inventory
into sets, applying successive features in turn, until every set has
. .
V
.
ə
. [high] . [high] . a . ɨ .
[back]
.
[back]
. i . u
Dresher (: )
SLIDE 7 The Successive Division Algorithm
- a. Begin with no feature specifications: assume all sounds are
allophones of a single undifferentiated phoneme.
- b. If the set is found to consist of more than one contrasting
member, select a feature and divide the set into as many subsets as the feature allows for.
- c. Repeat step (b) in each subset: keep dividing up the inventory
into sets, applying successive features in turn, until every set has
. .
V
.
ə
. [high] . [high] . a . ɨ .
[back]
.
[back]
. i . u
Dresher (: )
SLIDE 8 The Successive Division Algorithm
- a. Begin with no feature specifications: assume all sounds are
allophones of a single undifferentiated phoneme.
- b. If the set is found to consist of more than one contrasting
member, select a feature and divide the set into as many subsets as the feature allows for.
- c. Repeat step (b) in each subset: keep dividing up the inventory
into sets, applying successive features in turn, until every set has
.
V
.
ə
. [high] . [high] . a . ɨ .
[back]
.
[back]
. i . u
Dresher (: )
SLIDE 9 The Successive Division Algorithm
- a. Begin with no feature specifications: assume all sounds are
allophones of a single undifferentiated phoneme.
- b. If the set is found to consist of more than one contrasting
member, select a feature and divide the set into as many subsets as the feature allows for.
- c. Repeat step (b) in each subset: keep dividing up the inventory
into sets, applying successive features in turn, until every set has
.
V
.
ə
. [high] . [high] . a . ɨ .
[back]
.
[back]
. i . u
Dresher (: )
SLIDE 10 The Successive Division Algorithm
- a. Begin with no feature specifications: assume all sounds are
allophones of a single undifferentiated phoneme.
- b. If the set is found to consist of more than one contrasting
member, select a feature and divide the set into as many subsets as the feature allows for.
- c. Repeat step (b) in each subset: keep dividing up the inventory
into sets, applying successive features in turn, until every set has
.
V
.
ə
. [high] . [high] . a . ɨ .
[back]
.
[back]
. i . u
Dresher (: )
SLIDE 11 The Successive Division Algorithm
- a. Begin with no feature specifications: assume all sounds are
allophones of a single undifferentiated phoneme.
- b. If the set is found to consist of more than one contrasting
member, select a feature and divide the set into as many subsets as the feature allows for.
- c. Repeat step (b) in each subset: keep dividing up the inventory
into sets, applying successive features in turn, until every set has
.
V
.
ə
. [high] . [high] . a . ɨ .
[back]
.
[back]
. i . u
Dresher (: )
SLIDE 12 The Successive Division Algorithm
- a. Begin with no feature specifications: assume all sounds are
allophones of a single undifferentiated phoneme.
- b. If the set is found to consist of more than one contrasting
member, select a feature and divide the set into as many subsets as the feature allows for.
- c. Repeat step (b) in each subset: keep dividing up the inventory
into sets, applying successive features in turn, until every set has
.
V
.
ə
. [high] . [high] . a . ɨ .
[back]
.
[back]
. i . u
Dresher (: )
SLIDE 13 Specifications depend on hierarchical order
Two possible ways of dividing the vowel inventory /i u a/ with [high] and [back] :
high Q back back Q high . .
V
. [high] . [high] . a .
[back]
.
[back]
. i . u . .
V
. [back] . [back] . i .
[high]
.
[high]
. u . a i u a high
i u a high ∅
SLIDE 14 Specifications depend on hierarchical order
Two possible ways of dividing the vowel inventory /i u a/ with [high] and [back]:
high Q back back Q high . .
V
. [high] . [high] . a .
[back]
.
[back]
. i . u . .
V
. [back] . [back] . i .
[high]
.
[high]
. u . a i u a high
i u a high ∅
SLIDE 15 Specifications depend on hierarchical order
Two possible ways of dividing the vowel inventory /i u a/ with [high] and [back]:
high Q back back Q high .
V
. [high] . [high] . a .
[back]
.
[back]
. i . u .
V
. [back] . [back] . i .
[high]
.
[high]
. u . a i u a high
i u a high ∅
SLIDE 16 Specifications depend on hierarchical order
Two possible ways of dividing the vowel inventory /i u a/ with [high] and [back]:
high Q back back Q high .
V
. [high] . [high] . a .
[back]
.
[back]
. i . u .
V
. [back] . [back] . i .
[high]
.
[high]
. u . a i u a high
i u a high ∅
SLIDE 17 Specifications depend on hierarchical order
Two possible ways of dividing the vowel inventory /i u a/ with [high] and [back]:
high Q back back Q high .
V
. [high] . [high] . a .
[back]
.
[back]
. i . u .
V
. [back] . [back] . i .
[high]
.
[high]
. u . a i u a high
i u a high ∅
SLIDE 18 Specifications depend on hierarchical order
Two possible ways of dividing the vowel inventory /i u a/ with [high] and [back]:
high Q back back Q high .
V
. [high] . [high] . a .
[back]
.
[back]
. i . u .
V
. [back] . [back] . i .
[high]
.
[high]
. u . a i u a high
i u a high ∅
SLIDE 19 Specifications depend on hierarchical order
Two possible ways of dividing the vowel inventory /i u a/ with [high] and [back]:
high Q back back Q high .
V
. [high] . [high] . a .
[back]
.
[back]
. i . u .
V
. [back] . [back] . i .
[high]
.
[high]
. u . a i u a high
i u a high ∅
SLIDE 20 Contrast and phonological activity
The Contrastivist Hypothesis: Only contrastive features are active in the phonology. How do we know which features are contrastive? The SDA. But if the order of features can vary, how do we know what the right hierarchy is for any given language?
L If we observe that a feature is active, then by hypothesis it must
be contrastive.
L So every active feature must be high enough in the hierarchy to
be specified on the relevant segments.
Is this circular? (various reviewers, passim) No.
The SDA and the Contrastivist Hypothesis make testable predictions.
Hall (: )
SLIDE 21 Contrast and phonological activity
The Contrastivist Hypothesis: Only contrastive features are active in the phonology. How do we know which features are contrastive? The SDA. But if the order of features can vary, how do we know what the right hierarchy is for any given language?
L If we observe that a feature is active, then by hypothesis it must
be contrastive.
L So every active feature must be high enough in the hierarchy to
be specified on the relevant segments.
Is this circular? (various reviewers, passim) No.
The SDA and the Contrastivist Hypothesis make testable predictions.
SLIDE 22 Contrast and phonological activity
The Contrastivist Hypothesis: Only contrastive features are active in the phonology. How do we know which features are contrastive? The SDA. But if the order of features can vary, how do we know what the right hierarchy is for any given language?
L If we observe that a feature is active, then by hypothesis it must
be contrastive.
L So every active feature must be high enough in the hierarchy to
be specified on the relevant segments.
Is this circular? (various reviewers, passim) No.
The SDA and the Contrastivist Hypothesis make testable predictions.
SLIDE 23 Contrast and phonological activity
The Contrastivist Hypothesis: Only contrastive features are active in the phonology. How do we know which features are contrastive? The SDA. But if the order of features can vary, how do we know what the right hierarchy is for any given language?
L If we observe that a feature is active, then by hypothesis it must
be contrastive.
L So every active feature must be high enough in the hierarchy to
be specified on the relevant segments.
Is this circular? (various reviewers, passim) No.
The SDA and the Contrastivist Hypothesis make testable predictions.
SLIDE 24 Contrast and phonological activity
The Contrastivist Hypothesis: Only contrastive features are active in the phonology. How do we know which features are contrastive? The SDA. But if the order of features can vary, how do we know what the right hierarchy is for any given language?
L If we observe that a feature is active, then by hypothesis it must
be contrastive.
L So every active feature must be high enough in the hierarchy to
be specified on the relevant segments.
Is this circular? (various reviewers, passim) No.
The SDA and the Contrastivist Hypothesis make testable predictions.
SLIDE 25 Contrast and phonological activity
The Contrastivist Hypothesis: Only contrastive features are active in the phonology. How do we know which features are contrastive? The SDA. But if the order of features can vary, how do we know what the right hierarchy is for any given language?
L If we observe that a feature is active, then by hypothesis it must
be contrastive.
L So every active feature must be high enough in the hierarchy to
be specified on the relevant segments.
Is this circular? (various reviewers, passim) No.
The SDA and the Contrastivist Hypothesis make testable predictions.
see also Blaho (), de Lacy ()
SLIDE 26 Contrast and phonological activity
The Contrastivist Hypothesis: Only contrastive features are active in the phonology. How do we know which features are contrastive? The SDA. But if the order of features can vary, how do we know what the right hierarchy is for any given language?
L If we observe that a feature is active, then by hypothesis it must
be contrastive.
L So every active feature must be high enough in the hierarchy to
be specified on the relevant segments.
Is this circular? (various reviewers, passim) No.
The SDA and the Contrastivist Hypothesis make testable predictions.
SLIDE 27 Contrast and phonological activity
The Contrastivist Hypothesis: Only contrastive features are active in the phonology. How do we know which features are contrastive? The SDA. But if the order of features can vary, how do we know what the right hierarchy is for any given language?
L If we observe that a feature is active, then by hypothesis it must
be contrastive.
L So every active feature must be high enough in the hierarchy to
be specified on the relevant segments.
Is this circular? (various reviewers, passim) No.
The SDA and the Contrastivist Hypothesis make testable predictions.
SLIDE 28
Contrast and phonological activity
Given just a phonological inventory… We can’t predict exactly what the feature specifications are. The SDA is not deterministic. We can make predictions about how many features can be specified/active. We can make predictions about trade-offs between potentially contrastive features.
SLIDE 29
Contrast and phonological activity
Given just a phonological inventory… We can’t predict exactly what the feature specifications are. The SDA is not deterministic. We can make predictions about how many features can be specified/active. We can make predictions about trade-offs between potentially contrastive features.
SLIDE 30
Contrast and phonological activity
Given just a phonological inventory… We can’t predict exactly what the feature specifications are. The SDA is not deterministic. We can make predictions about how many features can be specified/active. We can make predictions about trade-offs between potentially contrastive features.
SLIDE 31
Contrast and phonological activity
Given just a phonological inventory… We can’t predict exactly what the feature specifications are. The SDA is not deterministic. We can make predictions about how many features can be specified/active. We can make predictions about trade-offs between potentially contrastive features.
SLIDE 32
Contrast and phonological activity
Back to our three-vowel example:
. V . [high] . [high] . a .
[back]
.
[back]
. i . u . V . [back] . [back] . i .
[high]
.
[high]
. u . a
We can’t use more than two features to specify three vowels. We can have [high] on /i/, or [back] on /a/, but not both. Our predictions are not [F] will be active and [G] will not, but rather if [F] is active then [G] cannot be.
SLIDE 33
Contrast and phonological activity
Back to our three-vowel example:
. V . [high] . [high] . a .
[back]
.
[back]
. i . u . V . [back] . [back] . i .
[high]
.
[high]
. u . a
We can’t use more than two features to specify three vowels. We can have [high] on /i/, or [back] on /a/, but not both. Our predictions are not [F] will be active and [G] will not, but rather if [F] is active then [G] cannot be.
SLIDE 34
Contrast and phonological activity
Back to our three-vowel example:
. V . [high] . [high] . a .
[back]
.
[back]
. i . u . . V . [back] . [back] . i . . .
[high]
.
[high]
. u . a
We can’t use more than two features to specify three vowels. We can have [high] on /i/, or [back] on /a/, but not both. Our predictions are not [F] will be active and [G] will not, but rather if [F] is active then [G] cannot be.
SLIDE 35
Contrast and phonological activity
Back to our three-vowel example:
. V . [high] . [high] . a .
[back]
.
[back]
. i . u . . V . [back] . [back] . i . . .
[high]
.
[high]
. u . a
We can’t use more than two features to specify three vowels. We can have [high] on /i/, or [back] on /a/, but not both. Our predictions are not [F] will be active and [G] will not, but rather if [F] is active then [G] cannot be.
SLIDE 36
Contrast and phonological activity
Back to our three-vowel example:
. V . [high] . [high] . a .
[back]
.
[back]
. i . u . . V . [back] . [back] . i . . .
[high]
.
[high]
. u . a
We can’t use more than two features to specify three vowels. We can have [high] on /i/, or [back] on /a/, but not both. Our predictions are not [F] will be active and [G] will not, but rather if [F] is active then [G] cannot be.
SLIDE 37
Contrast and phonological activity
Back to our three-vowel example:
. V . [high] . [high] . a .
[back]
.
[back]
. i . u . . V . [back] . [back] . i . . .
[high]
.
[high]
. u . a
We can’t use more than two features to specify three vowels. We can have [high] on /i/, or [back] on /a/, but not both. Our predictions are not [F] will be active and [G] will not, but rather if [F] is active then [G] cannot be.
SLIDE 38
Contrast and phonological activity
Back to our three-vowel example:
. V . [high] . [high] . a .
[back]
.
[back]
. i . u . . V . [back] . [back] . i . . .
[high]
.
[high]
. u . a
We can’t use more than two features to specify three vowels. We can have [high] on /i/, or [back] on /a/, but not both. Our predictions are not [F] will be active and [G] will not, but rather if [F] is active then [G] cannot be.
SLIDE 39
The contrastive hierarchy in Russian
Russian offers an exemplary case of a trade-off in the contrastive hierarchy. Our starting point is Halle (). In SPR, Halle uses a contrastive hierarchy , but does not adopt the Contrastivist Hypothesis. For Halle, the hierarchy primarily serves to simplify underlying representations. Redundant features are filled in during the derivation, allowing them to be phonologically active.
SLIDE 40
The contrastive hierarchy in Russian
Russian offers an exemplary case of a trade-off in the contrastive hierarchy. Our starting point is Halle (). In SPR, Halle uses a contrastive hierarchy , but does not adopt the Contrastivist Hypothesis. For Halle, the hierarchy primarily serves to simplify underlying representations. Redundant features are filled in during the derivation, allowing them to be phonologically active.
SLIDE 41
The contrastive hierarchy in Russian
Russian offers an exemplary case of a trade-off in the contrastive hierarchy. Our starting point is Halle (). In SPR, Halle uses a contrastive hierarchy: , but does not adopt the Contrastivist Hypothesis. For Halle, the hierarchy primarily serves to simplify underlying representations. Redundant features are filled in during the derivation, allowing them to be phonologically active.
SLIDE 42
The contrastive hierarchy in Russian
Russian offers an exemplary case of a trade-off in the contrastive hierarchy. Our starting point is Halle (). In SPR, Halle uses a contrastive hierarchy: , but does not adopt the Contrastivist Hypothesis. . . H (: ) . “The hierarchy of features seems to provide an explanation for the intuition that not all features are equally central to a given phonological system.” For Halle, the hierarchy primarily serves to simplify underlying representations. Redundant features are filled in during the derivation, allowing them to be phonologically active.
SLIDE 43
The contrastive hierarchy in Russian
Russian offers an exemplary case of a trade-off in the contrastive hierarchy. Our starting point is Halle (). In SPR, Halle uses a contrastive hierarchy, but does not adopt the Contrastivist Hypothesis. . . H (: ) . “[P rules] specify all features which play no distinctive role in the language but are not randomly distributed.” For Halle, the hierarchy primarily serves to simplify underlying representations. Redundant features are filled in during the derivation, allowing them to be phonologically active.
SLIDE 44
The contrastive hierarchy in Russian
Russian offers an exemplary case of a trade-off in the contrastive hierarchy. Our starting point is Halle (). In SPR, Halle uses a contrastive hierarchy, but does not adopt the Contrastivist Hypothesis. . . H (: ) . “[P rules] specify all features which play no distinctive role in the language but are not randomly distributed.” For Halle, the hierarchy primarily serves to simplify underlying representations. Redundant features are filled in during the derivation, allowing them to be phonologically active.
SLIDE 45
The contrastive hierarchy in Russian
Russian offers an exemplary case of a trade-off in the contrastive hierarchy. Our starting point is Halle (). In SPR, Halle uses a contrastive hierarchy, but does not adopt the Contrastivist Hypothesis. . . H (: ) . “[P rules] specify all features which play no distinctive role in the language but are not randomly distributed.” For Halle, the hierarchy primarily serves to simplify underlying representations. Redundant features are filled in during the derivation, allowing them to be phonologically active.
SLIDE 46
Voicing assimilation
Obstruents in clusters undergo regressive assimilation. Assimilation involves both voicing… …and devoicing. /__.: s-jexatʲ ‘move out’ iz-laɡatʲ ‘set out’ /__.: s-prositʲ ‘ask (for)’ is-klʲuʧatʲ ‘exclude’ /__.: z-dʲelatʲ ‘do’ iz-ɡnatʲ ‘drive out’ So [voice] is phonologically active on obstruents. (And it’s not active on sonorants.)
examples from Padgett ()
SLIDE 47
Voicing assimilation
Obstruents in clusters undergo regressive assimilation. Assimilation involves both voicing… …and devoicing. /__.: s-jexatʲ ‘move out’ iz-laɡatʲ ‘set out’ /__.: s-prositʲ ‘ask (for)’ is-klʲuʧatʲ ‘exclude’ /__.: z-dʲelatʲ ‘do’ iz-ɡnatʲ ‘drive out’ So [voice] is phonologically active on obstruents. (And it’s not active on sonorants.)
examples from Padgett ()
SLIDE 48
Voicing assimilation
Obstruents in clusters undergo regressive assimilation. Assimilation involves both voicing… …and devoicing. /__.: s-jexatʲ ‘move out’ iz-laɡatʲ ‘set out’ /__.: s-prositʲ ‘ask (for)’ is-klʲuʧatʲ ‘exclude’ /__.: z-dʲelatʲ ‘do’ iz-ɡnatʲ ‘drive out’ So [voice] is phonologically active on obstruents. (And it’s not active on sonorants.)
examples from Padgett ()
SLIDE 49
Voicing assimilation
Obstruents in clusters undergo regressive assimilation. Assimilation involves both voicing… …and devoicing. /__.: s-jexatʲ ‘move out’ iz-laɡatʲ ‘set out’ /__.: s-prositʲ ‘ask (for)’ is-klʲuʧatʲ ‘exclude’ /__.: z-dʲelatʲ ‘do’ iz-ɡnatʲ ‘drive out’ So [voice] is phonologically active on obstruents. (And it’s not active on sonorants.)
examples from Padgett ()
SLIDE 50
Voicing assimilation
Obstruents in clusters undergo regressive assimilation. Assimilation involves both voicing… …and devoicing. /__.: s-jexatʲ ‘move out’ iz-laɡatʲ ‘set out’ /__.: s-prositʲ ‘ask (for)’ is-klʲuʧatʲ ‘exclude’ /__.: z-dʲelatʲ ‘do’ iz-ɡnatʲ ‘drive out’ So [voice] is phonologically active on obstruents. (And it’s not active on sonorants.)
examples from Padgett ()
SLIDE 51
Voicing assimilation
Obstruents in clusters undergo regressive assimilation. Assimilation involves both voicing… …and devoicing. /__.: s-jexatʲ ‘move out’ iz-laɡatʲ ‘set out’ /__.: s-prositʲ ‘ask (for)’ is-klʲuʧatʲ ‘exclude’ /__.: z-dʲelatʲ ‘do’ iz-ɡnatʲ ‘drive out’ So [voice] is phonologically active on obstruents. (And it’s not active on sonorants.)
examples from Padgett ()
SLIDE 52
Voicing assimilation
Obstruents in clusters undergo regressive assimilation. Assimilation involves both voicing… …and devoicing. /__.: s-jexatʲ ‘move out’ iz-laɡatʲ ‘set out’ /__.: s-prositʲ ‘ask (for)’ is-klʲuʧatʲ ‘exclude’ /__.: z-dʲelatʲ ‘do’ iz-ɡnatʲ ‘drive out’ So [voice] is phonologically active on obstruents. (And it’s not active on sonorants.)
examples from Padgett ()
SLIDE 53
Voicing assimilation: Features
Most Russian obstruents come in voiced/voiceless pairs, and sonorants are all voiced. So if [sonorant] (or the equivalent) takes scope over [voice], voicing will be specified on obstruents but not on sonorants. Schematically: . C . [sonorant] . [sonorant] . n .
[voice]
.
[voice]
. t . d
(For Halle (), sonorants are distinguished by [vocalic], [consonantal], or [nasal].)
SLIDE 54
Voicing assimilation: Features
Most Russian obstruents come in voiced/voiceless pairs, and sonorants are all voiced. So if [sonorant] (or the equivalent) takes scope over [voice], voicing will be specified on obstruents but not on sonorants. Schematically: . C . [sonorant] . [sonorant] . n .
[voice]
.
[voice]
. t . d
(For Halle (), sonorants are distinguished by [vocalic], [consonantal], or [nasal].)
SLIDE 55
Voicing assimilation: Features
Most Russian obstruents come in voiced/voiceless pairs, and sonorants are all voiced. So if [sonorant] (or the equivalent) takes scope over [voice], voicing will be specified on obstruents but not on sonorants. Schematically: . C . [sonorant] . [sonorant] . n .
[voice]
.
[voice]
. t . d
(For Halle (), sonorants are distinguished by [vocalic], [consonantal], or [nasal].)
SLIDE 56 Voicing assimilation: The unpaired obstruents
For pairs like /t/ and /d/, [voice] must be contrastive. No matter how low [voice] is in the hierarchy, there’s no
- ther feature that could distinguish them.
But Russian also has three unpaired voiceless obstruents. () p pʲ t tʲ k kʲ b bʲ d dʲ ɡ ʦ ʧ f fʲ s sʲ ʃ x v vʲ z zʲ ʒ These unpaired obstruents were key to Halle’s (; ) argument against the structuralist separation of morphophonemic and allophonic patterns.
see also Dresher ()
SLIDE 57 Voicing assimilation: The unpaired obstruents
For pairs like /t/ and /d/, [voice] must be contrastive. No matter how low [voice] is in the hierarchy, there’s no
- ther feature that could distinguish them.
But Russian also has three unpaired voiceless obstruents. () p pʲ t tʲ k kʲ b bʲ d dʲ ɡ ʦ ʧ f fʲ s sʲ ʃ x v vʲ z zʲ ʒ These unpaired obstruents were key to Halle’s (; ) argument against the structuralist separation of morphophonemic and allophonic patterns.
see also Dresher ()
SLIDE 58 Voicing assimilation: The unpaired obstruents
For pairs like /t/ and /d/, [voice] must be contrastive. No matter how low [voice] is in the hierarchy, there’s no
- ther feature that could distinguish them.
But Russian also has three unpaired voiceless obstruents. () p pʲ t tʲ k kʲ b bʲ d dʲ ɡ ʦ ʧ f fʲ s sʲ ʃ x v vʲ z zʲ ʒ These unpaired obstruents were key to Halle’s (; ) argument against the structuralist separation of morphophonemic and allophonic patterns.
see also Dresher ()
SLIDE 59 Voicing assimilation: The unpaired obstruents
For pairs like /t/ and /d/, [voice] must be contrastive. No matter how low [voice] is in the hierarchy, there’s no
- ther feature that could distinguish them.
But Russian also has three unpaired voiceless obstruents. () p pʲ t tʲ k kʲ b bʲ d dʲ ɡ ʦ ʧ f fʲ s sʲ ʃ x v vʲ z zʲ ʒ These unpaired obstruents were key to Halle’s (; ) argument against the structuralist separation of morphophonemic and allophonic patterns.
see also Dresher ()
SLIDE 60 Voicing assimilation: The unpaired obstruents
For pairs like /t/ and /d/, [voice] must be contrastive. No matter how low [voice] is in the hierarchy, there’s no
- ther feature that could distinguish them.
But Russian also has three unpaired voiceless obstruents. () p pʲ t tʲ k kʲ b bʲ d dʲ ɡ ʦ ʧ f fʲ s sʲ ʃ x v vʲ z zʲ ʒ These unpaired obstruents were key to Halle’s (; ) argument against the structuralist separation of morphophonemic and allophonic patterns.
see also Dresher ()
SLIDE 61 Voicing assimilation: The unpaired obstruents
Unpaired /ʦ ʧ x/ undergo regressive assimilatory voicing:
‘father’ mox ‘moss’
‘father was’ moɣ bɨl ‘moss was’ ʒeʧ lʲi ‘should one burn?’ ʒeʤ bɨ ‘were one to burn’
examples from Halle (), Timberlake ()
SLIDE 62 Voicing assimilation: The unpaired obstruents
Unpaired /ʦ ʧ x/ undergo regressive assimilatory voicing:
‘father’ mox ‘moss’
‘father was’ moɣ bɨl ‘moss was’ ʒeʧ lʲi ‘should one burn?’ ʒeʤ bɨ ‘were one to burn’
examples from Halle (), Timberlake ()
SLIDE 63 Voicing assimilation: The unpaired obstruents
Unpaired /ʦ ʧ x/ undergo regressive assimilatory voicing:
‘father’ mox ‘moss’
‘father was’ moɣ bɨl ‘moss was’ ʒeʧ lʲi ‘should one burn?’ ʒeʤ bɨ ‘were one to burn’ (Thus Halle’s argument: If processes that produce alternations between phonemes are strictly separate from allophony, then there is no unified account of voicing assimilation.)
examples from Halle (), Timberlake ()
SLIDE 64 Voicing assimilation: The unpaired obstruents
Unpaired /ʦ ʧ x/ undergo regressive assimilatory voicing:
‘father’ mox ‘moss’
‘father was’ moɣ bɨl ‘moss was’ ʒeʧ lʲi ‘should one burn?’ ʒeʤ bɨ ‘were one to burn’ They also trigger regressive assimilatory devoicing: bʲez ozʲera ‘without a lake’ bʲes xlʲeba ‘without bread’ bʲes ʦenɨ ‘without price’ bʲes ʧestʲi ‘without honour’
examples from Halle (), Timberlake (), Calabrese ()
SLIDE 65
Specifying the unpaired obstruents
Since /ʦ ʧ x/ act like other [voice] obstruents, it would make sense for them to be specified as [voice]. But this is not what Halle does.
contrastive hierarchy for [consonantal] phonemes fron Halle (: )
SLIDE 66
Specifying the unpaired obstruents
Since /ʦ ʧ x/ act like other [voice] obstruents, it would make sense for them to be specified as [voice]. But this is not what Halle does.
contrastive hierarchy for [consonantal] phonemes fron Halle (: )
SLIDE 67
Specifying the unpaired obstruents
Since /ʦ ʧ x/ act like other [voice] obstruents, it would make sense for them to be specified as [voice]. But this is not what Halle does.
contrastive hierarchy for [consonantal] phonemes fron Halle (: )
SLIDE 68
Specifying the unpaired obstruents
Since /ʦ ʧ x/ act like other [voice] obstruents, it would make sense for them to be specified as [voice]. But this is not what Halle does.
contrastive hierarchy for [consonantal] phonemes fron Halle (: )
SLIDE 69
Specifying the unpaired obstruents
In Halle’s hierarchy: [low tonality] Q [continuant] Q [voiced] Q [sharped]
Strident dentals: Palatals and velars: . /ʦ s sʲ z zʲ/ . . [cont] ʦ . . [cont] . . [voice] . . [sharp] s . . [sharp] sʲ . . [voice] . . [sharp] z . . [sharp] zʲ . /ʧ ʃ ʒ k kʲ x ɡ/ . . [low ton] . . [cont] ʧ . . [cont] . . [voice] ʃ . . [voice] ʒ . . [low ton] . . [cont] . . [voice] . . [sharp] k . . [sharp] kʲ . . [voice] ɡ . . [cont] x
[continuant] cuts off /ʦ/, /ʧ/, and /x/ before [voiced] can be assigned to them.
SLIDE 70
Specifying the unpaired obstruents
In Halle’s hierarchy: [low tonality] Q [continuant] Q [voiced] Q [sharped]
Strident dentals: Palatals and velars: . /ʦ s sʲ z zʲ/ . . [cont] ʦ . . [cont] . . [voice] . . [sharp] s . . [sharp] sʲ . . [voice] . . [sharp] z . . [sharp] zʲ . /ʧ ʃ ʒ k kʲ x ɡ/ . . [low ton] . . [cont] ʧ . . [cont] . . [voice] ʃ . . [voice] ʒ . . [low ton] . . [cont] . . [voice] . . [sharp] k . . [sharp] kʲ . . [voice] ɡ . . [cont] x
[continuant] cuts off /ʦ/, /ʧ/, and /x/ before [voiced] can be assigned to them.
SLIDE 71
Specifying the unpaired obstruents
In Halle’s hierarchy: [low tonality] Q [continuant] Q [voiced] Q [sharped]
Strident dentals: Palatals and velars: . /ʦ s sʲ z zʲ/ . . [cont] ʦ . . [cont] . . [voice] . . [sharp] s . . [sharp] sʲ . . [voice] . . [sharp] z . . [sharp] zʲ . /ʧ ʃ ʒ k kʲ x ɡ/ . . [low ton] . . [cont] ʧ . . [cont] . . [voice] ʃ . . [voice] ʒ . . [low ton] . . [cont] . . [voice] . . [sharp] k . . [sharp] kʲ . . [voice] ɡ . . [cont] x
[continuant] cuts off /ʦ/, /ʧ/, and /x/ before [voiced] can be assigned to them.
SLIDE 72
Specifying the unpaired obstruents
In Halle’s hierarchy: [low tonality] Q [continuant] Q [voiced] Q [sharped]
Strident dentals: Palatals and velars: . /ʦ s sʲ z zʲ/ . . [cont] ʦ . . [cont] . . [voice] . . [sharp] s . . [sharp] sʲ . . [voice] . . [sharp] z . . [sharp] zʲ . /ʧ ʃ ʒ k kʲ x ɡ/ . . [low ton] . . [cont] ʧ . . [cont] . . [voice] ʃ . . [voice] ʒ . . [low ton] . . [cont] . . [voice] . . [sharp] k . . [sharp] kʲ . . [voice] ɡ . . [cont] x
[continuant] cuts off /ʦ/, /ʧ/, and /x/ before [voiced] can be assigned to them.
SLIDE 73
Specifying the unpaired obstruents
In Halle’s hierarchy: [low tonality] Q [continuant] Q [voiced] Q [sharped]
Strident dentals: Palatals and velars: . /ʦ s sʲ z zʲ/ . . [cont] ʦ . . [cont] . . [voice] . . [sharp] s . . [sharp] sʲ . . [voice] . . [sharp] z . . [sharp] zʲ . /ʧ ʃ ʒ k kʲ x ɡ/ . . [low ton] . . [cont] ʧ . . [cont] . . [voice] ʃ . . [voice] ʒ . . [low ton] . . [cont] . . [voice] . . [sharp] k . . [sharp] kʲ . . [voice] ɡ . . [cont] x
[continuant] cuts off /ʦ/, /ʧ/, and /x/ before [voiced] can be assigned to them.
SLIDE 74 Specifying the unpaired obstruents
For Halle, this is not a problem. The underlying representations of /ʦ ʧ x/ are kept simple, and redundant values for [voiced] can be filled in by rule. Rule P b Unless followed by an obstruent, /ʦ/, /ʧ/, and /x/ are voiceless. Rule P a If an obstruent cluster is followed […] by a sonorant, then with regard to voicing the cluster conforms to the last segment. E.g. без хлеба /bʲez xlʲeba/ [bʲes xlʲeba] ‘without bread’ : bʲez xlʲeba
[voiced] : g
rules: Halle (: –)
SLIDE 75 Specifying the unpaired obstruents
For Halle, this is not a problem. The underlying representations of /ʦ ʧ x/ are kept simple, and redundant values for [voiced] can be filled in by rule. Rule P b Unless followed by an obstruent, /ʦ/, /ʧ/, and /x/ are voiceless. Rule P a If an obstruent cluster is followed […] by a sonorant, then with regard to voicing the cluster conforms to the last segment. E.g. без хлеба /bʲez xlʲeba/ [bʲes xlʲeba] ‘without bread’ : bʲez xlʲeba
[voiced] : g
rules: Halle (: –)
SLIDE 76 Specifying the unpaired obstruents
For Halle, this is not a problem. The underlying representations of /ʦ ʧ x/ are kept simple, and redundant values for [voiced] can be filled in by rule. Rule P b Unless followed by an obstruent, /ʦ/, /ʧ/, and /x/ are voiceless. Rule P a If an obstruent cluster is followed […] by a sonorant, then with regard to voicing the cluster conforms to the last segment. E.g. без хлеба /bʲez xlʲeba/ [bʲes xlʲeba] ‘without bread’ : bʲez xlʲeba
[voiced] : g
rules: Halle (: –)
SLIDE 77 Specifying the unpaired obstruents
For Halle, this is not a problem. The underlying representations of /ʦ ʧ x/ are kept simple, and redundant values for [voiced] can be filled in by rule. Rule P b Unless followed by an obstruent, /ʦ/, /ʧ/, and /x/ are voiceless. Rule P a If an obstruent cluster is followed […] by a sonorant, then with regard to voicing the cluster conforms to the last segment. E.g. без хлеба /bʲez xlʲeba/ [bʲes xlʲeba] ‘without bread’ : bʲez xlʲeba
[voiced] : g
rules: Halle (: –)
SLIDE 78 Specifying the unpaired obstruents
For Halle, this is not a problem. The underlying representations of /ʦ ʧ x/ are kept simple, and redundant values for [voiced] can be filled in by rule. Rule P b Unless followed by an obstruent, /ʦ/, /ʧ/, and /x/ are voiceless. Rule P a If an obstruent cluster is followed […] by a sonorant, then with regard to voicing the cluster conforms to the last segment. E.g. без хлеба /bʲez xlʲeba/ [bʲes xlʲeba] ‘without bread’ : bʲez xlʲeba
[voiced] : g
rules: Halle (: –)
SLIDE 79 Specifying the unpaired obstruents
For Halle, this is not a problem. The underlying representations of /ʦ ʧ x/ are kept simple, and redundant values for [voiced] can be filled in by rule. Rule P b Unless followed by an obstruent, /ʦ/, /ʧ/, and /x/ are voiceless. Rule P a If an obstruent cluster is followed […] by a sonorant, then with regard to voicing the cluster conforms to the last segment. E.g. без хлеба /bʲez xlʲeba/ [bʲes xlʲeba] ‘without bread’: bʲez xlʲeba
[voiced] : g
rules: Halle (: –)
SLIDE 80 Specifying the unpaired obstruents
For Halle, this is not a problem. The underlying representations of /ʦ ʧ x/ are kept simple, and redundant values for [voiced] can be filled in by rule. Rule P b Unless followed by an obstruent, /ʦ/, /ʧ/, and /x/ are voiceless. Rule P a If an obstruent cluster is followed […] by a sonorant, then with regard to voicing the cluster conforms to the last segment. E.g. без хлеба /bʲez xlʲeba/ [bʲes xlʲeba] ‘without bread’: bʲez xlʲeba
[voiced] : g
rules: Halle (: –)
SLIDE 81 Specifying the unpaired obstruents
For Halle, this is not a problem. The underlying representations of /ʦ ʧ x/ are kept simple, and redundant values for [voiced] can be filled in by rule. Rule P b Unless followed by an obstruent, /ʦ/, /ʧ/, and /x/ are voiceless. Rule P a If an obstruent cluster is followed […] by a sonorant, then with regard to voicing the cluster conforms to the last segment. E.g. без хлеба /bʲez xlʲeba/ [bʲes xlʲeba] ‘without bread’: bʲez xlʲeba
[voiced] : g
rules: Halle (: –)
SLIDE 82
Specifying the unpaired obstruents
If we adopt the Contrastivist Hypothesis, then [voiced] must be contrastive on /ʦ ʧ x/ in order to be active. /ʦ ʧ x/ don’t have minimally different voiced counterparts */ʣ ʤ ɣ/ in the underlying inventory… …but they contrast with voiced obstruents in general. The flexibility of the SDA allows us to give [voiced] wider scope, so that it is specified on all Russian obstruents. But this doesn’t come for free. If [voiced] is promoted in the contrastive hierarchy, something else must be demoted. We predict a trade-off.
SLIDE 83
Specifying the unpaired obstruents
If we adopt the Contrastivist Hypothesis, then [voiced] must be contrastive on /ʦ ʧ x/ in order to be active. /ʦ ʧ x/ don’t have minimally different voiced counterparts */ʣ ʤ ɣ/ in the underlying inventory… …but they contrast with voiced obstruents in general. The flexibility of the SDA allows us to give [voiced] wider scope, so that it is specified on all Russian obstruents. But this doesn’t come for free. If [voiced] is promoted in the contrastive hierarchy, something else must be demoted. We predict a trade-off.
SLIDE 84
Specifying the unpaired obstruents
If we adopt the Contrastivist Hypothesis, then [voiced] must be contrastive on /ʦ ʧ x/ in order to be active. /ʦ ʧ x/ don’t have minimally different voiced counterparts */ʣ ʤ ɣ/ in the underlying inventory… …but they contrast with voiced obstruents in general. The flexibility of the SDA allows us to give [voiced] wider scope, so that it is specified on all Russian obstruents. But this doesn’t come for free. If [voiced] is promoted in the contrastive hierarchy, something else must be demoted. We predict a trade-off.
SLIDE 85
Specifying the unpaired obstruents
If we adopt the Contrastivist Hypothesis, then [voiced] must be contrastive on /ʦ ʧ x/ in order to be active. /ʦ ʧ x/ don’t have minimally different voiced counterparts */ʣ ʤ ɣ/ in the underlying inventory… …but they contrast with voiced obstruents in general. The flexibility of the SDA allows us to give [voiced] wider scope, so that it is specified on all Russian obstruents. But this doesn’t come for free. If [voiced] is promoted in the contrastive hierarchy, something else must be demoted. We predict a trade-off.
SLIDE 86
Specifying the unpaired obstruents
If we adopt the Contrastivist Hypothesis, then [voiced] must be contrastive on /ʦ ʧ x/ in order to be active. /ʦ ʧ x/ don’t have minimally different voiced counterparts */ʣ ʤ ɣ/ in the underlying inventory… …but they contrast with voiced obstruents in general. The flexibility of the SDA allows us to give [voiced] wider scope, so that it is specified on all Russian obstruents. But this doesn’t come for free. If [voiced] is promoted in the contrastive hierarchy, something else must be demoted. We predict a trade-off.
SLIDE 87
Specifying the unpaired obstruents
If we adopt the Contrastivist Hypothesis, then [voiced] must be contrastive on /ʦ ʧ x/ in order to be active. /ʦ ʧ x/ don’t have minimally different voiced counterparts */ʣ ʤ ɣ/ in the underlying inventory… …but they contrast with voiced obstruents in general. The flexibility of the SDA allows us to give [voiced] wider scope, so that it is specified on all Russian obstruents. But this doesn’t come for free. If [voiced] is promoted in the contrastive hierarchy, something else must be demoted. We predict a trade-off.
SLIDE 88
Specifying the unpaired obstruents
If we adopt the Contrastivist Hypothesis, then [voiced] must be contrastive on /ʦ ʧ x/ in order to be active. /ʦ ʧ x/ don’t have minimally different voiced counterparts */ʣ ʤ ɣ/ in the underlying inventory… …but they contrast with voiced obstruents in general. The flexibility of the SDA allows us to give [voiced] wider scope, so that it is specified on all Russian obstruents. But this doesn’t come for free. If [voiced] is promoted in the contrastive hierarchy, something else must be demoted. We predict a trade-off.
SLIDE 89
Specifying the unpaired obstruents
Revising Halle’s hierarchy… [low tonality] Q [continuant] Q [voiced] Q [sharped]
Strident dentals: Palatals and velars: . /ʦ s sʲ z zʲ/ . . [cont] ʦ . . [cont] . . [voice] . . [sharp] s . . [sharp] sʲ . . [voice] . . [sharp] z . . [sharp] zʲ . /ʧ ʃ ʒ k kʲ x ɡ/ . . [low ton] . . [cont] ʧ . . . [cont] . . [voice] ʃ . . . [voice] ʒ . . [low ton] . . [cont] . . [voice] . . [sharp] k . . [sharp] kʲ . . [voice] ɡ . . [cont] x
…gives us [voiced] on /ʦ ʧ x/… …but removes [continuant] from /z zʲ ʒ ɡ/.
SLIDE 90
Specifying the unpaired obstruents
Revising Halle’s hierarchy… [low tonality] Q [voiced] Q [continuant] Q [sharped]
Strident dentals: Palatals and velars: . /ʦ s sʲ z zʲ/ . . [voice] . . [cont] ʦ . . [cont] . . [sharp] s . . [sharp] sʲ . . [voice] . . [sharp] z . . [sharp] zʲ . /ʧ ʃ ʒ k kʲ x ɡ/ . . [low ton] . . [voice] . . [cont] ʧ . . [cont] ʃ . . [voice] ʒ . . [low ton] . . [voice] . . [cont] . . [sharp] k . . [sharp] kʲ . . [cont] x . . [voice] ɡ
…gives us [voiced] on /ʦ ʧ x/… …but removes [continuant] from /z zʲ ʒ ɡ/.
SLIDE 91
Specifying the unpaired obstruents
Revising Halle’s hierarchy… [low tonality] Q [voiced] Q [continuant] Q [sharped]
Strident dentals: Palatals and velars: . /ʦ s sʲ z zʲ/ . . [voice] . . [cont] ʦ . . [cont] . . [sharp] s . . [sharp] sʲ . . [voice] . . [sharp] z . . [sharp] zʲ . /ʧ ʃ ʒ k kʲ x ɡ/ . . [low ton] . . [voice] . . [cont] ʧ . . [cont] ʃ . . [voice] ʒ . . [low ton] . . [voice] . . [cont] . . [sharp] k . . [sharp] kʲ . . [cont] x . . [voice] ɡ
…gives us [voiced] on /ʦ ʧ x/… …but removes [continuant] from /z zʲ ʒ ɡ/.
SLIDE 92
The other unpaired obstruents
The revised hierarchy shows the gaps in the underlying inventory—*/ʣ ʤ ɣ/—in a new light. What’s missing from the inventory are not the voiced counterparts to /ʦ ʧ x/… …but the [α continuant] counterparts to /z zʲ ʒ ɡ/. We predict that [continuant] is not phonologically active on /z zʲ ʒ ɡ/. Minimally, we predict that omitting [continuant] from these segments will not lead to what Nevins () calls an ‘Oops, I Need That’ problem. More than this, though, there seems to be positive evidence for underspecification of [continuant].
SLIDE 93
The other unpaired obstruents
The revised hierarchy shows the gaps in the underlying inventory—*/ʣ ʤ ɣ/—in a new light. What’s missing from the inventory are not the voiced counterparts to /ʦ ʧ x/… …but the [α continuant] counterparts to /z zʲ ʒ ɡ/. We predict that [continuant] is not phonologically active on /z zʲ ʒ ɡ/. Minimally, we predict that omitting [continuant] from these segments will not lead to what Nevins () calls an ‘Oops, I Need That’ problem. More than this, though, there seems to be positive evidence for underspecification of [continuant].
SLIDE 94
The other unpaired obstruents
The revised hierarchy shows the gaps in the underlying inventory—*/ʣ ʤ ɣ/—in a new light. What’s missing from the inventory are not the voiced counterparts to /ʦ ʧ x/… …but the [α continuant] counterparts to /z zʲ ʒ ɡ/. We predict that [continuant] is not phonologically active on /z zʲ ʒ ɡ/. Minimally, we predict that omitting [continuant] from these segments will not lead to what Nevins () calls an ‘Oops, I Need That’ problem. More than this, though, there seems to be positive evidence for underspecification of [continuant].
SLIDE 95
The other unpaired obstruents
The revised hierarchy shows the gaps in the underlying inventory—*/ʣ ʤ ɣ/—in a new light. What’s missing from the inventory are not the voiced counterparts to /ʦ ʧ x/… …but the [α continuant] counterparts to /z zʲ ʒ ɡ/. We predict that [continuant] is not phonologically active on /z zʲ ʒ ɡ/. Minimally, we predict that omitting [continuant] from these segments will not lead to what Nevins () calls an ‘Oops, I Need That’ problem. More than this, though, there seems to be positive evidence for underspecification of [continuant].
SLIDE 96
The other unpaired obstruents
The revised hierarchy shows the gaps in the underlying inventory—*/ʣ ʤ ɣ/—in a new light. What’s missing from the inventory are not the voiced counterparts to /ʦ ʧ x/… …but the [α continuant] counterparts to /z zʲ ʒ ɡ/. We predict that [continuant] is not phonologically active on /z zʲ ʒ ɡ/. Minimally, we predict that omitting [continuant] from these segments will not lead to what Nevins () calls an ‘Oops, I Need That’ problem. More than this, though, there seems to be positive evidence for underspecification of [continuant].
SLIDE 97
The other unpaired obstruents
The revised hierarchy shows the gaps in the underlying inventory—*/ʣ ʤ ɣ/—in a new light. What’s missing from the inventory are not the voiced counterparts to /ʦ ʧ x/… …but the [α continuant] counterparts to /z zʲ ʒ ɡ/. We predict that [continuant] is not phonologically active on /z zʲ ʒ ɡ/. Minimally, we predict that omitting [continuant] from these segments will not lead to what Nevins () calls an ‘Oops, I Need That’ problem. More than this, though, there seems to be positive evidence for underspecification of [continuant].
SLIDE 98
The other unpaired obstruents: Variation
Circumstantially, we note that Russian /ɡ/ can be realized as [ɣ] or [ɦ] as well as [ɡ]. This is dialect variation, so it doesn’t necessarily show that the same U.R. surfaces as both stop and continuant in a single grammar. However, to the extent that different dialects of Russian show similar phonological patterns, we expect their inventories to have the same specifications. If this segment variously shows up as [ɡ] and [ɣ]/[ɦ], this is consistent with—but does not entail—the idea that it is unspecified for continuancy.
SLIDE 99
The other unpaired obstruents: Variation
Circumstantially, we note that Russian /ɡ/ can be realized as [ɣ] or [ɦ] as well as [ɡ]. This is dialect variation, so it doesn’t necessarily show that the same U.R. surfaces as both stop and continuant in a single grammar. However, to the extent that different dialects of Russian show similar phonological patterns, we expect their inventories to have the same specifications. If this segment variously shows up as [ɡ] and [ɣ]/[ɦ], this is consistent with—but does not entail—the idea that it is unspecified for continuancy.
SLIDE 100
The other unpaired obstruents: Variation
Circumstantially, we note that Russian /ɡ/ can be realized as [ɣ] or [ɦ] as well as [ɡ]. This is dialect variation, so it doesn’t necessarily show that the same U.R. surfaces as both stop and continuant in a single grammar. However, to the extent that different dialects of Russian show similar phonological patterns, we expect their inventories to have the same specifications. If this segment variously shows up as [ɡ] and [ɣ]/[ɦ], this is consistent with—but does not entail—the idea that it is unspecified for continuancy.
SLIDE 101
The other unpaired obstruents: Variation
Circumstantially, we note that Russian /ɡ/ can be realized as [ɣ] or [ɦ] as well as [ɡ]. This is dialect variation, so it doesn’t necessarily show that the same U.R. surfaces as both stop and continuant in a single grammar. However, to the extent that different dialects of Russian show similar phonological patterns, we expect their inventories to have the same specifications. If this segment variously shows up as [ɡ] and [ɣ]/[ɦ], this is consistent with—but does not entail—the idea that it is unspecified for continuancy.
SLIDE 102
The other unpaired obstruents: Alternations
Some (morpho)phonological evidence: Alternations resulting from the First Velar Palatalization The hierarchy that assigns [voiced] to /ʦ ʧ x/ also correctly identifies /ɡ/ and /ʒ/ as counterparts.
examples from Lightner ()
SLIDE 103 The other unpaired obstruents: Alternations
Some (morpho)phonological evidence: Alternations resulting from the First Velar Palatalization [low tonality] [low tonality] [voiced] [continuant] x
[voiced] [continuant] k
[voiced] g ɡ
The hierarchy that assigns [voiced] to /ʦ ʧ x/ also correctly identifies /ɡ/ and /ʒ/ as counterparts.
examples from Lightner ()
SLIDE 104 The other unpaired obstruents: Alternations
Some (morpho)phonological evidence: Alternations resulting from the First Velar Palatalization [low tonality] [low tonality] [voiced] [continuant] x
[voiced] [continuant] k
[voiced] g ɡ
The hierarchy that assigns [voiced] to /ʦ ʧ x/ also correctly identifies /ɡ/ and /ʒ/ as counterparts.
examples from Lightner ()
SLIDE 105 The other unpaired obstruents: Alternations
Some (morpho)phonological evidence: Alternations resulting from the First Velar Palatalization [low tonality] [low tonality] [voiced] [continuant] x
[voiced] [continuant] k
[voiced] g ɡ
The hierarchy that assigns [voiced] to /ʦ ʧ x/ also correctly identifies /ɡ/ and /ʒ/ as counterparts.
examples from Lightner ()
SLIDE 106 The other unpaired obstruents: Alternations
Some (morpho)phonological evidence: Alternations resulting from the First Velar Palatalization [low tonality] [low tonality] [voiced] [continuant] x
[voiced] [continuant] k
[voiced] g ɡ
The hierarchy that assigns [voiced] to /ʦ ʧ x/ also correctly identifies /ɡ/ and /ʒ/ as counterparts.
examples from Lightner ()
SLIDE 107 The other unpaired obstruents: Alternations
Some (morpho)phonological evidence: Alternations resulting from the First Velar Palatalization [low tonality] [low tonality] [voiced] [continuant] x
[voiced] [continuant] k
[voiced] g ɡ
Adjectives: tʲixij tʲiʃe ‘quiet(er)’ ʒarkij ʒarʧe ‘hot(ter)’ doroɡoj doroʒe ‘dear(er)’ The hierarchy that assigns [voiced] to /ʦ ʧ x/ also correctly identifies /ɡ/ and /ʒ/ as counterparts.
examples from Lightner ()
SLIDE 108 The other unpaired obstruents: Alternations
Some (morpho)phonological evidence: Alternations resulting from the First Velar Palatalization [low tonality] [low tonality] [voiced] [continuant] x
[voiced] [continuant] k
[voiced] g ɡ
Verbs: maxut maʃet ‘wave(s), wag(s)’ pekut peʧet ‘bake(s)’ striɡut striʒet ‘shear(s)’ The hierarchy that assigns [voiced] to /ʦ ʧ x/ also correctly identifies /ɡ/ and /ʒ/ as counterparts.
examples from Lightner ()
SLIDE 109 The other unpaired obstruents: Alternations
Some (morpho)phonological evidence: Alternations resulting from the First Velar Palatalization [low tonality] [low tonality] [voiced] [continuant] x
[voiced] [continuant] k
[voiced] g ɡ
Denominal adjectives: ʧerepaxa ʧerepaʃij ‘turtle’ / ‘testudinian’ volk volʧij ‘wolf’ / ‘lupine’ vraɡ̊ vraʒij ‘enemy’ / ‘hostile’ The hierarchy that assigns [voiced] to /ʦ ʧ x/ also correctly identifies /ɡ/ and /ʒ/ as counterparts.
examples from Lightner ()
SLIDE 110 The other unpaired obstruents: Alternations
Some (morpho)phonological evidence: Alternations resulting from the First Velar Palatalization [low tonality] [low tonality] [voiced] [continuant] x
[voiced] [continuant] k
[voiced] g ɡ
Denominal adjectives: ʧerepaxa ʧerepaʃij ‘turtle’ / ‘testudinian’ volk volʧij ‘wolf’ / ‘lupine’ vraɡ̊ vraʒij ‘enemy’ / ‘hostile’ The hierarchy that assigns [voiced] to /ʦ ʧ x/ also correctly identifies /ɡ/ and /ʒ/ as counterparts.
examples from Lightner ()
SLIDE 111
The other unpaired obstruents: Alternations
Some (morpho)phonological evidence: Alternations resulting from the First Velar Palatalization
. /ʧ ʃ ʒ k kʲ x ɡ/ . . [low ton] . . [voice] . . [cont] ʧ . . [cont] ʃ . . [voice] ʒ . . . [low ton] . . . [voice] . . [cont] . . [sharp] k . . . [sharp] kʲ . . . [cont] x . . [voice] ɡ
The hierarchy that assigns [voiced] to /ʦ ʧ x/ also correctly identifies /ɡ/ and /ʒ/ as counterparts.
examples from Lightner ()
SLIDE 112 The other unpaired obstruents: Alternations
Relics of the Second Palatalization pair velars with dentals: compact low tonality compact low tonality [voiced] [continuant] k
[voiced] g ɡ, ɡʲ
brjakatʲ ‘to let fall w/ a clang’ brjaʦatʲ ‘to clang’ voskliknutʲ ‘to exclaim’ (pf.) voskliʦatʲ ‘to exclaim’ (impf.) tjaɡatʲsja ‘to sue’ sostjazatʲsja ‘to contend with’ knjaɡʲinja ‘princess’ knjaz̥ʲ ‘prince’ These alternations are not productive in Modern Russian, but they are consistent with the prediction that /z zʲ/ are also unspecified for continuancy.
examples from Lightner ()
SLIDE 113 The other unpaired obstruents: Alternations
Relics of the Second Palatalization pair velars with dentals: compact low tonality compact low tonality [voiced] [continuant] k
[voiced] g ɡ, ɡʲ
brjakatʲ ‘to let fall w/ a clang’ brjaʦatʲ ‘to clang’ voskliknutʲ ‘to exclaim’ (pf.) voskliʦatʲ ‘to exclaim’ (impf.) tjaɡatʲsja ‘to sue’ sostjazatʲsja ‘to contend with’ knjaɡʲinja ‘princess’ knjaz̥ʲ ‘prince’ These alternations are not productive in Modern Russian, but they are consistent with the prediction that /z zʲ/ are also unspecified for continuancy.
examples from Lightner ()
SLIDE 114 The other unpaired obstruents: Alternations
Relics of the Second Palatalization pair velars with dentals: compact low tonality compact low tonality [voiced] [continuant] k
[voiced] g ɡ, ɡʲ
brjakatʲ ‘to let fall w/ a clang’ brjaʦatʲ ‘to clang’ voskliknutʲ ‘to exclaim’ (pf.) voskliʦatʲ ‘to exclaim’ (impf.) tjaɡatʲsja ‘to sue’ sostjazatʲsja ‘to contend with’ knjaɡʲinja ‘princess’ knjaz̥ʲ ‘prince’ These alternations are not productive in Modern Russian, but they are consistent with the prediction that /z zʲ/ are also unspecified for continuancy.
examples from Lightner ()
SLIDE 115 The other unpaired obstruents: Alternations
Relics of the Second Palatalization pair velars with dentals: compact low tonality compact low tonality [voiced] [continuant] k
[voiced] g ɡ, ɡʲ
brjakatʲ ‘to let fall w/ a clang’ brjaʦatʲ ‘to clang’ voskliknutʲ ‘to exclaim’ (pf.) voskliʦatʲ ‘to exclaim’ (impf.) tjaɡatʲsja ‘to sue’ sostjazatʲsja ‘to contend with’ knjaɡʲinja ‘princess’ knjaz̥ʲ ‘prince’ These alternations are not productive in Modern Russian, but they are consistent with the prediction that /z zʲ/ are also unspecified for continuancy.
examples from Lightner ()
SLIDE 116 The other unpaired obstruents: Alternations
Relics of the Second Palatalization pair velars with dentals: compact low tonality compact low tonality [voiced] [continuant] k
[voiced] g ɡ, ɡʲ
brjakatʲ ‘to let fall w/ a clang’ brjaʦatʲ ‘to clang’ voskliknutʲ ‘to exclaim’ (pf.) voskliʦatʲ ‘to exclaim’ (impf.) tjaɡatʲsja ‘to sue’ sostjazatʲsja ‘to contend with’ knjaɡʲinja ‘princess’ knjaz̥ʲ ‘prince’ These alternations are not productive in Modern Russian, but they are consistent with the prediction that /z zʲ/ are also unspecified for continuancy.
examples from Lightner ()
SLIDE 117 The other unpaired obstruents: Alternations
Relics of the Second Palatalization pair velars with dentals: compact low tonality compact low tonality [voiced] [continuant] k
[voiced] g ɡ, ɡʲ
brjakatʲ ‘to let fall w/ a clang’ brjaʦatʲ ‘to clang’ voskliknutʲ ‘to exclaim’ (pf.) voskliʦatʲ ‘to exclaim’ (impf.) tjaɡatʲsja ‘to sue’ sostjazatʲsja ‘to contend with’ knjaɡʲinja ‘princess’ knjaz̥ʲ ‘prince’ These alternations are not productive in Modern Russian, but they are consistent with the prediction that /z zʲ/ are also unspecified for continuancy.
examples from Lightner ()
SLIDE 118 The other unpaired obstruents: Alternations
Relics of the Second Palatalization pair velars with dentals: compact low tonality compact low tonality [voiced] [continuant] k
[voiced] g ɡ, ɡʲ
brjakatʲ ‘to let fall w/ a clang’ brjaʦatʲ ‘to clang’ voskliknutʲ ‘to exclaim’ (pf.) voskliʦatʲ ‘to exclaim’ (impf.) tjaɡatʲsja ‘to sue’ sostjazatʲsja ‘to contend with’ knjaɡʲinja ‘princess’ knjaz̥ʲ ‘prince’ These alternations are not productive in Modern Russian, but they are consistent with the prediction that /z zʲ/ are also unspecified for continuancy.
examples from Lightner ()
SLIDE 119 The other unpaired obstruents: Alternations
Relics of the Second Palatalization pair velars with dentals: compact low tonality compact low tonality [voiced] [continuant] k
[voiced] g ɡ, ɡʲ
brjakatʲ ‘to let fall w/ a clang’ brjaʦatʲ ‘to clang’ voskliknutʲ ‘to exclaim’ (pf.) voskliʦatʲ ‘to exclaim’ (impf.) tjaɡatʲsja ‘to sue’ sostjazatʲsja ‘to contend with’ knjaɡʲinja ‘princess’ knjaz̥ʲ ‘prince’ These alternations are not productive in Modern Russian, but they are consistent with the prediction that /z zʲ/ are also unspecified for continuancy.
examples from Lightner ()
SLIDE 120 The other unpaired obstruents: Alternations
Relics of the Second Palatalization pair velars with dentals: compact low tonality compact low tonality [voiced] [continuant] k
[voiced] g ɡ, ɡʲ
brjakatʲ ‘to let fall w/ a clang’ brjaʦatʲ ‘to clang’ voskliknutʲ ‘to exclaim’ (pf.) voskliʦatʲ ‘to exclaim’ (impf.) tjaɡatʲsja ‘to sue’ sostjazatʲsja ‘to contend with’ knjaɡʲinja ‘princess’ knjaz̥ʲ ‘prince’ These alternations are not productive in Modern Russian, but they are consistent with the prediction that /z zʲ/ are also unspecified for continuancy.
examples from Lightner ()
SLIDE 121
Elsewhere in Slavic
Other Slavic languages show similarly asymmetrical inventories, and similar phonological patterns:
SLIDE 122
Elsewhere in Slavic
Other Slavic languages show similarly asymmetrical inventories, and similar phonological patterns: Serbian: /ɡ/ has no continuant counterpart, and alternates with /ʒ/ and with /z/. Radišić () argues for a contrastive hierarchy that leaves /ɡ/ unspecified for continuancy.
source: Radišić () on Serbian
SLIDE 123
Elsewhere in Slavic
Other Slavic languages show similarly asymmetrical inventories, and similar phonological patterns: Lower Sorbian: /ɡ/ has no continuant counterpart. Where /k/ alternates with /ʦ/ and /x/ with /ʃ/… ruk-a ruʦ-e ‘hand’ mux-a muʃ-e ‘fly’ noɡ-a noz-e ‘leg’ rozɡ-a rozʣ-e ‘twig’ …/ɡ/ becomes either /z/ or /ʣ/, whichever is phonotactically less marked (/ʣ/ aer /z/; /z/ elsewhere).
sources: Radišić () on Serbian; Schaarschmidt () on Sorbian
SLIDE 124
Elsewhere in Slavic
Other Slavic languages show similarly asymmetrical inventories, and similar phonological patterns: Lower Sorbian: /ɡ/ has no continuant counterpart. Where /k/ alternates with /ʦ/ and /x/ with /ʃ/… ruk-a ruʦ-e ‘hand’ mux-a muʃ-e ‘fly’ noɡ-a noz-e ‘leg’ rozɡ-a rozʣ-e ‘twig’ …/ɡ/ becomes either /z/ or /ʣ/, whichever is phonotactically less marked (/ʣ/ aer /z/; /z/ elsewhere).
sources: Radišić () on Serbian; Schaarschmidt () on Sorbian
SLIDE 125
Elsewhere in Slavic
Other Slavic languages show similarly asymmetrical inventories, and similar phonological patterns: Lower Sorbian: /ɡ/ has no continuant counterpart. Where /k/ alternates with /ʦ/ and /x/ with /ʃ/… ruk-a ruʦ-e ‘hand’ mux-a muʃ-e ‘fly’ noɡ-a noz-e ‘leg’ rozɡ-a rozʣ-e ‘twig’ …/ɡ/ becomes either /z/ or /ʣ/, whichever is phonotactically less marked (/ʣ/ aer /z/; /z/ elsewhere).
sources: Radišić () on Serbian; Schaarschmidt () on Sorbian
SLIDE 126 Elsewhere in Slavic
Other Slavic languages show similarly asymmetrical inventories, and similar phonological patterns: Ukrainian: Historical */ɡ/ has become /ɦ/, making its alternations with coronal continuants more transparent phonetically. A new, marginally contrastive stop /ɡ/ is emerging through borrowings.
. . [voice] . . [cont] k . . [cont] x . . [voice] ɡ
. [voice] . . [cont] k . . [cont] x . . [voice] ɦ
. [voice] . . [cont] k . . [cont] x . . [voice] . . [cont] ɡ . . [cont] ɦ
sources: Radišić () on Serbian; Schaarschmidt () on Sorbian; Shevelov () on Ukrainian
SLIDE 127 Elsewhere in Slavic
Other Slavic languages show similarly asymmetrical inventories, and similar phonological patterns: Ukrainian: Historical */ɡ/ has become /ɦ/, making its alternations with coronal continuants more transparent
- phonetically. A new, marginally contrastive stop /ɡ/ is
emerging through borrowings.
. . [voice] . . [cont] k . . [cont] x . . [voice] ɡ
. [voice] . . [cont] k . . [cont] x . . [voice] ɦ
. . . . [voice] . . [cont] k . . . [cont] x . . . [voice] . . . [cont] ɡ . . . [cont] ɦ
sources: Radišić () on Serbian; Schaarschmidt () on Sorbian; Shevelov () on Ukrainian
SLIDE 128
Conclusions
The Successive Division Algorithm is not deterministic. It does not stipulate the order of features, and so it cannot predict exactly which features will be active based on the inventory alone. This makes it compatible with the proposition that features themselves are emergent (Mielke ), as discussed by Dresher () and Cowper & Hall (). But it does make predictions about how many features can be specified, and about trade-offs between potential specifications. These predictions are, in principle, falsifiable. As regards voicing and continuancy in Slavic, though, it appears that they are not actually false.
SLIDE 129
Conclusions
The Successive Division Algorithm is not deterministic. It does not stipulate the order of features, and so it cannot predict exactly which features will be active based on the inventory alone. This makes it compatible with the proposition that features themselves are emergent (Mielke ), as discussed by Dresher () and Cowper & Hall (). But it does make predictions about how many features can be specified, and about trade-offs between potential specifications. These predictions are, in principle, falsifiable. As regards voicing and continuancy in Slavic, though, it appears that they are not actually false.
SLIDE 130
Conclusions
The Successive Division Algorithm is not deterministic. It does not stipulate the order of features, and so it cannot predict exactly which features will be active based on the inventory alone. This makes it compatible with the proposition that features themselves are emergent (Mielke ), as discussed by Dresher () and Cowper & Hall (). But it does make predictions about how many features can be specified, and about trade-offs between potential specifications. These predictions are, in principle, falsifiable. As regards voicing and continuancy in Slavic, though, it appears that they are not actually false.
SLIDE 131
Conclusions
The Successive Division Algorithm is not deterministic. It does not stipulate the order of features, and so it cannot predict exactly which features will be active based on the inventory alone. This makes it compatible with the proposition that features themselves are emergent (Mielke ), as discussed by Dresher () and Cowper & Hall (). But it does make predictions about how many features can be specified, and about trade-offs between potential specifications. These predictions are, in principle, falsifiable. As regards voicing and continuancy in Slavic, though, it appears that they are not actually false.
SLIDE 132
Conclusions
The Successive Division Algorithm is not deterministic. It does not stipulate the order of features, and so it cannot predict exactly which features will be active based on the inventory alone. This makes it compatible with the proposition that features themselves are emergent (Mielke ), as discussed by Dresher () and Cowper & Hall (). But it does make predictions about how many features can be specified, and about trade-offs between potential specifications. These predictions are, in principle, falsifiable. As regards voicing and continuancy in Slavic, though, it appears that they are not actually false.
SLIDE 133
Conclusions
The Successive Division Algorithm is not deterministic. It does not stipulate the order of features, and so it cannot predict exactly which features will be active based on the inventory alone. This makes it compatible with the proposition that features themselves are emergent (Mielke ), as discussed by Dresher () and Cowper & Hall (). But it does make predictions about how many features can be specified, and about trade-offs between potential specifications. These predictions are, in principle, falsifiable. As regards voicing and continuancy in Slavic, though, it appears that they are not actually false.
SLIDE 134 References I
Blaho, Sylvia (). The syntax of phonology: A radically substance-free
- approach. PhD dissertation, Universitetet i Tromsø.
Calabrese, Andrea (). A constraint-based theory of phonological markedness and simplification procedures. Linguistic Inquiry :. –. Cowper, Elizabeth & Daniel Currie Hall (). Reductiō ad discrīmen: Where features come from. Nordlyd :. –. Dresher, B. Elan (). The contrastive hierarchy in phonology. Cambridge: Cambridge University Press. Dresher, B. Elan (). The phoneme. In Marc van Oostendorp, Colin J. Ewen, Elizabeth Hume & Keren Rice (eds.) The Blackwell companion to phonology, volume . Oxford: Wiley-Blackwell, –. Dresher, B. Elan (). The arch not the stones: Universal feature theory without universal features. Nordlyd :. –. Dresher, B. Elan (). The motivation for contrastive feature hierarchies in phonology. Linguistic Variation :. –.
SLIDE 135 References II
Hall, Daniel Currie (). The role and representation of contrast in phonological theory. PhD dissertation, University of Toronto. Hall, Daniel Currie (forthcoming). Contrastive specification in phonology. In Mark Aronoff (ed.) Oxford research encyclopedia of linguistics. Oxford: OUP. Halle, Morris (). On the phonetic rules of Russian. Presented to the Linguistic Society of America, Chicago, . Halle, Morris (). The sound pattern of Russian: A linguistic and acoustical investigation. The Hague: Mouton. de Lacy, Paul (). Review of Dresher (). Phonology :. –. Lightner, Theodore (). Segmental phonology of Modern Standard
- Russian. PhD dissertation, Massachusetts Institute of Technology.
Mielke, Jeff (). The emergence of distinctive features. Oxford: Oxford University Press. Nevins, Andrew Ira (). Triumphs and limits of the contrastivity-only
- hypothesis. Linguistic Variation :. –.
SLIDE 136 References III
Padgett, Jaye (). Russian voicing assimilation, final devoicing, and the problem of [v] (or, the mouse that squeaked). Ms., University of California, Santa Cruz. ROA #. Radišić, Milica (). The double nature of the velar /g/ in Serbian. Toronto Working Papers in Linguistics . –. Schaarschmidt, Gunter (). The historical phonology of the Upper and Lower Sorbian languages. Heidelberg: C. Winter. Shevelov, George Y. (). On the chronology of h and the new g in
- Ukrainian. Harvard Ukrainian Studies :. –.
Timberlake, Alan (). Russian. In Bernard Comrie & Greville G. Corbett (eds.) The Slavonic languages, first paperback edition. London: Routledge, –.
