LIGN 111 - Will Styler

---

# Phonemic Analysis

### Will Styler - LIGN 111

---

### Today's Plan

- Phonemic Analysis from three different perspectives

- How to tell if your instructor is actually Batman

- How to actually DO phonemic analysis

---

## You've just jumped out of a plane

---

### Now, you collect data, write a grammar, and write a dictionary

- What are the morphemes in this language?

- What are the words in this language?

- How should this language be written?

- Decisions will need to be made

---

### One of the key questions you'll face is 'which sounds matter to speakers'

- Which sounds carry a *contrast*

- Which sounds *define words*

- Which sounds *cannot be changed without changing meanings*

---

### Phonemic Analysis

- Determining which sound changes *affect the meaning* of a word in a language

- Phonemic Sounds

- ... which sound changes are *predictable*, and don't change word meanings

- Allophonic Sounds

- ... and which sound variations are completely unpredictable and meaningless

- Free variation
    
---

### A note on notation

- Phonemes go in // slashes

- /t/ means 'the /t/ phoneme'
    
    - /t/ exists in the speaker's mind*
    
- Allophones go in [] brackets

- [t] means 'a [t] on the surface', from the speaker's mouth

- This [t] may be an allophone of /t/ or some other phoneme

---

### We're going to look at this process using three different perspectives

- 1) Based on looking at data

- 2) Based on looking at perception

- 3) Based on looking at distributions

---

### This is a "threshold concept"

- That's why I'm covering it from three directions

- If you're struggling here, come to office hours

---

### Phonemic Analysis I: Let's use some data!

---

### "Oh no. There are sounds everywhere!"

- "People use a huge set of sounds, and I don't know which differences matter!"

- "What changes in sounds are *random*, and don't affect the meanings?"

- "What changes in sounds are *meaningful*, and change the meaning of a word?"

- "What changes in sounds can be *predicted* based on the other elements of the word, and don't change the meaning?"

- "What should I write down in my grammar or dictionary?"

---

### Does the difference between [t] and [t̪] change the meaning?

---

### No!  It doesn't change the meaning!

- In English, it doesn't matter whether a /t/ is made as a dental sound [t̪] or an alveolar sound [t]

- Speakers can do two things and nobody particularly cares

- There's no pattern, no standard, just maddening chaos

- This is **free variation**.

---

### Does the difference between [k] and [p] change the meaning?

---

### Yes! The meaning is changed

- When you change from /k/ to /p/, the meaning of the word changes

- We see 'minimal pairs' (e.g. [ki] and [pi]) where that segment is the only thing that's changed.

- /k/ and /p/ are in a **contrastive** distribution

- They are **two different phonemes**

---

### Does the difference between [ej] and [ej:] change the meaning?

<center>
<style type="text/css">
.tg  {border-collapse:collapse;border-spacing:0;}
.tg td{font-family:Arial, sans-serif;padding:10px 5px;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;border-color:black;}
.tg th{font-family:Arial, sans-serif;font-weight:normal;padding:10px 5px;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;border-color:black;}
.tg .tg-1wig{font-weight:bold;text-align:left;vertical-align:top}
.tg .tg-0lax{text-align:left;vertical-align:top}
.tg .tg-xldj{border-color:inherit;text-align:left}
.tg .tg-0pky{border-color:inherit;text-align:left;vertical-align:top}
</style>
<table class="tg">
  <tr>
    <td class="tg-xldj">[sejf]</td>
    <td class="tg-xldj">[sej:v]</td>
  </tr>
  <tr>
    <td class="tg-xldj">[ejp]</td>
    <td class="tg-xldj">[ej:b]</td>
  </tr>
  <tr>
    <td class="tg-xldj">[mejt]</td>
    <td class="tg-xldj">[mej:d]</td>
  </tr>
  <tr>
    <td class="tg-xldj">[ejtʃ]</td>
    <td class="tg-xldj">[ej:dʒ]</td>
  </tr>
  <tr>
    <td class="tg-xldj">[ejk]</td>
    <td class="tg-xldj">[ej:g]</td>
  </tr>
  <tr>
    <td class="tg-0pky">[mejs]</td>
    <td class="tg-0pky">[mej:z]</td>
  </tr>
</table>
</center>

---

### No!

- Changing from a short to long [ej] doesn't change the meaning for speakers

- Vowel length is *predictable* based on the voicing of the next consonant

- Short and long vowels are in a **complementary** distribution

- They are **allophones of the same phoneme**

---

### How are you feeling about this material?

---

### Phonemic Analysis II: Let's think about perception

---

### Spanish speakers hear...

- # "Cabo"

- When somebody says either

- # [kabo] or [kaβo]
    
---
    
### Spanish speakers hear...

- # /b/

- When somebody says either

- # [b] or [β]

---

### English has two /l/ sounds

- Light l ([l] as in 'lip')

- 'Dark' or Velarized l ([ɫ] as in 'pill')

- The 'Dark' L happens at the end of a syllable

---

### English speakers hear...

- # "Pill"

- When somebody says either...

- # [pɪl] or [pɪɫ]
    
---

### English speakers hear...

- # /l/

- When somebody says either...

- # [l] or [ɫ]
    
---

### Speakers of language hear...

- # The phoneme

- When somebody says...

- # Any of the allophones of that phoneme

---

### Phonemes are groups of sounds which trade places predictably!

- ... and that trading is opaque to speakers

- The /l/ phoneme has two allophones in English: [l] and [ɫ]

- As in 'lip' and 'pill'

- The /t/ phoneme has many allophones in English: [t], [tʰ], [ʔ], [t˺], [ɾ], [ɾ̃]

- As in 'stop', 'top', 'button', 'cat', 'later', 'winter'

- Every sound produced is an allophone of some phoneme

- ... but not every phoneme has multiple allophones

---

### How are you feeling about this material?

---

### Phonemic Analysis III: Looking at distributions

---

### Allophones are *predictable*

- They always show up in specific environments, and don't affect the meaning of the word

- [ɱ] shows up in English *only* when there's an /m/ before an /f/

- 'Same' [sejm] but 'Symphony' [sɪɱfəni]

- [n̪] shows up in English *only* when there's an /n/ before an Interdental sound

- 'Pine' [pajn] but 'Pine thug' [pajn̪ θʌɡ]

- [ɫ] shows up in MUSE *only* at the end of a syllable

- 'lip' [lɪp] but 'pill' [pɪɫ]

---

### Allophones show up according to rules
    
- If a sound is *variant* that only shows up according to a *rule*, it's an allophone of another phoneme

---

### Phonemes are *unpredictable*

- There is no pattern which dictates where they show up

- They can occur in the same environments

- They form 'minimal pairs', different words which differ only in that sound

- (or sometimes near minimal pairs e.g., te[ð]er vs. mea[ʒ]ure)

---

### If a sound has its own identity in the language’s structure, it’s a phoneme.

- If it’s just another 'persona', it’s an allophone

---

### To determine this, consider a simple question...

---

## Is Will Batman?

- <img class="r-stretch" src="img/people_will_batman.jpg" alt="$txt">

---

### How do you find out if your instructor is secretly Batman?

- You look at the distributions!

---

### If you see Batman and Will in the same context, they’re two different people.

- Finding two people talking to each other in the same place is a good indication that they're independent entities

- Two different personas can't be in the same place at the same time!

- They're in **contrastive** distribution

- They happen in the same contexts

- If two sounds show up in an identical context, *they're independent phonemes*

---

### If you only see Will when there's no crime, and only see Batman where there's crime...

- This is a potential sign that Will *could be* the same being as Batman

- If Will runs into the bathroom when crime happens and Batman emerges, that's pretty good evidence

- We call this an "alternation"

- They're in **complementary** distribution

- One shows up in one context, the other in another context

- *If two sounds only show up in different contexts, or a sound suddenly changes, they're probably allophones of one phoneme*

---

### Complementary Distributions

(Batman and his alter-ego, Bruce Wayne)

---

### Contrastive Distribution

(Batman and Superman, two different heroes)
<img class="r-stretch" src="img/batmansuperman.jpg" alt="$txt">

---

### So, to sum up the pop culture references...

- Superman is an Allophone of Clark Kent

- Batman is an Allophone of Bruce Wayne

- Batman, Superman, and Will Styler are Phonemes

- ... or so Will wants you to think!
    
---

### So, that's three different perspectives on phonemic analysis

- You can look at the data and see when the meaning changes

- You can look at perception, and see what changes listeners "hear"

- You can look at the distribution, and see what sounds occur when.

---

### How are you feeling about this material?

---

## Cool.  So... how do we do it?

---

### Step 0: Check for minimal pairs

If you have a minimal pair where the meaning changes, the sounds are different phonemes and your work is done.  Always.

---

### If you have a minimal pair, the sounds are different phonemes and your work is done.

---

### Remember, the vault might be unlocked

---

### Phonemic Analysis in four easy steps!

- 0: Check for Minimal Pairs, if none...

- 1: Collect all the environments the sound you’re interested in can occur in

- 2: State the distribution of the sounds.

- 3: Decide which allophone is the basic *underlying* form
    
- 4: Write rules to derive the other allophone(s) from it based on environment

---

### 0: Check for Minimal Pairs, if none...

- Please.  PLEASE.

---

### 1: Collect all the environments the sound you’re interested in can occur in

- Write down what follows and precedes them

- Use __ to help focus your brain on the context

---

### 2: State the distribution of the sounds.

- "This one occurs before/after/around/near __"
    
- Sometimes you can only describe where one happens

- Test hypotheses!

---

### 3: Decide which allophone is the basic *underlying* form

- Choose the one you can't predict

- Often it's the “everywhere else” allophone

---

### 4: Write rules to derive the other allophone(s) from it based on environment

- We'll talk about writing rules more next week

---

### All sorts of environments trigger changes

- Adjacent sound or sounds

- Adjacent *types* of sound

- Word boundaries

- Syllable boundaries

- Sounds or later earlier in the word

- This is a bit more rare, but really neat!

---

# Let's check out some data!

---

### Are [l] and [n] allophones in Russian?

<center>
<style type="text/css">
.tg  {border-collapse:collapse;border-spacing:0;}
.tg td{font-family:Arial, sans-serif;padding:10px 5px;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;border-color:black;}
.tg th{font-family:Arial, sans-serif;font-weight:normal;padding:10px 5px;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;border-color:black;}
.tg .tg-fymr{font-weight:bold;border-color:inherit;text-align:left;vertical-align:top}
.tg .tg-0lax{text-align:left;vertical-align:top}
.tg .tg-xldj{border-color:inherit;text-align:left}
.tg .tg-0pky{border-color:inherit;text-align:left;vertical-align:top}
</style>
<table class="tg">
  <tr>
    <th class="tg-fymr">[l]</th>
    <th class="tg-0lax">Gloss</th>
    <th class="tg-fymr">[n]</th>
    <th class="tg-fymr">Gloss</th>
  </tr>
  <tr>
    <td class="tg-xldj">[lʲet]</td>
    <td class="tg-0lax">'year'</td>
    <td class="tg-0pky">[sin]</td>
    <td class="tg-xldj">'son'</td>
  </tr>
  <tr>
    <td class="tg-xldj">[lʲublʲu]</td>
    <td class="tg-0lax">'I love'</td>
    <td class="tg-0pky">[nʲet]</td>
    <td class="tg-xldj">'no'</td>
  </tr>
  <tr>
    <td class="tg-xldj">[slon]</td>
    <td class="tg-0lax">'elephant'</td>
    <td class="tg-0pky">[maʃina]</td>
    <td class="tg-xldj">'car'</td>
  </tr>
  <tr>
    <td class="tg-xldj">[dʲelal]</td>
    <td class="tg-0lax">'did'</td>
    <td class="tg-0pky">[novʲi]</td>
    <td class="tg-xldj">'new'</td>
  </tr>
  <tr>
    <td class="tg-0pky">[bʲeli]</td>
    <td class="tg-0lax">'white'</td>
    <td class="tg-0pky">[rajon]</td>
    <td class="tg-0pky">'district'</td>
  </tr>
</table>
</center>

---

### Nyet! There's a minimal pair!

- If 'lʲet' and 'nʲet' differ in meaning and differ only by that segment, they can't be allophones of the same phoneme!

---

### What about vowel length in this language?

---

### Whoa, English vowels are longer before voiced sounds!

---

### You're doing Phonemic Analysis!

- Amazing!

---

### Wrapping up

- Phonemic analysis is how we determine which sounds have an identity in the language

- ... and which are just personas of other sounds
    
- Will *may* be Batman

- Check for minimal pairs first, damnit.

---

### Next time

- We'll jump out of a plane!

- Not really.

---

<huge>Thank you!</huge>

---