chapter 11.10: features and rules

As discussed in the sections above on consonants and vowels, phonemes can be categorized
by certain distinctive features. Understanding these features will be very important
when it comes time to see how sounds can change into other sounds.

Features are given in a binary form between square brackets, using the plus (+)
or minus (-) signs to indicate whether a particular phoneme has or does not have
that feature. For example, we learned that the phoneme /p/ was a bilabial voiceless
stop. Because it is a labial consonant, we can use the notation [+labial], whereas
a consonant like /t/ would be [-labial]. Because it is voiceless, we can use the
notation [-voice], whereas a sound like /b/ is [+voice]. Because it is a stop,
in which the air flow through the oral cavity is completely obstructed, it can
be notated by the feature [-continuant], where a sound like /f/ or /l/ or /e/
would all be [+continuant].

The advantage of this type of binary notation is that it makes the specific processes
of various sound changes more transparent. As an example, let us continue with the sound
/p/, a labial, voiceless, stop. It was mentioned earlier that the sound /p/ in
Proto-Indo-European became /f/ in Germanic languages like English. This change
explains why we have Latin words like piscis, pedem and pater,
all beginning with /p/, while their English counterparts, fish, foot, and
father, all begin with /f/. It is one thing to say that /p/ becomes /f/,
but it is another thing to explain how this process takes place. How can one
sound simply become another?

As already described, using feature notation we can write /p/ as

[+labial] [-voice] [-continuant]

If we look at the features for the sound [f] we have

[+labial][-voice][+continuant]

The only difference between the two sounds is in the [continuant] feature. In other
words, we are not dealing so much with one sound becoming an entirely different
sound, but with a sound altering a single feature. We can write the change thus:

[+labial][-voice][-continuant] → [+labial][-voice][+continuant]

In fact, the change from [p] to [f] is part of a larger shift, called Grimm’s Law,
in which all three voiceless stops, [p], [t], and [k], became voiceless fricatives [f], [θ],
[h]. Because this is a universal change for all stops, we can omit the place of
articulation feature and write the change thus:

[-voice][-continuant] → [-voice][+continuant]

This change covers all voiceless, non-continuants, that is, all voiceless stops, and says
that they all become voiceless continuants. In other words, it says that [p],
[t], [k] become [f], [θ], [h].

A simpler example of the use of features to describe changes can be seen if the
following instance of assimilation. We have already learned how nasals are easily
assimilated to following consonants. Take the phrase “on call.” In everyday speech,
the alveolar nasal [n] in “on” is assimilated to the velar stop [k] that begins
the word “call,” and becomes pronounced [ŋ]. Using feature notation we can write the
change [n] > [ŋ] as

[+nasal][+alveolar] → [+nasal][+velar]

List of Distinctive Features

The following features are some (from a
larger list)
that we will occasionally see in this course as we discuss phonological changes. You will notice
that most of them are terms you have already seen in the description of phonemes from Week 2.

• [+/- consonant] Segments with this feature include all consonants, whether stops, fricatives, affricates, nasals, or liquids. Vowels and semivowels lack this feature.
• [+/- continuant] This feature describes consonantal segments in which the airflow is allowed to pass through the oral cavity without stopping. It includes fricatives and liquids, but not stops or affricates.
• [+/- approximant] As learned already, approximants are segments in which there is little to no obstruction of the air. This class includes vowels, semivowels, liquids. Segments without this feature are stops, fricatives,
  affricates, and nasals.
• [+/- sonorant] Sonorant segments include vowels, semivowels, liquids and nasals. Segments without this feature are stops, fricatives and affricates.

In addition to these, there are place features such as [+labial] or [-velar] and voicing [+/- voice].

For vowels, the features [+/-high][+/-mid][+/-low], [+/-front][+/-central][+/-back], [+/-tense] and [+/-rounded] can be used.

Phonological rules

The change in Grimm’s Law wherein [p] > [f] is an unconditioned change. It
happened throughout all occurrences of Indo-European [p]. But the change from
[n] to [ŋ] is a conditioned change. It only occurs in certain phonetic environments,
specifically, when the alveolar nasal comes before a velar sound. When we use
features to describe the change like we did above, we can incorporate notation
which specifies the type of phonetic environment that is necessary to cause the change.

The change we want to describe says that an alveolar nasal becomes a dental nasal
whenever the alveolar nasal comes before a velar stop. So we write a phonological
“rule” that describes this:

[+nasal][+alveolar] → [+nasal][+velar] /____ [+velar]

Let’s break down this equation. The first part, [+nasal][+alveolar], describes the
features of the phoneme before the sound change. The arrow, → (or you can use >) represents
the process of change itself, and is followed by the features after the change
has occurred, in this case [+nasal][+velar]. This change is followed by a slash /
that means “in the following condition,” and is followed by a description of the
phonetic environment that must exist for the change to occur. In this case we have
____ [+velar]. The underscore here represents the phoneme that is being affected
by the sound change. So we can read this equation as whenever a nasal alveolar phoneme
occurs directly before a velar phoneme, then it will change to a nasal velar phoneme.

In effect, this rule is the same as simply saying [n] > [ŋ] before velars. By using
the notation of features, we can see the underlying mechanisms that cause the change.

As second example, let’s take the “rule” that describes plurals in English. With a
word like “dog,” we add the [z] to the end, so that we have [dɔgz] (or [dɑgz]).
But if we add a plural to a word like “hat” we have [hæts]. One of these words
has [z] and the other has [s]. The difference is that “dog” ends in a voiced sound
while “hat” ends in a voiceless sound. The plural is normally voiced but becomes
voiceless by assimilation when it follows a voiceless sound. We can represent
this change in the following way:

[+alveolar][+continuant][+voice] → [+alveolar][+continuant][-voice] /[-voice] _____

The environment is given as /[-voice] _____, meaning that the change occurs when the sound
[+alveolar] [+continuant][+voice] occurs directly after a voiceless sound. The
change shows that the voiced alveolar continuant, that is, [z], becomes a voiceless,
alveolar continuant, that is, [s].

In this course we will most often use the simple method for ease, but I will assume
that you know the features of the sounds that we are talking about, and in some
instances it will be useful to discuss the changes through the distinctive features.

chapter 11.9: vowel sound changes

Vowels are fundamentally different from consonants in several ways that affect how
they can change. First, all vowels are voiced, so there cannot be a change in voicing.
Second, all vowels are said in the same manner, so there cannot be changes in manner
either. Finally, vowels are not produced by placing the tongue in contact with discrete
speech organs, so there can be no movement from one part to another either. The features
of vowels that can change are frontness, height,
roundedness, and tenseness. The causes of these
changes can be neighboring consonants, although that is rare. More often, the change
is a “long-distance” change in that it is produced by assimilating with vowels in a
preceding or following syllable (since vowels form the nucleus of a syllable, other
vowels are most often in other syllables). This type of change, where one vowel is
influenced by another, is called metaphony.

One type of metaphony vowels undergo is called mutation, where a
vowel in one syllable “anticipates” a vowel in a following syllable and changes to
become more like it. The structure of a two-syllable word can be written like
this CV₁CV₂C, where C = Consonant and V = Vowel. In mutation
V₁ changes to become more like V₂, because it is easier for
the tongue to anticipate the shape it will need to be in for the following vowel.
For example, we have a word in which V₁ is a low vowel and V₂
is a high vowel, V₁ could be “raised” to a high vowel to anticipate
the height of V₂ in the following syllable.

Perhaps the most common example of mutation in English occurred just before the Old
English period, when V₂ was an [i], a high front vowel. This caused the
V₁ vowels to mutate towards the front of the mouth. We call this change
either front mutation, or i-mutation, or in German,
Umlaut, and it has had major effects throughout English.
Compare the list of words with back vowels in the first column, followed by related
words with fronted vowels in the second:

In all these examples the words on the left hand are the original words and contain
back vowels. In the period just before Old English, new words were made out of
them by adding suffixes with the high front vowel [i], and this pulled the back vowel
in the original word to the front.

Vowels can also be raised or lowered (depending on
tongue height). For example, Standard American, the pronunciation of pen
is [pɛn] but in the South it is often [pɪn], so that it sounds more like the word
pin. The mid front lax vowel /ɛ/ has been raised to the high front lax vowel
/ɪ/ because of the following nasal.

Chain shifts

Vowels can also move in what are called chain shifts. This is not a
form of assimilation or metaphony. Rather, it is a process of movement that starts
with the raising or lowering of one vowel, which results in a chain of all other
vowels moving to fill in the resulting gaps. The most famous one in English is the
Great Vowel Shift of the 15th through the 18th centuries, but other
chain shifts are currently going on in regional dialects of American English.
If you are interested, you might look at the charts and descriptions (with audio
examples) of the Northern
California vowel shift that is currently underway. The
Northern
Cities Shift explains why vowels of speakers from places like Chicago
or Detroit sound so distinctive. We will not be covering these shifts in
much detail in this class, mostly because they are still evolving, and therefore
are not part of the past history of English.

Reduction

Reduction is a change in the tenseness of vowel sounds, resulting
in less tenseness. Vowels that are reduced change in a progression from tense to
lax to schwa to zero or null, i.e., they are lost entirely. This is the standard
progression in all English vowels in unaccented syllables, although not all vowels
have or will undergo each stage of this progression, and it has had more of an impact
on English grammar than anything other sound change.

An example of reduction is the vowel in the word “man,” [mæn], a low, front vowel.
When the word stands alone and is stressed, the vowel remains in position. When the
word is joined in a compound, as in “chairman,” it is no longer stressed and therefore
is reduced. It may be pronounced as a schwa (i.e., moved from the front to the central,
that is, neutral, position) [čɛrmən], or even reduced to zero: [čɛrmn̩].

When reduction of a vowel is complete (i.e., the vowel is lost completely), then
it is called elision: the loss of a sound because it is unstressed
(elision can be used of both consonants and vowels). Elision is widespread in two of
our major grammatical categories: the s-plural and the ed-past tense. The plural
ending used to be a full syllable, as it remains in some words (“dishes”), but in
most words the -e- has reduced to zero, that is, it has been elided. The same reduction
has happened in the past tense -ed, which usually takes the form of either a [t]
or a [d] on the ends of verbs.

There are specific words to describe different types of elision, depending on where
in the word the sound has been lost:

• apheresis: loss of an unstressed vowel at the front of words:
  ’bout (for about), ’round (for around)
• syncope: loss of vowel sounds in the middle of a word:
  laboratory, comfortable [kəɱftərbəl] (which includes metathesis of the [r] and
  [t]), family, contractions such as “can’t” or “isn’t” for “cannot” and “is not”
• apocope: loss of vowels from the end of a word (this
  has happened historically in many English words; we will see examples later).

Note: these terms, apheresis, syncope and apocope, can refer to loss of
both vowels and consonants. Thus, “chile” as a pronunciation of “child” with simplification
of the final consonant cluster is an example of apocope, or “fith” as a pronunciation
of “fifth” is an example of syncope, as is “won’t” for “will not.” The pronunciations
‘im or ‘e as a pronunciation of “him” or “he” are examples of consonant apheresis.
But the most common losses are with unstressed vowels.

Epenthesis: In addition to the loss of sounds, sometimes sounds are
added. Epenthesis can be the addition of vowels, such as when “athlete” is pronounced like
“athalete,” or of consonants, such as in the following examples: “drownded” for “drowned,”
“fambly” for “family” (compare also “chimbly” for chimney” with dissimulation and epenthesis),
“hampton” for what was originally “hamton,” “thunder” for what used to be “thunor,”
and “sherbert” for “sherbet.”

chapter 11.7: Unconditioned sound changes

One of the primary areas of change within a language is the change in its system
of sounds, and much of the nineteenth-century work in linguistics was spent in
figuring out exactly how the sounds in one language corresponded to those in another.
People had long recognized that there were similarities between languages, even
when the pronunciation differed. For example, English “hound” and German “Hund”
both mean “dog” and are so similar that there must be some type of relation between
them, but the precise nature of the connection was unclear. As the nineteenth
century progressed, linguists gathered such correspondences
between the sounds of as many related languages as they could and began to notice
more and more patterns. The greatest example of this is what we call
Grimm’s Law.
Grimm noted that there were
correspondences between the consonants in Germanic languages and those in languages
like Greek and Latin, for example the correspondence between the “t” in Latin
tres, tu and tenuis and the “th” in English three,
thou and thin, or the relationship between the “p” in pater, piscis,
and pedemand the “f” in father, fish, and foot. Grimm
suggested that this correspondence was 1) a
regular one that occurred throughout the language, rather than being entirely
random, and 2) that it had happened at a historical point in time.

As ground-breaking as these two suggestions were, it was a later generation of
scholars, the Neogrammarians, who pushed them to their full conclusion. They
suggested that sound changes like this were not only regular, but that they occurred without
exception. Thus, it wasn’t just that some instances of the phoneme /t/ in Indo-European
languages become /θ/ in Germanic languages, but every single /t/ turned into /θ/.
As they put it, Die Lautgesetze kennen keine Ausnahmen: “Sound Laws know of no exception.”
They modeled the name “Sound Law” on the observable laws that had been discovered
in other sciences like Newton’s Laws of Motion. For the Neogrammarians,
laws could be formulated describing the production and changes of sounds in human
speech in the same way that Newton and others could observe the natural world
and formulate laws that described natural phenomena. The difference between a law
like Grimm’s and one like Newton’s is that Grimm’s Law occurred at one point
in time (i.e., it was historical) and then stopped. Newton’s Law of Inertia
always works.

Sound Laws like these are known as unconditioned
changes, meaning they occur throughout the language without the necessity of
there being any special phonetic conditions the word.
For example, the change from /t/ to /θ/ that we see in a correspondence like Latin
tres and English three did not happen because of the specific
vowel in the word, or the presence of the
phoneme /r/. No special conditions
were needed for it to take place. This separated these types of changes
from the conditioned changes that only occurred under specific
conditions.

Perhaps the most famous unconditioned sound changes in English are Grimm’s Law and
the Great Vowel Shift,
but there are others occurring today. Cities like Buffalo and
Rochester in the Northern US are undergoing what is known as the Northern Cities Shift,
and while parts of California
and the southern US
are in the midst of their own unconditioned vowel changes.
The “cot-caught”
merger in which the vowels in these two words start to rhyme
has affected a large part of the US.

chapter 11.5: vowels

Like consonants, vowels are made by expelling air from the lungs through the
vibrating vocal cords and into the oral cavity. Unlike the consonants, however,
when we make the vowel sounds we do not close or greatly obstruct the air flow.
Instead, we move the tongue into various places and raise or lower our lower jaw,
and by doing so, we alter the frequency of the vibrations caused by the air flow
and vocal cords. The variations in frequency are perceived as different phonemes.
In General American English twelve distinct vowel phonemes can be distinguished,
as shown in the following chart.

The chart represents the inside of the mouth with the teeth towards the left.
Where the chart says “front” is the front of the mouth, specifically the area
just behind the teeth, the alveolar ridge. The right-hand side of the chart,
which is labeled “back” is the back of the mouth, roughly where the velum, or
soft palate, is. Similarly, the three sections labeled “high,” “mid,” and “low”
represent the roof of the mouth down to the lower jaw. The symbols are placed
at the various locations within the mouth where the tongue moves in order to
make the relevant vowel sound. The symbols in the white area represent the
vowels in which the tongue is fairly “tense” while saying them, while the symbols
within the grayed area are made with a relaxed tongue. Finally, the four vowel
symbols within the box in the back, high, and mid area are vowels which are
pronounced while the lips are “rounded.” Further explanation of this terminology
will be found below.

An interactive vowel chart, with audio examples, can be found here.

Distinctive Features

Vowels are distinguished by four distinctive features:

• height: Vowel height can be high,
  mid, and low, depending on whether the tongue
  is near the roof of the mouth (i.e., high), or near the floor of the mouth. For
  example, say just the vowels of the words beet, bait, bat, concentrating
  on where your tongue is. The first has a high vowel, the second a mid vowel,
  and the third a low vowel.
• frontness: Vowels can be front,
  central and back, depending on where the
  tongue is. English generally distinguishes only front and back vowels.
  For example, say just the vowels in the words bait and bought
  and feel the position of the tongue sliding forward and back against the
  roof of the mouth. Both are mid vowels, but the first is a front vowel
  and the second is a back vowel.
• tenseness: Vowels can be tense or
  lax, depending again on the tongue. As examples, say
  beet, which has a tense vowel, and then say bit, which has a
  lax vowel. Tense and lax only applies to high and mid vowels. The low vowels
  do not have tense distinctions.
• roundedness: Vowels can be rounded or unrounded
  depending on the shape of the lips. In English generally the front vowels are unrounded
  and the back vowels (at least the mid and high ones) are rounded. For example,
  say the vowels in beet and boot. Besides the tongue
  sliding back and forth to the front and back, you will notice that your lips
  become rounded to say boot but they are not rounded when you say
  beet.

In what follows I will describe the 15 vowel phonemes of English. All
vowels are produced by expelling air through the oral cavity, with minimal
restriction provided by tongue placement and lip rounding. All vowels are
voiced.

High Vowels

• front, tense, unrounded /i/. Examples: bee, beat, machine
• front, lax, unrounded /ɪ/. Examples, bit, flip
• back, tense, rounded /u/. Examples: moon, mood, rude
• back, lax, rounded /ʊ/. Examples: put, foot

Mid Vowels

• front, tense, unrounded /e/. Examples: bait, day, made, name.
• front, lax, unrounded /ɛ/. Examples: bet, head, peck.
• back, tense, rounded /o/: Examles: boat, hope, moan.
• back, lax, rounded /ɔ/: Examples: bought, caught,
  hawk (Note: this phoneme is often not part of the
  Western American dialect)

Low Vowels

• front, unrounded /æ/. Examples: cat, bat.
• back, unrounded /ɑ/. Examples: cot, hock,
  pot, bother, father.
  (People with a Western American dialect will also pronounce
  bought, caught with this vowel
  instead of /ɔ/).

Two other vowel sounds occur

• /ʌ/. This symbol represents is a central, mid, unrounded, lax vowel.
  For many Americans it does not contrast with the following vowel, but is used
  only for stressed vowels. Examples: cup, luck,
  love.
• /ə/. This is the schwa, the reduced central >mid vowel
  found in most unaccented syllables. It comprises several different vowel
  sounds, all found in unaccented syllables, specifically [ə], as in
  about, sofa, occur,
  nation. And a slightly higher vowel, [ɨ], as in
  dishes, messes, rated,
  raided. For the purposes of this course you can transcribe
  both of these as /ə/ if you wish, but you should be aware of both symbols
  in case you see them elsewhere.

Diphthongs

All the above vowel sounds are monophthongs, that is,
they are single vowel sounds. As such, they are distinguished from
diphthongs, two vowels sounds spoken together in a
single syllable. There are three phonemic diphthongs in American English.
Each consists of a main vowel, followed by an offglide which is either a
front or back high vowel. (Note: The “ph” in the
words monophthong and diphthong can be pronounced either as an [f] or a [p]).

• /aɪ/. This sound starts with /a/, a low front vowel very similar to /æ/,
  followed by a glide toward /ɪ/. Examples: rice,
  wife, eye.
• /aʊ/. This sound also starts with /a/, but the glide is the rounded back vowel
  /ʊ/. Examples: house, brown, cow.
• /ɔɪ/. This sound starts with the mid back rounded vowel /ɔ/ before
  moving to the front unrounded /ɪ/. Examples: joy,
  boy, boil.

A note on the pronunciations of monophthongs as diphthongs in American
English: Most speakers of American English pronounce the simplex vowels like
/e/ and /o/ as the diphthongs [eɪ] and [oʊ]. As a result, some textbooks and
online resources give represent these vowels in IPA as diphthongs. In British
resources /o/ is often represented as [ɛʊ]. We will make these distinctions when
necessary, but otherwise will treat these vowels as if they were pronounced
simply /e/ and /o/.

chapter 11.4: consonants

The following chart gives the IPA symbols for each of the 24 consonants used in
General American English. An interactive chart with sound examples of each phoneme
tags: phonology can be found here

One way to understand phonemes is to categorize them according to their
distinctive features. For consonants, the distinctive features
are categorized under several broad headings:

• Place of Articulation: The organ of speech which is the
  primary location for the production of the phone
• Manner of Articulation: How the air flow is obstructed,
  whether stopped entirely or restricted by one of the organs of speech
• Voicing: Whether the vocal cords vibrate or not during
  the production of the phone

Place of Articulation

The places of articulation are given across the top of the chart above. They are

• bilabial: phones that are made by obstructing the air
  flow with both lips
• labiodental: phones that are made by obstructing the air
  with a combination of the lips and teeth
• dental: phones that are made by obstructing the air with
  the tongue against the teeth
• alveolar: phones that are made by obstructing the air with
  the tongue against the alveolar ridge
• palatal: phones that are made by obstructing the air with
  the tongue against the hard palate
• velar: phones that are made by obstructing the air with
  the tongue against the soft palate or velum
• glottal: phones that are made by obstructing the air
  with the glottis

These features can be referred to with notation such as [+alveolar],
signifying an alveolar sound, or [-velar], signifying that the sound is not
velar.

Manner of Articulation

The manners of articulation are given in the left-hand column of the chart. They are

• Plosive (also called stops): phones that
  are made by stopping the air flow completely (and then releasing it)
• Fricative: phones that are made by restricting but not
  stopping the air flow
• Affricate: phones that are made by stopping the air flow
  and then restricting it when released (i.e., a combination of a stop and a
  fricative)
• Nasal: phones that are made by opening the nasal cavity
  so that the air flow by resonate inside it
• Approximant: phones that are made when the articulators
  approach each other, but not as close as in the fricatives
• Semivowel: a subset of approximants, phones that act as
  consonants or vowels depending on the environment they are in.

As with the places of articulation, the manner of a particular sound can also
be specified with feature notation such as [+stop]. In order to refer to a
phoneme that is made by stopping the air with both lips, we can use the features:
[+stop][+bilabial]. One more feature is the term continuant,
which can be referred to in feature notation as [+continuant] or [-continuant].
Continuants are the sounds in which the air is not stopped from going through
the oral cavity. Thus, plosives are [-continuant] because they stop the air
entirely. Nasals are also [-continuant] because in them the air resonates in
the nasal not the oral cavity. Affricates begin by being [-continuant] but
end as [+continuant]. All other consonants are [+continuant].

Voicing

The final distinctive feature of consonants is voicing, that is, whether the
vocal cords vibrate. You can test whether or not a sound is voiced by placing
your hand on the front of your throat while you speak. If you feel a vibration,
as when you say [b] or [d], then the sound is voiced, or
[+voice]. If you do not feel a vibration, as in /p/ or /t/ or /f/, then the
sound is voiceless, or [-voice].

English Consonant Phoneme Symbols

I will now discuss each of these 24 consonant symbols given above, listing them
according to their Manner of Articulation. Make sure to learn each one according
to its three features (place of articulation, manner of articulation and
voicing) and its IPA symbol.

Plosives, or Stops: /p/ /b/ ~ /t/ /d/ ~ /k/ /g/ (The first of each pair is voiceless, the second voiced).

• Bilabial: /p/ and /b/. The first of these symbols
  represents the voiceless bilabial plosive. Bilabial plosives are made
  by stopping the air with both lips (bi-labial) and then releasing it.
  Examples: pin, spin, apt,
  stop. The second symbol, /b/, is made in the same way,
  except it is voiced by the vibration of the vocal cords. Examples:
  bin, tab, table.
• Alveolar: /t/ and /d/. These two plosives are made by
  stopping the air flow by placing the apex (tip) of the tongue against the
  alveolar ridge and then releasing it. The first is voiceless and the second is
  voiced. Examples: top, stop, bat
  and dot, bad, ladder.
• Velar: /k/ and /g/. These sounds are made by stopping the
  air flow by placing the dorsum (back) of the tongue against the velum and then
  releasing it. The first is voiceless, and the second is voiced. Examples:
  kick, making; gall,
  ogre, hog. (Note, the symbol /g/ never
  represents the so-called “soft” g-sound as in gel).

Fricative: /f/ /v/ ~ /θ/ /ð/ ~ /s/ /z/ ~ /ʃ/ /ʒ/ ~ /h/
(The first of each pair is voiceless, the second voiced). Fricatives are made by
creating a restriction in the oral cavity through which air must pass. The
friction of the air passing creates the specific sound.

• Labiodental: /f/ and /v/. These sounds are made by placing
  the upper teeth against the lower lip and breathing through the restricted
  space. The first is voiceless and the second is voiced. Examples:
  fall, after, awful,
  off, and vine, over,
  leave.
• Dental (sometimes called interdental): /θ/ /ð/ (the first
  of these symbols is the Greek letter theta; the second is an Icelandic
  letter called “eth”). These sounds are made by placing the tongue between
  the teeth and breathing through the restricted space. The first is voiceless
  and the second is voiced. Examples of /θ/: thin,
  breath, wealth. Examples of /ð/:
  then, breathe, weather.
• Alveolar : /s/ /z/. These sounds are made by air passing
  between either the apex of the tongue and the alveolar ridge. Examples of /s/:
  sit, cell, lessen,
  ice, hiss. Examples of /z/: zoo,
  wisdom, is.
• Post-alveolar: /ʃ/ /ʒ/ (the first of these is called an “esh”
  or a “long s”; the second is an “ezh”). These sounds are made by air passing
  rapidly between the tongue and the area just behind the alveolar ridge.
  Examples of /ʃ/: shoe, ocean, nation,
  ash. Examples of /ʒ/: measure, azure,
  decision.
• Glottal: /h/. Examples of /h/: hello,
  high. There is no voiced counterpart to this glottal fricative
  in Modern English. It is more historically accurate to consider /h/ a velar
  sound, since it is generally a reduced sound of the velar fricative /x/ which
  English used to have, and still does in some dialects (as in Scottish
  loch), or which you will hear in foreign words like German
  Bach. There was at one time a voiced velar fricative /ɣ/, which will
  come up in discussions of Old English. To make this sound, make the sound in
  Scottish loch, and then vibrate your vocal cords.

Affricate: /tʃ/ /dʒ/ (The first is voiceless, the second voiced).
As you can tell from the symbols, these sounds are really a combination of two
others, but in English we still consider them distinct phonemes.

• Alveo-palatal affricate /tʃ/ /dʒ/. The onset of the sound
  is made by stopping the airflow with the tongue
  against the alveolar ridge, as in /t/ and /d/, but instead of a full release,
  the airflow is released into the restricted flow of the fricatives /ʃ/ or /ʒ/.
  Examples of /tʃ/: church, chin,
  hatch, nature. Examples of /dʒ/:
  judge, jelly, edge.

Nasal: /m/ /n/ /ŋ/ (All three are voiced). Nasals are made
by stopping the air in the same way as in plosives, but rather than releasing
the air, the air is allowed to vibrate in the nasal cavity.

• Bilabial: /m/. This sound is made by stopping the air flow
  by closing both lips and then allowing the air to resonate within the nasal
  cavity. It is voiced. Examples: man, ham,
  hammer.
• Alveolar: /n/. This sound is made by stopping the airflow
  by placing the apex of the tongue against the alveolar ridge and then allowing
  the air to resonate within the nasal cavity. It is voiced. Examples:
  no, man, manner.
• Velar: /ŋ/ (the symbol is called an “eng” or an “engma”).
  This sound is made by stopping the airflow with the dorsum of the tongue
  against the velum and then allowing the air to resonate within the nasal
  cavity. It is voiced. Examples: sing, singer,
  finger. (Note: although we spell this sound with two letters,
  it is a single phoneme).

Approximant: /l/ /
r/ (both voiced)
An approximant is a consonant sound in which the restriction of airflow is not enough to cause any turbulence as described in the previous sounds.

• Lateral alveolar: /l/. The /l/ is made by placing the tip
  of the tongue against the alveolar ridge, and allowing the air to escape
  around the sides of the tongue (i.e., laterally). Examples of /l/:
  lamp, fall.
• Central alveolar: /r/. This sound is produced by placing
  the tip or blade of the tongue near the alveolar ridge and allowing air to
  flow centrally through the gap. Some people make it by curling the tip
  of the tongue backwards, so that it is called a retroflex approximant.
  Examples of /r/: red, very,
  near. (Note: in the chart above I use the symbol /ɹ/, which
  is technically a more accurate symbol for the type of “r” we make in American
  English, but for ease of typing it will be fine to use /r/.)

Semivowel: /j/ /w/ (both voiced) These sounds are made in the same way as an approximant, allowing barely restricted air to flow through, in a manner
very similar to vowels but in environments in which a consonant is needed, i.e., they are usually followed by vowels rather than consonants..

• Palatal: /j/ This sound is made by placing the
  tongue near the hard palate and allowing air to flow through. It is called a
  semivowel because it is produced in the same way as the vowel /i/.
  Examples of /j/: year, yellow,
  yes, you. (Do not confuse the symbol
  /j/ with the letter <j> in words like “judge”).
• Labiovelar: /w/. This sound is made with the back of the
  tongue near the velum and with rounded lips, in the same way as the vowel
  /u/. Examples of /w/: west, wall.

These 24 phonemes represent the consonant phonemes of General American English. You must learn each symbol and the sound it represents, and be able to describe each phoneme according to its three features: voicing, place of
articulation, manner of articulation. I advise you to say these sounds out loud over and over (whispering will not do) until you understand each of them.

chapter 11.3: phonemes

A phoneme is the smallest meaningful unit of sound.
The key word in this definition is meaningful. Although speech is continuous,
as we listen to someone speaking, we are able to break the words and phrases
into individual segments, to which we ascribe meaning. For example, we analyze
the word “pit” as composed of three distinct segments, each of which is a meaningful
phoneme: /p/, /ɪ/, and /t/, or simply /pɪt/. To signify that we are talking about
phonemes and using the IPA, we write the symbols between slashes / /. When we
discuss actual pronunciation, regardless of phonemic status, we use square brackets
[ ].

Minimal pairs

In order to determine if a sound is a phoneme, we can set up a pair of words in
which the only difference is one single sound. For example, take the word “pit”
again, written in IPA as [pɪt], and let us pair it with the word “bit”, in IPA
[bɪt]. These words differ by only a single sound, the first one;
there is the minimal amount of difference between the two words. Thus, we call
the set a minimal pair. Because we as English speakers can distinguish
the utterances [pɪt] and [bɪt] as two distinct words, we can determine that
/p/ and /b/ must have a meaningful
difference to us. This difference allows us to establish that /p/ and /b/
are both phonemes in English. Another way to think of the word “meaningful” is
as a contrastive difference. The sound /p/ has a contrastive difference from
other sounds such as /b/ or /t/ or /f/.

We can establish other phonemes of English in the same way, by taking a word
like “pit,” and changing it by one sound at a time. Take “pit,” and change the
first sound to “b”: “bit.” Now change the first sound to “h”: “hit.” Now change
the vowel to the “ee” sound: “heat.” Now change the last sound to “d”: “heed.”
Now change the first sound to “m”: “mead.” Now change the vowel to the long “a”
sound: “made.” And so on. You can see here that the spelling changes in
various unexpected ways, but if we write these words in IPA it is clearer that
only one sound is changing at a time: [pɪt] ~ [bɪt] ~ [hɪt] ~ [hit] ~ [hid] ~
[mid] ~ [med]. This method can be used to build an inventory of phonemes of a
particular language. This short chain of minimal pairs allows us to determine
that the following are phonemes of English: /p b h ɪ i t d m e/.

It may seem obvious that all these sounds are “meaningful,” but there are many
different ways to pronounce sounds that have no meaningful distinction between
them. Besides all the grunts, burps, squeaks, laughs, etc. that we can make with
our mouths, we have many different types of sounds we can make that are not
phonemes. For example, pronounce the word “bid” with a long, drawn-out vowel.
We could try to convey this vowel in English spelling as “biiiiiiiid,” or we
could write it in IPA by adding the symbol [ː], which is used to indicate a
lengthened vowel: [bɪːd]. Now let’s contrast them as a minimal pair: [bɪd] and
[bɪːd]. These are two very different ways to say the word, and yet the word
remains “bid.” Therefore, these two vowels fail the minimal pair test, which
tells us that [ɪː] and [ɪ] and not two distinct phonemes in English. They are
simply two non-meaningfully different ways to say the phoneme /ɪ/. These variations
within a phoneme are called allophones, a concept that will
be discussed in more detail later.

One more example of non-meaningful variants within a phoneme will suffice. Again,
take the word “bid,” but this time pronounce the initial consonant with a heavy
breath of air. This is an “aspirated” [b], (from the word “aspiration” which means
breathing). We can signify it in IPA by adding a superscript h like this:
[b^h]. If we contrast “bid” said in the usual way with the word with
an aspirated b: [bɪd] and [b^hɪd], we find that the pair do not create
two distinct words and therefore the sounds fail the minimal pair test. Therefore
there is no contrastive difference between [b] and [b^h] and they are
not two distinct phonemes. Instead, they are simply two allophones of the
phoneme /b/.

Every language has its own set of phonemes.
There are languages in which the length of a vowel changes the meaning of a word.
This was the case in English 1000 years ago, when the word “ac” with a short vowel
[ɑk] meant “but” and the word “ac” with a long vowel [ɑ:k] meant “oak tree.” Thus,
in Old English /ɑ/ and /ɑ:/ were two distinct phonemes, whereas now in Modern
English they are two allophones of the same phoneme /ɑ/. Similarly, there are
languages like Hindi in which /b/ and /b^h/ are two distinct phonemes. The
number of phonemes in different languages can also vary widely. Humans have the
ability to produce about 600 different consonant sounds and 200 vowel sounds,
but English only uses about 24 consonant phonemes and 15 vowel phonemes, for a
total of 39 distinct meaningful sounds.
Ubykh,
an extinct language from the Caucasus, had 86 phonemes–84 consonants and only 2
vowels! Hawaiian,
on the other hand, has only 13 distinct phonemes, 8 consonants and 5 vowels. It
does not matter how many phonemes a language has; what matters is that the set of
phonemes function as a system, where those sounds and only those are meaningful and
capable of carrying information. (If you are interested in how babies begin to
distinguish the phonemes of their own language, see
“Does an 11-week-old
know what English sounds like,” from which the figures above on the total
number of consonant and
vowel sounds humans can make were taken).