Ling 403/603 Introduction to Phonology - University of ...udel.edu/~koirala/phonology/day23.pdf ·...

22
DAY 23 CESAR KOIRALA Ling 403/603 Introduction to Phonology

Transcript of Ling 403/603 Introduction to Phonology - University of ...udel.edu/~koirala/phonology/day23.pdf ·...

D A Y 2 3

C E S A R K O I R A L A

Ling 403/603

Introduction to Phonology

Nonlocal Environments in Phonology

Nonlocal Environments are the cases in which trigger of a phonological process can occur at an extended distance from the target (For example: long distance harmony)

Autosegmentalism (Goldsmith 1976):

Nonlocal processes are actually local.

All nonlocal operations are local on a tier; the set of tiers constrains the set of nonlocal environments.

Autosegmental Phonology

Introduced by Goldsmith (1976) in his dissertation.

Initially formalized to address tonal phenomenon.

Tones:

Pitch contrast, which is distinctive for words/word forms

Segments Tone

ma high ‘mother’

ma mid-high ‘hemp’

ma mid-low-high ‘horse’

ma high-low ‘scold’

The absolute slicing hypothesis:

In traditional phonology, a phonological representation is a complete transitive order of segments.

This means that for the all sounds in the representation:

1. Either sound A precedes sound B or sound B precedes sound A

2. If sound A precedes sound B and sound B precedes sound C, then sound A precedes sound C.

The consequence: Speech can be divided phonologically into isolated segments.

Autosegmental Phonology

Goldsmith observed that there was not always a one-to-one mapping of tones to syllables:

1. Two tones could occupy one syllable

2. Two syllables could be occupied by one tone.

Tones:

Pitch contrast, which is distinctive for words/word forms

Segments Tone

ma high ‘mother’

ma mid-high ‘hemp’

ma mid-low-high ‘horse’

ma high-low ‘scold’

Autosegmental Phonology

Non-linearity: Goldsmith proposed that phonological representations should be composed of multiple tiers of segments. The tiers correspond to the different gestures of speech and differ according to the features that are specified for the segments on them.

Here, we have a tonal tier where the segments are specified only for tones

and a second tier where segments are specified for all other features

Autosegmental Phonology

The lines that link segments on the tonal tier to the segments on the segmental tier are called association lines.

An association line between two segments on two tiers means they are articulated simultaneously.

Crucially, there need not be one to one mapping between segments of two tiers.

Here, we have a tonal tier where the segments are specified only for tones

and a second tier where segments are specified for all other features

Possible mapping relations

• One-to-one mapping corresponds to the traditional view of tonal association.

(Each tone is mapped to a single vowel)

• Many-to –one mapping tells that a single syllable is produced during the period

of two distinct tones.

• Multiple linking shows that 3 syllable are produced during the period of a single

low tone.

Rules in Autosegmental Phonology

• Though Autosegmental phonology is non-linear in its approach, it is still derivational in

nature.

(surface forms are derived from underlying forms via the application of rules)

• Linking: Lets look at the the case of nasal spreading in English: nasality spreads

from nasal consonant to a preceding vowel. For example, the vowel in can is nasalized

due to the following nasal consonant.

• An unbroken association line in the rule forms part of the structural description and is

to be read as already being present.

• The dashed line is the change and denotes spreading.

v

Rules in Autosegmental Phonology

• The case of nasal spreading in English: nasality spreads from nasal consonant to a

preceding vowel. For example, the vowel in can is nasalized due to the following nasal

consonant.

Advantage of this approach:• IT shows that there are not two [+nas] features which associate themselves with

two adjacent segments, but rather the same single feature applies to both

segments.

Rules in Autosegmental Phonology

• Delinking: An autosegmental rule can also delete an association line. Lets look at an

example of High tone shift in the Sukuma language of East Africa.

• The rule shows that the high tone shifts one syllable to the right.

• Broken association line represents deletion.

• The circle around the segment means that it is unassociated with a segment on

the adjacent tier.

Stability:Stability is a property of autosegments which means they remain present in the

representation even if those segments with which they are associated on another

tier are deleted via a rule.

Conventions

No-crossing Constraint

• Lines associating segments on two tiers may not cross each other.

Contrast [can] vs. [film]

• This constraint is not language specific, but a constraint of the autosegmental theory in

general.

• The idea is to constrain the power of the theory. It serves to block the rule from

applying.

The treatment of tones

Evidence: Luganda Vowel Coalescence

Note:

1. The tone associated with [a] did not delete even when [a] deleted.

2. The tone associated with [a] shifts to the next syllable

“Tones operate on a separate level from vowels and consonants”

Example: Mende

Step 1: Tonal association – Associate tone to TBU

Step 2: Left to Right Tone spreading – Spread the tone to adjacent unassociated

TBUs

Example: Mende

Universal Linking Conventions (WFC)

Vowel Harmony

Vowel harmony is a requirement found in many different languages by which

vowels in a certain domain agree in one or more phonetic features.

Turkish Vowel Harmony

1. Allomorphs of plural morpheme?

2. Rule that explains the alternation?

Vowel Harmony

Vowel harmony is a requirement found in many different languages by which

vowels in a certain domain agree in one or more phonetic features.

Turkish Vowel Harmony

1. Allomorphs of plural morpheme?

2. Rule that explains the alternation?

Back harmony:

V →[α back] / [α back] C0___

Advantages of doing Vowel Harmony autosegmentally