Features!

Will Styler - LIGN 111

Today’s Plan

• Phonological Features

Lists of segments aren’t great

• They’re inefficient

• They obscure relations among segments

• We can do better

Natural Classes are better!

• They represent common groupings of sounds

• They’re more efficient than lists

• They get at the idea that some rules apply to groups because of some shared characteristic

… but why are natural classes natural?

• What unites /w ɹ j/ as approximants?

• Why are /m n ɲ ŋ/ grouped?

• What’s the meaningful link between [pʰ tʰ kʰ]?

  • Why do we treat them as different from [p t k] in English?

  • … or from [b d g]?

These differences in groups matter to speakers

• Languages place these groups in opposition

• In English, [pʰ tʰ kʰ] are different from [p t k]

  • … but [p t k] and [b d g] aren’t in opposition

• In Spanish, [p t k] are different from [b d g]

  • … but [pʰ tʰ kʰ] and [p t k] aren’t in opposition

But is this actually about groups of sounds?

• In Spanish, [p t k] AND [pʰ tʰ kʰ] are different from [b d g]

• This sure seems to be about voicing

• In English, [pʰ tʰ kʰ] are different from [p t k] AND [b d g]

• Could we say that this contrast is about aspiration (greater glottal airflow)?

English vowels are nasalized before /m n ŋ/

• So, it’s before nasal sounds?

• … but also before [ɾ̃] as in “winter”

• This is about nasality!

It’s often easier to talk not about the classes of sounds, but the more abstract features that make the classes distinct

• Which is why we’re going to use…

Phonological Features

Phonological Features

Abstract representations of the properties of sounds which make them phonologically distinct in languages

Features have three main principles

• They capture linguistic contrasts

• They are what natural classes are made of

• They’re phonetically grounded

Features capture contrasts

• In Spanish, stops contrast on the basis of [+/- voice]

• In English, vowel nasality is triggered only by [+nasal] sounds

• Feature theories are designed to capture the contrasts present in languages

Features are the underpinnings of natural classes

• Sounds in a natural class share features

• This is the nature of a natural class

Features are phonetically meaningful

• Phonetic realization remains a crucial part of sounds’ phonological nature

Features in Practice

Features can be unary or binary

• Unary features have one state (e.g. [+coronal])

• Binary features have two states (e.g. [+voice] or [-voice])

Features overlap across segments and segment types

• Each feature describes many segments

• … but every segment can be specified using one specific combination

Sounds can be specified as a series of features

We can ‘toggle’ or change individual features as a part of rules

• [-sonorant] -> [-voice] / __#

• [-cont] -> [+nasal] / V__V

Features can also trigger rules

• [+syllabic] -> [+nasal] / __[+nasal]

• [-cont, +coronal] -> ∅ / __[+lateral]

One important note!

Feature Sets are theory!

• You can split sounds up in many different ways

• Each approach will have some benefits, and some downsides

A certain feature theory will describe the entire set of sounds in play

• Every sound is specified in terms of distinctive features

• … but some sets will use different features to do it

There are many proposed feature theories

• We’re going to use a relatively vanilla set of features

• Designed to work effectively in a variety of problems across languages

• … but there are other approaches

• Please use the feature specifications we’ve given you for this class!

Will was trained in a different theory!

• … but he wants you to learn the local standard

• If you see him pausing or checking the chart, that’s what’s up!

Now, let’s talk about the feature set we’re using!

Wrapping up

• Features allow us to access some of the contrasts made in language directly.

• We can use features to write our rules more elegantly

• From here out, you’ll want to use features to write your rules

• Language seems to use features, so we will too!

Next time

• We’ll get some practice applying features to sounds and rules

Thank you!