These Lessons introduce you to pronunciation systems in world languages. They build on the IPA lessons.

  1. Allophones & phonemes
  2. Finding phonemes
  3. Syllables & accent
  4. Environments & rules
  5. Answers to the exercises

Allophones & phonemes

Introduction

At this point, you've already studied the International Phonetic Alphabet and you know something about how humans produce speech sounds. In short, you're familiar with the basics of how speech sounds work, which gets called phonetics. Now you will learn how to apply that knowledge to specific languages. Studying phonology will allow you to make sense of pronunciation systems. Along the way, you'll have plenty of chances to practice and refine your IPA skills.

If you've followed my other lessons in linguistics for language learners, you've noticed that I focused on concepts and terms, and highlighted those concepts with examples from a variety of languages. In this course on basic phonology, we'll take a different route. We'll get our hands dirty by studying the pronunciation of a single language.

Throughout this page, we will go about piecing together the pronunciation of standard Latin American Spanish. Again, I repeat that you need to understand basic IPA to follow along. This will enable you to pronounce the example words and phrases (the data) below.

Since we're considering a pronunciation system, we'll look for speech sounds that speakers treat as distinct sounds within that system. Phonemes are sounds that are treated as distinct from each other in a language's pronunciation system. We can test for phonemes with minimal pairs, which are explained in the next section.

We will also look for speech sounds that are treated as versions of the same sound, known as allophones. Near the end of the intro to the IPA, I brought up the example "inbox". That word contains the phoneme /n/. The phoneme /n/ is pronounced [n] in [ɪnbɑks], but can also be heard as an [m] when someone says [ɪmbɑks]. Here [n] and [m] are two allophones of the phoneme /n/.

In the next section, you'll learn some simple tests that will help you compare sounds in a single spoken language to determine its phonemes and allophones.

Complementary distribution, free variation & contrastive phonemes

If we want to identify the phonemes of a language, we must find how sounds are distinguished from each another. When switching one sound for another sound results in a different word, the two sounds are said to contrast. We can test for phonemes by using pairs of distinct words that differ in one sound and one sound only. For instance, the word pair /sej/ and /dej/ differs by one sound only. Since English speakers hear "say" and "day" as two different words, they must also hear /s/ and /d/ as two distinct sounds (phonemes), not as versions of the same sound (allophones). Such sets of words that allow us to test for contrastive phonemes are called minimal pairs.

When one sound can be switched out for another without changing meaning, we have an example of free variation. A well-known example in English is the pair /tə'mɑɾow/, /tə'meɾow/. We would write the two pronunciations as [tə'meɾow] ~ [tə'mɑɾow], or "tomayto varies with tomahto". In this case, [e] and [ɑ] are heard as two versions of the same sound. Free variants are non-contrastive, since they are not heard as two distinct sounds but as variants of the same sound. They are allophones of the same phoneme.

When one sound is found in one context and another in a different context, the two sounds are in complementary distribution. For instance, English speakers pronounce plural /s/ as voiceless [s] after voiceless consonants but voiced [z] after vowels or voiced consonants: [bʊks] versus ['fownijmz] or [dejz]. Sounds in complementary distribution are also non-contrastive, since they are not heard as two distinct sounds but as variants of the same sound. Like free variants, they are allophones of the same phoneme.

Practice Exercise

1) Let's check that your IPA reading skills are up to the task! Give the standard English spelling of every word I transcribed in IPA above.

2) Hawaiian speakers may pronounce "w" in "Hawaiʻi" like "v" or like "w". Either way. Is this an example of contrastive sounds, complementary distribution or free variation? Explain your answer.

3) French speakers pronounce "c" like an "s" before /i/ and /e/ but like a "k" before /y/, /a/, /o/ and /u/. Is this an example of free variation, complementary distribution or contrastive sounds? Explain.

4) Japanese speakers pronounce one word as /kani/ and a different word as /kaŋi/. Is this an example of complementary distribution, contrastive sounds or free variation? Explain.

5) Think for a moment about the difference between complementary distribution and free variation. Notice that American English speakers pronounce [tə'mejɾow] with the tap sound [ɾ]. Explain how the mispronunciation *[ɾə'mejɾow] suggests that [t] and [ɾ] are allophones of /t/ in complementary distribution for these speakers.

Finding phonemes

Vowel phonemes

Let's use the concept of minimal pairs to begin working out the sound structure of Spanish. Let's start by finding vowel phonemes. To be thorough, we need to compare every vowel sound with every other vowel sound in Spanish. Fortunately, we only run into five vowels: a, e, i, o and u. Are these distinct phonemes? We'll simplify the task by finding five unique words that differ in one and only one vowel sound:

Spanish word IPA translation
piso /piso/ 'floor'
peso /peso/ 'weight'
paso /paso/ 'step'
pozo /poso/ 'well'
puso /puso/ '(s)he put'

When we focus on the IPA, which shows us how the words are pronounced, we can see that the five vowels are distinct vowel phonemes. At first glance, the entire vowel system of Spanish seems to be /i/, /e/, /a/, /o/, /u/. If these are the only vowel sounds, it follows that there are no separate vowel allophones.

minimal pairs
/ase/ /ise/
'does' 'did'
/si/ /se/
'yes' 'I know'
/uje/ /oje/
'flees' 'hears'

Further minimal pairs confirm our current understanding of the Spanish vowel system.

Diphthongs

Now let's look at Spanish diphthongs (vowel + vowel combinations). Consider the following pairs. Recall that the symbol ~ stands for "varies with".

[kwento] ~ [kuento]
'story'
[kwal] ~ [kual]
'which'
[kwiða] ~ [kuiða]
'cares'
[kwota] ~ [kuota]
'toll road'
[sjento] ~ [siento]
'I feel'

Practice Exercise

1) How many vowel phonemes does Spanish have? List them in a chart arranged by their features (height and backness).

2) Look at the word list under "diphthongs" above. In your own words, what's going on in these pairs?

3) Spanish dipthongs are "rising" if they begin in u/i (i/u + vowel), but "falling" if they end in u/i (vowel + i/u). Apply your generalization from question 2 above to the diphthongs in aceite and cautivo. Give two ways to pronounce each of those words.

4) Spanish speakers hear /kae/ and /kaje/ as two different words. Similarly, they hear /le/ and /lej/ as two different words. What does this tell you about the relation between diphthongs and monophthongs ("simple" vowels) in Spanish? How is this different from English?

5) What kind of evidence would we need to disprove our claim that Spanish has the vowel phonemes above, with no distinct allophones? Imagine and write out two examples.

Consonant phonemes

Let's turn to Spanish consonants. While it's harder to compare all consonants side-by-side like we did with vowels, we'll start by considering this minimal set:

Spanish words IPA translations
de /de/ 'from'
fe /fe/ 'faith'
le /le/ 'to him/her/it'
me /me/ 'to me'
se /se/ 'him/her/itself'
te /te/ 'to you'

This list represents only a fraction of all Spanish consonant sounds. Ideally, we would test every potential phoneme against every other in minimal pairs. Let's get started by using our best judgement - here we'll test sounds that sound close enough that they're potentially confused.

minimal pairs
/lej/ /rej/
'law' 'king'
/umo/ /uno/
'smoke' 'one'
/uɲa/ /una/
'(finger)nail' 'an'
/kaɾo/ /karo/
'expensive' 'car'

Pairs like /te/ and /de/, /peso/ and /beso/ show us that Spanish regularly differentiates voiceless/voiced plosives, so we can expect /k/ and /g/ to be distinct phonemes in Spanish, too. Crucially, the pair /kaɾo/ vs. /karo/ shows us that the alveolar tap and trill are heard as two distinct sounds.

Finally, we could find even more sets of words proving the contrastiveness of the remaining phonemes in Spanish, starting with words like /oho/, /ojo/, /otʃo/ and /oso/.

Consonant allophones

If you took my course on basic syntax, you're familiar with the linguistic function of the symbol * (asterisk). It marks a word or phrase considered negative evidence. Consider the two word lists below. Pay attention to the differences between each pair of words, then answer the questions in the practice exercises below.

Consonant allophones #1
[empaɾis] *[enparis]
'in Paris'
*[emespaɲa] [enespaɲa]
'in Spain'
[gasto] *[ɣasto]
'waste'
*[kegasto] [keɣasto]
'what a waste'
[dwele] *[ðwele]
'it hurts'
*[medwele] [meðwele]
'it hurts me'
[baka] *[βaka]
'cow'
*[labaka] [laβaka]
'the cow'

Consonant allophones #2
[hale] ~ [xale]
'pull!'
[jamas] ~ [dʒamas]
'you call'
[mutʃo] ~ [muʃo]
'a lot'

Practice Exercise

1) How many consonant phonemes does Spanish have? List them in a chart arranged by their features (place of articulation, manner of articulation and voicing).

2) Look at the word pairs in the table "consonant allophones #1" above. What is the relation between n and m in Spanish?

3) Look again at the word pairs in the table "consonant allophones #1" above. What is the relation between b d g versus β ð ɣ in Spanish? How do these sounds relate to their phonemes?

4) Pay attention to the word pairs in the table "consonant allophones #2". What is the relation between [x] and [h], [tʃ] and [ʃ], [dʒ] and [j] in Spanish?

5) We've already seen that Spanish speakers distinguish between /karo/ (with trill) and /kaɾo/ (with tap), and reasonably concluded that /r/ and /ɾ/ are two distinct phonemes. BUT we find that speakers say /rie/ and not */ɾie/ - in fact, /ɾ/ doesn't show up at the beginning of words. That has the ring of complementary distribution, with /r/ showing up at the beginning of words and /ɾ/ elsewhere. In your own words, explain how we could reconcile this apparent discrepancy.

6) Some linguists propose that all "free variants" are actually socially conditioned variants. Can you guess what that might mean for cases like [mutʃo] versus [muʃo]?

Accent & Syllables


This video covers both "Accent" and "Phonotactics" below

Stress accent

Take a look at these Spanish words and phrases, then answer the questions in the practice activity below.

Spanish stress
['si] [si]
'yes' 'if'
[o'hala] [oha'la] *['ohala]
'or pulls' 'hopefully'
['kwento] ['kwento] *[kwen'to]
'story' 'I tell'
[ali'βjo] [a'liβjo] *['aliβjo]
'it relieved' 'relief'
[me'ðwele] *[meðwe'le] *['meðwele]
'it hurts me'

Practice Exercise

1) Does stress play a role in distinguishing words? Does it always distinguish words?

2) Which does Spanish stress distinguish most clearly and consistently - consonant phonemes, vowel phonemes, syllables, or words? Give examples and justify your answer.

Phonotactics (Syllable structure)

Look at this word list, paying close attention to the way Spanish divides syllables, then answer the questions in the practice exercise below.

Spanish syllables
[es'pe.ɾo] 'I hope'
[es'tɾe.ja] 'star'
[jo'si.lo'se] 'I do too know!'
[o.ha'la] 'hopefully'
[baj'laɾ] 'to dance'
[kon'ta.βas] 'you were telling'
['e.ja.pla'ne.a] 'she plans'
[pa.ɾa'tɾas] 'to the back'
[me'fal.tas'tu] 'I miss you'
[la.pa'la.βɾa'floɾ] 'the word "flower"'

Practice Exercise

1) Returning to the abbreviations C & V from our last course in basic phonetics, list every syllable type you find in the table above. (Let V stand for monophthongs or diphthongs).

2) Which type of syllable is more common above: open or close? Use this to make a generalization about Spanish syllables.

3) Which of the syllable types you identified in #1 is the most common?

4) Take a look at syllables that begin with more than one consonant. Thinking about features of consonants, can you make a generalization about those syllables?

Environments & rules

Environments

Speakers don't use sounds in isolation, but in context. Phonemes occur next to other phonemes in longer streams of speech. The immediate context of a specific phoneme is called its environment.

[empa'ɾis] between "e" and "p"
[embɾe'taɲa] between "e" and "b"
[eŋko'ɾea] between "e" and "k"
[eŋgɾambɾe'taɲa] between "e" and "g"
[enes'paɲa] between "e" and "e"
[jano] between "a" and "o"
[tan] at the end of a word
[no] at the beginning of a word

The table above shows allophones of the Spanish phoneme /n/. Those allophones occur in different environments. If we broaden our understanding of these environments to account for the component features of sounds, we can come up with more general ways to describe which allophones occur in which environments.

[empa'ɾis] /n/ before a bilabial plosive
[embɾe'taɲa] /n/ before a bilabial plosive
[eŋko'ɾea] /n/ before a velar plosive
[eŋgɾambɾe'taɲa] /n/ before a velar plosive
[enes'paɲa] /n/ between vowels
[jano] /n/ between vowels
[tan] /n/ before or after silence
[no] /n/ before or after silence

Rules

In the context of phonology, rules don't tell you what to do. They encapsulate and represent the mechanics of a language. I explained descriptive vs. prescriptive approaches to rules in the basic syntax lessons.

Rules allow us to take our simple observations in "environments" above and give them a sort of predictive power. If our rules turn out to be correct, they will account for the allophones of a phoneme in an environment.

When it comes to the Spanish phoneme /n/, we found three allophones in complementary distribution: [n] [ŋ] [m]. If we extend the observations we made above, we can propose the following generalizations:

Allophones of /n/
[m] before a bilabial plosive
[ŋ] before a velar plosive
[n] between vowels, initally & finally

Notice that we're still just observing the pronunciation of sounds in their environments. If we want to turn these into rules, we need to give some indication of how to apply those rules. Let's focus on the pronunciation of /n/ as [m] in some environments.

Labialization of /n/
/n/ becomes [m] between a vowel and a bilabial plosive

We would write such a rule this way:

Labialization of alveolar nasal (/n/)
[n] -> [m] / [+vowel] __ [+bilabial, +plosive]
"[n] becomes [m] in the environment where it is preceded by a vowel and followed by a bilabial plosive."

Since we have very few examples of /n/ before a bilabial plosive, we must look for other examples to confirm that this rule applies. Further examples from Spanish, from [tan] + ['poko] = [tam'poko] to [un] + ['bajle] = [um'bajle], will confirm this rule.

Confirmations alone don't guarantee that this rule is correct. It's just as important to look for Spanish examples that break this rule. Specifically, we listen for environments that don't match the one we've defined above where /n/ shows up as [m]. It won't take long to run into examples like [en] + ['fɾansja] = [em'fɾansja].

The example [em'fɾansja] shows us that our rule is too narrow - /f/ has the features [+fricative, +labiodental]. Labiodental and bilabial sounds both involve lips, so they are collectively called "labial". It looks like our rule applies to labials in general, not just bilabials. Also, stops, affricates and similar sounds that highly restrict airflow are collectively called "obstruents". Perhaps our rule applies to all labial obstruents.

Labialization of alveolar nasal (/n/)
[n] -> [m] / [+vowel] __ [+labial, +obstruent]
"[n] becomes [m] in the environment where it is preceded by a vowel and followed by a labial obstruent."

This rule holds up well in Spanish. Keep in mind that this, like all linguistic rules, is best seen as a work in progress.

Thank you for taking the time to learn with me. If you appreciate this approach and want more practice opportunities, consider my IPA workbook, particularly the chapters "Phonemes and Allophones", "Syllables" and "Words, Phrases and Utterances".

Practice Exercise

1) Follow the same steps in "rules" above to come up with a rule that applies to the pronunciation of /n/ as [ŋ] in certain environments. (Use Spanish examples from "environments" above.)

2) Which of the following examples confirm your rules? Which of the following examples break your rules?

[en] + [kana'da] = [eŋkana'da]
[en] + [watemala] = [eŋwate'mala]
[un] + ['gato] = [uŋ'gato]

3) Broaden the rule you made in #1 to account for any examples in #2.

Extra challenge

Continue thinking about phonological rules and environments in Spanish. Consider the allophones [b] [d] [g] and [β] [ð] [ɣ] that we noticed in "consonant phonemes" above. Begin with a simple rule for one phoneme (like /b/), then expand that rule to cover /b/ /d/ and /g/. Call the rule "Fricativization of voiced plosives".

Answers to the practice exercises

Practice Exercise 1 (Complementary distribution, free variation & contrastive sounds)

1) say, day, tomato, books, phonemes, days

2) free variation, since allophones [v] and [w] are freely swappable

3) complementary distribution, as only allophone [s] is found in one context and only allophone [k] in another.

4) the two nasal phonemes [n] and [ŋ] are contrastive

5) *[ɾə'mejɾow] is a mispronunciation because [t] and [ɾ] do not occur in the same contexts. Specifically, [ɾ] is not found at the beginning of words. Yet both [t] and [ɾ] are heard as variants of the same sound /t/ (see the "environments & rules" section in this course).

Practice Exercise 2 (Vowels)

1) Five vowel phonemes:

front central back
open i u
mid e o
close a

2) These pairs exhibit diphthongs with free variation. The diphthongs have an /i/ or /u/ component, which may be pronounced [j] and [w] or [i] and [u].

3) [ei] and [ej]; [au] and [aw]

4) Based on /kae/ and /kaje/, diphthongs and monophthongs are contrastive in Spanish. English speakers, on the other hand, hear many monophthongs as free variants of diphthongs: [se] and [sej], [fud] and [fuwd].

5) To disprove the claim that Spanish has only five vowel phonemes, we would need to identify other phonemes in the speech of native speakers by testing minimal pairs. For example, we might find a separate word */'pasy/, or that */ɑlto/ and /al.to/ are two different words. To disprove the claim that these vowel phonemes have no distinct allophones, we would need to identify variant pronunciations. One example might be if we heard *[ra'sun] for [ra'son]. If we listened carefully, we would find at least two instances of complementary distribution in Spanish: [e] for [i] 'and' before a word beginning in [i], and [u] for [o] 'or' before a word beginning in [o].

Practice Exercise 3 (Consonants)

1)

labial dental alveolar palatal velar glottal
nasal m n ɲ
plosive p b t d k g
affricate
fricative f s h
approximant j
trill r
tap ɾ
lateral l

2) [n] and [m] are in complementary distribution - where one appears, the other cannot. [m] and [p] share the feature +labial, which suggests that [n] assimilates to following consonant. In this case, we can assume an underlying phoneme /n/ with at least two allophones: [n] and [m]. That the phoneme is /n/ not /m/ is confirmed by the fact that /m/ can occur before vowels, as we saw in /me/. It is further confirmed by written Spanish, which has en París 'in Paris' and en España 'in Spain'.

3) [b] and [β] are in complementary distribution - where one appears, the other cannot. [d] and [ð] are in complementary distribution. [g] and [ɣ] are also in complementary distribution. The less restricted (less plosive) allophones [β] [ð] [ɣ] appear between vowels, while the plosives occur at the beginning of words. It looks as though the underlying phonemes are /b/ /d/ /g/.

4) [x] and [h] are free variants of the underlying phoneme /h/. [tʃ] and [ʃ] are variants of the phoneme /tʃ/. [dʒ] and [j] are free variants of the phoneme /j/.

5) There is no contradiction if /ɾ/ has two allophones, one [r] which shows up at the beginning of words and another [ɾ] between vowels, while /r/ is a separate phoneme. The two allophones of /ɾ/ are in complementary distribution. In a separate scenario, there is no contradiction if Spanish phonology simply does not allow /ɾ/ to appear at the beginning of words (that is, if its distribution is restricted to word-medial and word-final position).

6) Socially conditioned here means that the use of a specific variant will reveal things like the social status, formality level or place of origin of the speaker. For instance, [mutʃo] includes the standard or prestige pronunciation of the affricate /tʃ/, while [muʃo] contains a variant [ʃ] judged rustic, local or less educated by some speakers.

Practice Exercise 4 (Stress accent)

1) Stress can play a role in distinguishing words, but it doesn't always. Two words are pronounced ['kwento] (the accent fails to distinguish them), and strings of phonemes don't always correspond to another word when we "move" the accent mark.

2) Stress applies very clearly and consistently to syllables. You may argue for "vowel phonemes", too, since the vowel at the nucleus is louder when stressed or softer when unstressed. Any set of words from the table will do for justifying your answer - you can see that the stressed syllable is always marked with a straight apostrophe, and we are comparing words by moving the stressed syllable.

Practice Exercise 5 (Syllables)

1) V (like ['e.ja]), VC (as in [es'pe.ɾo]), CV (as in [kon'ta.βas]), CVC (like [kon'ta.βas]), CCV (as in [es'tɾe.ja]), and CCVC (like ['floɾ]).

2) 26 out of the 35 syllables in the table above end in vowels, with 9 of 35 ending in consonants. A preliminary generalization: "Spanish prefers open syllables" or "the majority of Spanish syllables are open syllables".

3) 20 of the 35 syllables have the structure CV.

4) All of syllables of the type CCV and CCVC have a plosive or affricative + liquid l or tap r.

Practice Exercise 6 (Environments & rules)

1)

Velarization of /n/
/n/ becomes [ŋ] between a vowel and a velar plosive

Velarization of /n/
[n] -> [ŋ] / [+vowel] __ [+velar, +plosive]
"[n] becomes [ŋ] in the environment where it is preceded by a vowel and followed by a velar plosive."

2) [eŋkana'da] and [uŋ'gato] confirm the rule. [eŋwate'mala] breaks the rule, since /w/ is not a plosive.

3) We can broaden the rule to account for /n/ before any velar consonant. Since the velar phonemes we identified include /g/ /k/ and /w/, this covers the example [eŋwate'mala] above:

Velarization of alveolar nasal (/n/)
[n] -> [ŋ] / [+vowel] __ [+velar, +consonant]
"[n] becomes [ŋ] in the environment where it is preceded by a vowel and followed by a velar consonant."