Você está na página 1de 8

Week 2 Phonemic Analysis with Features

September 15, 2011

Hayes 2009 and Features

In this course, we will rely on Hayes (2009) for the feature system used in this course. This feature system is based on the one from Chomsky and Halle (1968). The total set of features is identical with the set of phonetic properties that can in principle be controlled in speech; they represent the phonetic capabilities of man, and we would assume, are therefore the same for all languages.(Chomsky & Halle 1968:29495)

1.1

Comprehensive

This system is comprehensive in two ways: 1. Every IPA symbol (except for clicks) is translated into phonological features. 2. Every IPA symbol draws from the same set of features. This last one is helped along by the use of the 0 feature value.

1.2

Zero valued features

Some features only make sense in the presence of others. For example, consider vowel features like [high] and [back], which can be dened in terms of the relative position of the tongue body in the oral cavity. What values for these features would sounds [p] and [s] have? The zero value is used in these situations. It means this feature does not care how it realized. In the context of motor planning, it means the phonetic dimension associated with the feature is realized in whatever way is most convenient. The zero feature manifests itself most prominently with respect to place features. The following diagram illustrates these features.

Week 2: Phonemic Analysis with Features PLACE

J. Heinz

Labial labiodental round

Coronal anterior distributed lateral strident

Dorsal back front high low

Figure 1: Place features

Phonemic analysis with distinctive features

Now we return to phonemic analysis we did earlier, but this time making use of distinctive features. To summarize the hypothesis we have been pursuing, we state the organization of the phonological system as follows: 1. The phonological component of a lexical item is a sequence of phonemes. 2. The phonemes themselves are specied in terms of distinctive features. Perhaps, as Jakobson suggests, they are somewhat abstract. 3. However, the phones themselves carry all phonological features. 4. The mapping between the phonemes and the phones may: (a) change the distinctive-feature make-up of a sound segment (b) add non-distinctive features of which there are two kinds: i. the kind we have discussed in terms of complementary distribution ii. redundant features (this kind is often referred to as redundancy rules) 5. The output of this mapping is transformed further into language-specic motor control plans. Recall our observations from Maasai:

/p,t,k/

abstract/phonemic level

[p,t,k]

[b,d,g]

[B,D,G] 2

concrete/allophonic level

Week 2: Phonemic Analysis with Features We had two (preliminary) generalizations: 1. Voiceless stops become voiced post nasally. 2. Voiceless stops become voiced and spirantize intersyllabically.

J. Heinz

2.1

Rewrite Rules

Here is a feature chart for Maasai phones. Table 1: Phonetic Features for Maasai consonants b B m t d D n l r s S 0 0 + + 0 0 + + + 0 0 + + + 0 0 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +

p implosive nasal continuant trill coronal lateral anterior labial dorsal voice 0 0 +

k 0 0 +

g 0 0 +

G + 0 0 + +

+ 0 0 + +

Here are distinctive features for Maasai phones that I obtained using the SDA (Dresher, 2009) with the hierarchy shown in the appendix (if we have time we can discuss this). Table 2: Distinctive Features for Maasai phonemes p m t n l r s S k implosive nasal continuant trill lateral anterior labial dorsal + + + + + + + + +

+ + + +

First attempt at our rules: 1. Postnasal Voicing (PNV): -continuant -implosive 3 [+voice] / [+nasal]

Week 2: Phonemic Analysis with Features -continuant -implosive +continuant +voice

J. Heinz

2. Spirantization:

/ [+syllabic]

[+syllabic]

Lets apply these to some lexical items in Maasai. Underlying form Postnasal Voicing Spirantization Surface form /impala/ papers imbala [imbala] patan skilled in shooting paDan [paDan]

Because everything is feature driven, the IPA symbols above are just convenient shorthand, and actually hide a lot of the analysis. Lets derive [imbala] papers in excruciating detail.
-syllabic -syllabic -implosive +syllabic -implosive -nasal input +high +nasal -continuant -back +labial +labial -syllabic -implosive -nasal PNV -continuant +labial +voice -syllabic -implosive -syllabic +syllabic -implosive -nasal output +high -continuant +nasal -back +labial +labial +voice +syllabic +low -syllabic -nasal +continuant +lateral +syllabic +low -syllabic -nasal +continuant +lateral

+syllabic +low

+syllabic +low

Can we remove the [-implosive] feature from the target of Postnasal Voicing? Why or why not?

Can we remove the [+voice] feature from the structural change of Spirantization? Why or why not?

Week 2: Phonemic Analysis with Features So we can nalize the rules as follows: 1. Postnasal Voicing (PNV): [-continuant] [+voice] / [+nasal] 2. Spirantization: [-continuant] +continuant +voice / [+syllabic]

J. Heinz

[+syllabic]

This in turn suggests we can sharpen our generalizations to 1. Stops become voiced post nasally. 2. Stops become voiced and spirantize intersyllabically. This output is not yet quite the surface form. There are a number of other, context-free, feature ll-in rules, which must apply before we acquire the phonological surface form. These are often omitted from analyses because they are considered to be obvious and implicit. 1. [+syllabic] [+voice] 2. [+nasal] [+voice] 3. [+implosive] [+voice] 4. [+labial] [-dorsal] 5. [+labial] [-coronal] 6. [-coronal] [0anterior] 7. . . . How would the default rule that makes noncontinuants voiceless be written?

2.2

Introduction to Optimality Theory

Lets not forget about constraints! If we are interested in describing the licit phonotactic patterns in Maasai, we would make the following two generalizations (now using our phonetic features from Table 2.1): 1. *[+nasal][-voice] 2. *[+syllabic][-continuant][+syllabic] Optimality Theory (Prince and Smolensky, 1993, 2004) derives surface forms from underlying forms in a dierent manner than rewrite rules. 5

Week 2: Phonemic Analysis with Features

J. Heinz

1. Constraints like the ones above are a central component of the theory. They are called markedness constraints because they describe ill-formed sequences. 2. OT divorces the ill-formed structure from how it could be repaired. 3. In addition to markedness constraints, there are faithfulness constraints, which penalizes dierences between the underyling and surface form. 4. The derivation proceeds by nding the surface form which best satises the markedness and faithfulness constraints. 5. Best satisfaction can be determined because the constraints are prioritized: it is more important to avoid violating certain constraints than others. 6. Distinctive features are thrown out the window. /impala/ a. impala b. imbala *[+nasal][-voice] Ident(Voice) * *

Table 3: This is an OT Tableau. The asterisks show that the surface form [impala] violates the markedness constraint, but not the faithfulness constraint, and that the surface form [imbala] violates the faithfulness constraint, but not the markedness constraint. The Ident(voice) constraint incurs a violation for evert value of a voicing feature in the underlying form which does not match their values in the surface form. This means that in underlying forms, sound segments have all their features! /impala/ a. impala *[+nasal][-voice] Ident(Voice) * *

b.  imbala

Table 4: Here we have ranked the markedness constraint above the faithfulness constraint (indicated by the left-to-right ordering and the solid vertical line). This tableau shows then that [impala] is more harmonic than [imbala] with respect to this ranking. The pointy nger indicates that this is the most harmonic candidate surface form in the tableau. Constraint ranking determines the sound pattern of languages. What other ways could the underlying form change to avoid violating this markedness constraint?

Week 2: Phonemic Analysis with Features /impala/ a.  impala b. imbala * Ident(Voice) *[+nasal][-voice] *

J. Heinz

Table 5: Here we have reversed the ranking of the constraints. It is now more important NOT to change the value of a voicing features than it is to avoid the markedness constraint. Hence here [imbala] is more harmonic than [impala].

References
Chomsky, Noam, and Morris Halle. 1968. The Sound Pattern of English. New York: Harper & Row. Dresher, Elan. 2009. The Contrastive Hierarchy in Phonology. Cambridge University Press. Prince, Alan, and Paul Smolensky. 1993. Optimality Theory: Constraint interaction in generative grammar. Tech. Rep. 2, Rutgers University Center for Cognitive Science. Prince, Alan, and Paul Smolensky. 2004. Optimality Theory: Constraint Interaction in Generative Grammar . Blackwell Publishing.

Você também pode gostar