The International Phonetic Alphabet

If this is your first encounter with phonetics, you had better start reading the general phonetic notes before trying to make sense of the tables.

Table of contents

The tables

The International Alphabet was invented by the International Phonetic Association in 1888 and has undergone constant revision since. Here is the official chart.

The tables shown here follow the manner of the official chart with the addition of an “epiglottal” column in the consonants table because there is no reason not to do so (there are as many symbols in the epiglottal as in the glottal locations, and it is interesting to parallel the pharyngeal, epiglottal and glottal consonants), and similarly an “implosives” row.

When the consonants' table shows two characters in a square, the first is the voiceless sound and the second is voiced; when only one character is shown, it is voiced, except for the glottal and epiglottal plosives, which are voiceless (it is — probably — not possible to pronounce a voiced pharyngeal, glottal or epiglottal plosive). A grey rectangle (if your browser displays it, that is) means that the corresponding sound is impossible to pronounce, or meaningless; the voiceless pharyngeal plosive is possible (only the voiced counterpart is not), it just doesn't have a symbol. An empty square means that the sound is (presumably) possible, but no symbol has been defined (because no language uses it, or because it is just as convenient to use diacritics over an existing symbol). Dentals, alveolar and postalveolar consonants use the same symbols except for fricatives: if necessary, diacritics can be used to mark them apart; the standard version is alveolar (though in my opinion, the approximant used to mark the English ‘r’ (lowercase turned r, number 151) is distinctly postalveolar, even slightly retroflex).

The vowels' table attempts to map the vowel symbols. However, when used phonemically, their value varies greatly from language to language, much more than consonants. When two vowels are shown in a square, the first is unrounded and the second is rounded (though it is by no means true that all “unrounded” vowels are equally unrounded, or even that all “rounded” vowels are equally rounded). When a single vowel is shown, it varies in roundedness, being generally of somewhat neutral value; however, the ash (lowercase ae ligature, number 325) is a distinctly unrounded front semi-open vowel, and the upsilon (number 321, which looks like a turned small capital omega) is a distinctly rounded back-center semi-closed vowel.

Regular IPA

The following tables show the regular IPA symbols.

Displayed with graphics

These tables show one small (PNG) graphics file for each character. These graphics file were prepared from the “TIPA” TeX/LaTeX fonts written by Rei Fukui (fkr@tooyoo.l.u-tokyo.ac.jp) using an ad hoc Perl script to convert the METAFONT output to individual PNG files.

The images are numbered using the International Phonetic's Association standard numbering of IPA symbols. If your browser is text-only, you will see the corresponding numbers in place of the images.

CONSONANTSBilabialLabiodental DentalAlveolarPostalveolarRetroflex PalatalVelarUvular PharyngealEpiglottalGlottal
Plosive[101] [102] [103] [104][105] [106] [107] [108][109] [110][111] [112]   [XXX][173] [XXX][113] [XXX]
Nasal[114][115] [116][117] [118][119][120] [XXX][XXX][XXX]
Trill[121] [122] [XXX][123] [XXX][XXX]
Flap [124][125] [XXX] [XXX][XXX]
Fricative[126] [127][128] [129] [130] [131][132] [133][134] [135][136] [137] [138] [139][140] [141][142] [143] [144] [145][172] [174][146] [147]
Lateral fricative[XXX][XXX] [148] [149] [XXX][XXX][XXX]
Approximant[150] [151][152] [153][154] [XXX][XXX]
Lateral approximant[XXX][XXX] [155][156] [157][158] [XXX][XXX][XXX]
Implosive[160] [162] [164][166][168] [XXX][XXX][XXX]
VOWELSFront CentralBack
Close[301] [309] [317] [318][316] [308]
Semi-close[319] [320]   [321]
Close-mid[302] [310] [397] [323][315] [307]
Mid [322]
Open-mid[303] [311] [326] [395][314] [306]
Semi-open[325]   [324]
Open[304] [312] [305] [313]
CLICKS
Bilabial[176]
Dental[177]
Alveolar or postalveolar[178]
Palatoalveolar[179]
Alveolar lateral[180]
EXTRA SYMBOLS
Voiceless labial-velar fricative[169]
Voiced labial-velar approximant[170]
Voiced labial-palatal approximant[171]
Voiceless postalveolo-velar fricative[175]
Alveolar lateral flap[181]
Alveolo-palatal fricatives[182] [183]

Using Unicode characters

These tables contain the Unicode characters of the various phonetic symbols. If you are lucky enough to have a functional Unicode web browser with the appropriate fonts correctly installed (something rather unlikely!), it should look just as pretty as the previous table, and it should work even if your browser is text-only. If not, you will probably see lots of question marks, unreadable or incorrect characters, or simply blank squares. This has been tested using Mozilla on Linux, with an impressive number of fonts; and also using lynx on an xterm with Unicode support.

CONSONANTSBilabialLabiodental DentalAlveolarPostalveolarRetroflex PalatalVelarUvular PharyngealEpiglottalGlottal
Plosivep b t dʈ ɖ c ɟk gq ɢ    ʡ  ʔ  
Nasalmɱ nɳ ɲŋɴ    
Trillʙ r  ʀ   
Flap ɾɽ     
Fricativeɸ βf v θ ðs zʃ ʒʂ ʐ ç ʝx ɣχ ʁ ħ ʕʜ ʢh ɦ
Lateral fricative   ɬ ɮ    
Approximantʋ ɹɻ jɰ   
Lateral approximant   lɭ ʎʟ    
Implosiveɓ ɗ ʄɠʛ    
VOWELSFront CentralBack
Closei y ɨ ʉɯ u
Semi-closeɪ ʏ   ʊ
Close-mide ø ɘ ɵɤ o
Mid ə
Open-midɛ œ ɜ ɞʌ ɔ
Semi-openæ   ɐ
Opena ɶ ɑ ɒ
CLICKS
Bilabialʘ
Dentalǀ
Alveolar or postalveolarǃ
Palatoalveolarǂ
Alveolar lateralǁ
EXTRA SYMBOLS
Voiceless labial-velar fricativeʍ
Voiced labial-velar approximantw
Voiced labial-palatal approximantɥ
Voiceless postalveolo-velar fricativeɧ
Alveolar lateral flapɺ
Alveolo-palatal fricativesɕ ʑ

ASCII IPA

Representing IPA characters on a computer is very inconvenient: the characters themselves are unusual and consequently not found in most fonts, and diacritics are many. Although the progressive acceptance of Unicode will probably alleviate the problem, it is still a few years before tables like the previous ones are displayed correctly everywhere. Besides, even when it works correctly, there are places where using Unicode representations is inconvenient, for example in Usenet articles. For this reason, a pure-ASCII representation of the IPA has been devised, mostly for use on the sci.lang Usenet group. Its specification is by Evan Kirshenbaum (kirshenbaum@hpl.hp.com).

I have made some changes in the following table with respect to Evan's specification to correct some (IMHO) illogical features (or contradictions) in the document, and to bring the representation in line with the latest standard of the IPA itself (1996). I have added an “epiglottal” feature, represented by “epg”. I have represented the voiced uvular fricative by ‘Q"’, rather than ‘g"’, which ought to be a voiced plosive (note that the same objection could be made about the ‘r"’ symbol which is here shown as a trill whereas it ought to be an approximant, but in fact the corresponding IPA symbol is often used as an approximant, so it does not matter very much). I have represented the labiodental approximant by ‘v<apr>’ because it seems much more appropriate than the proposed ‘r<lbd>’ for a language such as Hindi (of course, different symbols can be used for different languages, even, in strict phonetic transcription, when they represent the same sound).

CONSONANTSBilabialLabiodental DentalAlveolarPostalveolarRetroflex PalatalVelarUvular PharyngealEpiglottalGlottal
Plosivep b t dt. d. c Jk gq G    ?<epg>  ?  
NasalmM nn. n^Nn"    
Trillb<trl> r<trl>  r"   
Flap **.     
FricativeF Vf v T Ds zS Zs. z. C C<vcd>x QX Q" H H<vcd>h<epg> Q<epg>h h<vcd>
Lateral fricative   s<lat> z<lat>    
Approximantv<apr> rr. jj<vel>   
Lateral approximant   ll. l^L    
Implosiveb` d` J`g`G`    
VOWELSFront CentralBack
Closei y i" u"u- u
Semi-closeI I.   U
Close-mide Y @<umd> o"o- o
Mid @
Open-midE W V" O"V O
Semi-open&   @<sml>
Opena a. A A.
CLICKS
Bilabialp!
Dentalt[!
Alveolar or postalveolart!
Palatoalveolarc!
Alveolar laterall!
EXTRA SYMBOLS
Voiceless labial-velar fricativew<vls>
Voiced labial-velar approximantw
Voiced labial-palatal approximantj<rnd>
Voiceless postalveolo-velar fricativeS"
Alveolar lateral flap*<lat>
Alveolo-palatal fricativess^ z^

General phonetic notes

Segments

One of the fundamental (but perhaps debatable) postulates behind the International Phonetic Alphabet is that speech, or, at any rate, the meaningful and “abstract” content of speech, can be divided in atomic sound utterances called phones or segments. For example, the English word “dog” consists of three segments, which happen to be represented by one letter each; the middle one is a vowel (a rounded back vowel that is relatively open, though the degree of openness depends on the variant of English being considered) and the first and last are consonants (a voiced alveolar plosive, and a voiced velar plosive, respectively). Another important postulate is that some phones are common between languages. For example, the segment [v], corresponding to the first consonant (‘v’) in the English word “view” is the same as that which starts the French word “ver”, the German word “weg” or the Russian word “вода”.

Naturally, the validity of these hypotheses is not complete. First, spectral analysis of sounds show that dividing a continuous utterance into well-delimited segments is hopeless: there exists all sorts of complicated transition patterns, and a sound only makes sense in a certain phonetic context. Second, phones do not fall in neatly defined little bins, and there exists a continuum of possibilities (this is even more true for vowels than for consonants). Third, no two phones are exactly identical: within a same language, people pronounce things differently, and between different languages, no two sounds are ever exactly identical. Still, despite all these limitations, the hypotheses we have formulated are a useful basis that works well in practice.

Although there is a slight correspondance between letters and segments, it is by no means true that every letter corresponds to a single segment or vice versa. In English, this is “particularly false”, in fact. In the name “Vaughan”, the letter sequence ‘augha’ corresponds to a single segment; in the word “music”, the letter ‘u’ corresponds to two segments (the first being a consonant and the second a vowel). Furthermore, a given letter or letter sequence can correspond to several different segments or segment sequences according to context, and a given segment or segment sequence can be represented by several different letters or letter sequences in various words. Again, English is particularly badly behaved in this respect: compare “laughter” and “daughter” or “how” and “low” on the one hand, “bury” and “berry” or “son” and “sun”.

All segments are pronounced using the vocal tract, from the larynx to the lips, including the nose. Sounds which could be produced by other parts of the body (such as by snapping one's fingers) are not included in phonetic studies: it seems that no language uses them for articulated communication, though there is no reason why they should not be. (Of course, body gestures are used by the hearing impaired, but the language in question is visual and not phonetic, its very ratio operandi in fact.)

Different kinds of segments

Voiced and voiceless phones

During normal speech, in particular when one is not whispering, the vocal cords are vibrating part, but not all, of the time. Phones which are pronounced with the vocal cords vibrating are called voiced, and the others are called voiceless. This is illustrated by the difference between the English words “view” [vju:] and “few” [fju:]: the initial segment (a labiodental fricative) is voiced in the former and voiceless in the latter. This should be perfectly clear if you put your hand to your throat: you should feel your vocal cords vibrating when saying “view” whereas when saying “few” they start vibrating only after the ‘f’.

Or, to say things differently: voiced segments are pronounced with a simultaneous “humming” sound (vibration of the cords) whereas voiceless segments are pronounced without it.

Despite this, even when whispered, the [v] and [f] segments do not sound exactly identical (though the differences is much slighter than in normal speech): this is presumably because the vocal cords, although not vibrating, are not quite in the same position when [v] is pronounced as when [f] is pronounced; if the difference between a devoiced [v] and an [f] is to be emphasized, the former can be called a lenis and the latter a fortis.

Consonants and vowels

Vowels are sounds which are produced with the vocal tract unobstructed, whereas consonants are those which are produced with some obstruction to the flow of air. For example, if the mouth is opened widely, with the tongue kept at rest, completely out of the way of the air, while the vocal cords are made to vibrate, the sound [a] is produced: it is a vowel (not found in English, but in French, as in the word “ami”), and it is the furthest removed from any consonant. On the other hand, if the lips are closed so as to interrupt the flow of air, and then opened again, all while the vocal cords are vibrating, the sound [b] is produced: it is a consonant, and specifically a bilabial plosive.

Vowels are voiced except when they are whispered. Consonants, on the other hand, can be voiced or voiceless.

The distinction between vowels and consonants is not always clear-cut, and there exists a fuzzy boundary region. First note that any consonant (except perhaps a plosive) can be used as a syllable nucleus just as if it were a vowel: this is particularly common in English for the consonants [l] and [n]. For example, in way the word “bottle” is usually pronounced, the second syllable consists of the consonant [l] all by itself (though it is always possible to add a weak schwa vowel before it to serve as syllable nucleus in its stead). This does not concern the sound of the segment itself but rather its duration and phonetic environment.

Second, there exists sounds which are just in between vowels and consonants. In the English word “yeast”, the ‘y’ is clearly a consonant (a voiced palatal fricative, number 139). In “year”, it is still a consonant, but it is much more open: it has changed from a fricative to an approximant. On the other hand, the ‘e’ in “Seattle” is ever so slightly more open, and, this time, it is a vowel.

Kinds of consonants

Plosives (also called “occlusives” or “stops”) are consonants which are produced by completely interrupting the flow of air by obstructing it in some way. For example, the two consonants in the English word “copy” are (voiceless) plosives. A plosive consists of three parts: first, the flow is interrupted (this is the attack), then for some time there is no flow (and thus no sound if the plosive is voiceless), and finally comes the release which terminates the plosive.

If the plosive is the first phone pronounced, then there is no attack: if the plosive is voiced, it starts with a vibration of the vocal cords, and if it is voiceless it simply starts with the release. The release is usually the most audible part of the plosive, and it is for this reason that final plosives tend to vanish, especially voiceless ones (their release is suppressed, and then the entire plosive is reduced to very little).

The release might not be complete. If the plosive is the last segment, the release might simply be omitted or inaudible. If the next segment is a nasal consonant articulated in the same place, the release is nasal: the air is allowed to flow through the nose. If the next segment is a lateral consonant articulated in the same place, the release is lateral: the air is allowed to flow around the obstacle. The English words “button” and “bottle” show examples of nasal and lateral release respectively, if the final segment is pronounced syllabically: once the tip of the tongue is in place against the alveolar ridge (when pronouncing the ‘t’ sound), it remains there, and the release of air is made either through the nose or by the sides of the tongue.

Note that if a plosive is lengthened, it is the period between attack and release which is lengthened. In the case of a voiceless plosive, this is merely a period of silence.

See this section for a detailed segment-by-segment discussion of plosives.

Nasals are very much like plosives, except that the air is allowed to escape through the nose: the attack diverts the flow of air through the nose rather than interrupting it. The sound commonly represented by the letter ‘n’ in English is a nasal consonant. A detailed description of nasals is found in this section.

Trills (or “rolled consonants”) are produced by repeatedly interrupting the airstream. The “rolled r” of various languages are trills; but in practice, trills, being rather hard to produce, are often simplified to flaps or approximants. We will say more about trills and how to articulate them further on.

Flaps (or “taps”) are obtained by shortening trills to a single interruption, which therefore comes very close to a “light” plosive (but it is much shorter, and the interruption is usually not complete). Some British speakers will pronounce intervocalic ‘r’ (of such words as “very”) as a flap; Americans, on the other hand, will pronounce some intervocalic ‘t’'s (such as that of “butter”) as a kind of (voiced) flap. More about this later.

Fricatives (also called “spirants” or “constrictives”) are probably the most important kind of consonants, in that it permits the greatest number of easily producible and distinguishable places of articulation. A fricative is obtained by narrowing the airstream so as to make the flow turbulent and produce a kind of hissing sound (frication). A fricative does not usually have a clear attack and release; it can be arbitrarily prolonged. The two consonants in the English word “fuss” are fricatives.

When the airflow is no longer turbulent, fricatives becomes approximants (also called “continuants” or “semivowels” — although the latter term is better reserved for non-lateral approximants). This can be because there is less narrowing, or because the air flows more slowly (fluid dynamics teaches us that the Reynolds number, which determines how turbulent the flow is, is proportional to the speed of the fluid).

Although approximants are classified as consonants, there is no real distinction with vowels (at least for non-lateral approximants between the palatal and pharyngeal regions): any vowel that is sufficiently close or back can be made into an approximant. The letter ‘y’ in the English words “year”, “you” or “yet” is a palatal approximant (on the other hand, in the word “yeast” it comes closer to a voiced fricative); similarly, the letter ‘w’ in “worm” or “wet” is a labio-velar approximant (on the other hand, in some American pronunciations of “where”, the ‘wh’ is realized as a voiced fricative). The letter ‘u’ in the French word “nuit” is a labio-palatal approximant, and the letter ‘w’ in the Dutch word “wind” is a labio-dental approximant (whereas in the corresponding German word it is a voiced labio-dental fricative). Note that the letter ‘r’ found in most American and some British pronunciations of English is also an (alveolar) approximant; but analysis of the English ‘r’ is by no means easy. Finally, the ‘l’ of English and French (the two are not the same) are lateral approximants.

One sequence commonly found in many languages is the succession of a plosive by the corresponding fricative. It then often happens that the release of the plosive merges with the attack of the fricative to form an affricate. In other words, an affricate pair is a plosive with constrictive release. Whether an affricate consists of a single segment or two is a theological question. Examples of affricates found in English are the ‘ch’ sequences in the word “church” (voiceless postalveolar affricates) and the ‘j’ and ‘dg’ sequences in “judge” (the voiced counterpart). Further, an affricate can lead to an approximant instead of a fricative: the ‘dr’ sequence in some English words is of this kind, at least in some pronunciations.

The voiced / voiceless distinction is mainly found in plosives and fricatives (and affricates). Nasals and approximants are voiced (except when whispered), because it is hard to hear them when they are not: a laminar (the opposite of turbulent) flow of air not accompanied by a vibration of the vocal cords is all but inaudible. In fact, voiceless approximants tend very strongly to be replaced by the corresponding fricative: for instance, when the word “wet” is whispered, or when the word “when” is pronounced by those speakers who make a difference between voiced ‘w’ and voiceless ‘wh’, the voiced (labio-velar) approximant is changed to a voiceless (labio-velar) fricative; even the approximant ‘r’ tends to turn into a fricative (not unlike ‘sh’) when whispered, at least in initial and post-consonantal positions. Trills and flaps are, also, generally voiced (although there is no real reason for this).

Places of articulation

Bilabials refer to consonants that are produced using the lips, by narrowing or closing them. For example, the sounds usually represented in English by the letters ‘p’, ‘b’ and ‘m’ are bilabials (the first two are plosives, voiceless and voiced respectively, and the third is a nasal).

Labiodentals are consonants which are formed with the upper teeth against the lower lip. English has two such consonants, [f] and [v], both being fricatives; since it knows no bilabial fricatives and the only labiodentals are fricatives, we can say that English does not distinguish bilabials and labiodentals. There also exists a linguolabial kind of sounds, made with the tongue against the upper lip; but apart from some Austronesian languages, I don't think they ever occur.

Dentals, alveolars and postalveolars are all made with the tongue in the front of the mouth: dentals use the tip of the tongue against the teeth, alveolars are articulated with the tip or blade of the tongue against the alveolar ridge (that is, the ridge of the gums of the upper teeth), and postalveolars are articulated with the tip or blade of the tongue a little behind the ridge. The ‘th’ sounds of English (both the “soft”, i.e. voiced, one, and the “hard” or voiceless one) are dental fricatives. The sounds [s], [z] such as found in English for example are alveolar. The approximant ‘r’, the fricative ‘sh’ (as well as its voiced counterpart) and the affricate ‘ch’, in English, are all postalveolar.

One difficulty of dentals, alveolars and postalveolars is that the tongue, being very mobile in its front, has much freedom of placement, so there are many subtle differences between one sound and another. For example, the [t] and [d] found in French are clearly alveolar, perhaps even dental, whereas those in English are articulated further back, and could be classified as postalveolars. The sound can also be made with the tip or the blade of the tongue: in the formar case it is called apical and in the latter, laminal.

Retroflex consonants are articulated with the tongue's tip curled back against the soft palate. Retroflex sounds are not found in English as such, but some postalveolar consonants tend to have a certain retroflex quality about them, for example the ‘r’.

Lateral consonants are pronounced with the air escaping on the side of the tongue rather than on the front. The lateral quality is not really a “place of articulation” and can be combined with other properties of the consonants: for example, [l] is a (laminal or apical) alveolar lateral approximant (which is further velarized in some cases in English).

Palatals are articulated by bringing together the front of the tongue and the hard palate (i.e. the front part of the palate). The sound [j] found at the start of the English word “yet” is a palatal approximant; the soft ‘ch’ in German (found in such words as “ich”) is a (voiceless) palatal fricative; the sound represented in Italian by ‘gli’ (for example in the word “figlio”) is a palatal lateral approximant; the sound represented in French by ‘gn’ (such as in “mignon”) is a palatal nasal.

Velars are articulated further back than palatals: this time, the back of the tongue approaches the soft palate (i.e. the back of the palate, sometimes called its “veil”, or velum, hence the name). The term “guttural” is also sometimes found. The letters ‘k’, ‘g’ and ‘ng’ in the English words “key”, “get” and “ring” represent respectively a voiceless velar plosive, a voiced velar plosive and a velar nasal.

Uvulars are articulated even further back, with the very back of the tongue coming up to the uvula (the uvula is “that thing which hangs down at the back of our throats”). There are no uvular sounds in English, but the French ‘r’ sound is uvular (generally it is a voiced fricative, but it can be an approximant in certain contexts, and a trill if spoken carefully).

Pharyngeals are produced by constricting the pharynx, which is the region extending from the uvula down to the larynx (where the vocal cords are found). A strictly pharyngeal sound is produced with the root of the tongue, whereas an epiglottal is made with the epiglottis, i.e. that cartilage which prevents food from entering the trachea when one swallows. Finally, glottals or “laryngeals” are made at the level of the glottis. There are no pharyngeals (let alone epiglottals) in English, but there are some glottal sounds: the normal ‘h’ sound is a voiceless glottal fricative, and, although it is not really part of the English phonemic system, one finds some glottal stops (plosives) in certain circumstances in many varieties of English.

Description of segments

Consonants

Plosives

Plosives are probably the easiest to pronounce.

[101] [p]: voiceless bilabial plosive

This is the letter ‘p’ of many languages. There is no particular difficulty about this segment. Nearly every language in the world makes use of this sound (or its voiced counterpart), with the notable exception of Iroquois. The English ‘p’ is slightly aspirated, except when preceded by an ‘s’ in the same syllable.

[102] [b]: voiced bilabial plosive

This is the voiced counterpart of the previous segment. It is the letter ‘b’ of many languages, such as English.

[103] [t]: voiceless alveolar plosive

This is the letter ‘t’ of many languages. However, there is much variability, both between languages and within certain languages, in the articulation of this segment. In English, it is alveolar or even postalveolar, especially when followed by the letter ‘r’. In French, it can be dental, as it is in Hindi (in opposition to retroflex counterparts). Furthermore, the articulation can be apical or laminal.

The English ‘t’, as for the sound [p], is slightly aspirated, except when preceded by an ‘s’ in the same syllable.

[104] [d]: voiced alveolar plosive

This is the letter ‘d’ of many languages. As its voiceless counterpart, it can be articulated in a great variety of ways.

[105] [ʈ] ([t.] in ASCII IPA): voiceless retroflex plosive

This sound is like [t] except that it is retroflex, i.e. articulated with the tongue curled back. It actually sounds very much like [t], and I think the difference mostly comes from the way the following vowel gets “colored” by the retroflex tongue position (rather than from the consonant's articulation itself). Some languages such as Hindi (and other Indic languages) oppose the dental and retroflex ‘t’ and ‘d’.

The IPA symbol for this segment is a ‘t’ with a tail (or hook) on the right. ASCII IPA uses a dot to indicate retroflex quality, so the representation of this character is [t.].

[106] [ɖ] ([d.] in ASCII IPA): voiced retroflex plosive

This is the voiced counterpart of the retroflex ‘t’. Similar comments apply: it sounds very much like a plain [d].

The IPA symbol for this segment is a ‘d’ with a tail (or hook) on the right.

[107] [c]: voiceless palatal plosive

This is one of these mysterious sounds of which nobody knows exactly what they're supposed to sound like. Depending on tongue placement, one can get a sound resembling somewhat [kj] (if it is apical) or [tj] (if it is laminal). In any case, there always seems to be some amount of affrication about this segment (but of course it is really like [kç] or [tç] because the frication is voiceless), but a pure articulation should reduce it to a minimum. However, the attack is pretty much indistinguishable from an alveolar or velar attack (according to the way the tongue is placed), and it is only the slight frication during release that reveals the sound as palatal.

[108] [ɟ] ([J] in ASCII IPA): voiced palatal plosive

Pretty much the same as for its voiceless counterpart applies to this one, mutatis mutandis: it sounds quite like [gj] or [dj] according to tongue placement.

The IPA symbol for this segment is a barred dotless ‘j’. This looks somewhat like a turned ‘f’ (and this printing is sometimes found in old texts), but the stroke is higher than it would be for a turned ‘f’.

[109] [k]: voiceless velar plosive

This is the ‘k’ (or “hard ‘c’”) of many languages. Contrary to its palatal counterpart, it is unambiguously a plosive (though ‘qu’, in English for example, can be a labialized velar affricate).

The English ‘k’, as for the sound [p], is slightly aspirated, except when preceded by an ‘s’ in the same syllable.

[110] [g] (or more properly [ɡ]): voiced velar plosive

This is the ‘g’ (“hard ‘g’”) of many languages.

The IPA symbol is an opentail (i.e. italic-type, or “script”) ‘g’. A looptail ‘g’ is tolerated, if it is the font's normal type of ‘g’, but the opentail variant is strongly preferred. Of course, in ASCII IPA, it is simply ‘g’.

[111] [q]: voiceless uvular plosive

This sounds much like [k] but it is articulated at the very back of the tongue, against the uvula.

[112] [G] (or more properly [ɢ]): voiced uvular plosive

The voiced counter part of [q] sounds very much like [g] but it is articulated at the very back of the tongue, against the uvula.

The IPA symbol is a small capital ‘G’. Naturally, in ASCII IPA, it is simply ‘G’.

[113] [ʔ] ([?] in ASCII IPA): glottal plosive

This “glottal stop” sound is made by abruptly closing the vocal cords (attack) and opening them again (release). As for any plosive, but particularly for this one, the attack may be inexistant if the segment is initial (see below), and the release may be inexistant if the segment is final. The central part (between attack and release) is a period during which the vocal cords are kept closed: for this reason, they are unable to vibrate, so this sound is necessarily voiceless.

To pronounce an initial vowel, one can maintain the vocal cords in a neutral (relaxed) position, perhaps letting an ever-so-slight flow of air go through them, and then get them vibrating with the tongue in correct position: this is the normal “soft” attack. Contrarily, one can also keep the vocal cords firmly shut, and get them to vibrate just as one opens them: this is the “hard” attack, which corresponds to pronouncing a glottal stop before the vowel (note that the attack of the vowel is qualified as hard, and it is the release of the glottal stop; but the attack of the glottal stop itself is inexistant or inaudible, as the vocal cords were already closed to start with).

Hard attacks are not normally found in English, and never in French. German, on the other hand, makes a systematic use of them. In some versions of English (notably British), some consonants (notably the letter ‘t’) tend to be replaced by glottal stops (this is known as “glottal replacement”). Glottal stops are also an integral part of certain interjections, such as “uh-uh” (both parts have hard attack).

The IPA symbol for this segment is reminiscent of a dotless question mark.

Note that there exists a (voiceless) pharyngeal plosive, but it has no symbol. There is also a (voiceless) epiglottal plosive, which will be described later.

Similarly, there exist labiodental plosives, but they have no IPA symbol.

Nasals

[114] [m]: bilabial nasal

This is the normal ‘m’ sound of such languages as English. It is pronounced like the sound [b] but with the palate lowered so that the air can flow freely through the nose.

[115] [ɱ] ([M] in ASCII IPA): labiodental nasal

This sound is somewhat like [m] except that it is pronounced with the lower teeth (rather than the lower lip) against the upper lip.

Plosive counterparts to this nasal are possible, but they are not included in the IPA.

The IPA symbol is an ‘m’ with a left (“elephant trunk”) hook on the right. The ASCII IPA uses a capital ‘M’.

[116] [n]: alveolar nasal

This is the ‘n’ of many languages. The same remarks as for [t] (concerning the variety of articulation) apply to this nasal.

[117] [ɳ] ([n.] in ASCII IPA): retroflex nasal

This is like [n] but pronounced with the tongue curled back: see the discussion on the retroflex plosives.

[118] [ɲ] ([n^] in ASCII IPA): palatal nasal

Just as the corresponding plosive sounds a little like [tj] or [kj], this segment sounds somewhat like [nj]; however, it should not be confused with it.

It is found notably in French, where it is transcribed ‘gn’.

The IPA symbol is an ‘n’ with a left hook on left. No not confuse it with the eng.

[119] [ŋ] ([N] in ASCII IPA): velar nasal

The velar nasal is frequently found in place of the alveolar nasal before a velar plosive; sometimes it replaces both segments in one: so it is often written ‘ng’.

In English, ‘ng’ can be either a plain velar nasal or a velar nasal followed by a (voiced) velar plosive: one should be careful to distinguish the two. Present participles ending in ‘ng’ are pronounced with a plain velar nasal, as is (consequently) the word “singer”, whereas in the words “finger” or “English”, the ‘ng” combination is a velar nasal followed by a velar plosive. Practically, the two are distinguished by the fact that in the second case the flow of air through the nose has been interrupted before the release, and the latter is stronger.

The IPA symbol is important enough to merit a special name: it is called an “eng” (from the sound itself) or “agma” (from the Greek word “ἄγμα” meaning “fragment”). It is an ‘n’ with a left hook at right. The ASCII IPA uses a capital ‘N’ to represent this segment, despite the fact that the IPA itself uses the (small) capital ‘N’ for the uvular version.

[120] [N] (or more properly [ɴ]; [n"] in ASCII IPA): uvular nasal

This sounds quite like [ŋ] but it is articulated at the very back of the tongue, against the uvula, as are the plosive counterparts of this sound.

This is the normal (word-final) pronunciation of the Japanese syllabic nasal, e.g. in such words as “日本” (“にほん” = “nihon”): [ɲihoɴ].

Note that it is not possible to pronounce a pharyngeal, epiglottal or glottal nasal, because the communication between the throat and the nose is at the back of the soft palate, and pharyngeals &al are articulated even further back in the throat (so that blocking the flow of air there will block it even for the nose).

Trills

Trills take some practice before they can be correctly pronounced. The native English speaker may have particular difficulties on these, because none are found in this language. Once you know how to articulate one trill, the others should be much easier.

[121] [B] (or more properly [ʙ]; [b<trl>] in ASCII IPA): bilabial trill

This sound is essentially that of the interjection (or should I rather say, onomatopeia) “brrr”, used (though perhaps not in English?) to signify, in essence, “I'm cold”. Essentially, keep your lips shut (relatively tightly), and blow hard enough through them so as to make them flutter. Now, do the same while humming slightly (to make the sound voiced): this is the segment we are discussing. (At least I hope it is… I must admit I have never heard it spoken — the sound is used, as far as I know, only in some Austronesian languages — but it certainly is a voiced bilabial trill, and I don't think there can be much variety as to that.)

[122] [r]: alveolar trill

This is the “rolled ‘r’” of such languages as Russian (yet even in Russian there is a tendency for this segment to be replaced by a simple flap). Spanish maintains a difference between the full trill (‘rr’) and the simple flap (‘r’).

It is not evident to describe how to articulate this segment: pronouncing a “rolled ‘r’” can take some practice, and cheap imitations should be shunned :-). I can nevertheless try offering some advice. First, you should not try pronouncing anything that sounds like ‘sh’ (a fricative) or an American ‘r’ (an approximant). Second, although we classify this trill as alveolar, I find it rather postalveolar; in other words, keep your tongue about where you pronounce ‘sh’ (as opposed to ‘s’). Put the tip of your tongue in that place, against the palate, and build up a very slight excess of pressure behind it (i.e. above the tongue), so that the tongue will be forced away from the palate for a brief moment before coming back in place, thus producing a short ‘t’ sound (do not try to produce a ‘t’ sound as such because if you do, your tongue will not come back fast enough: no muscular action will be fast enough to produce a trill). This is a simple flap. Now try to keep the flutter going, by maintaining the air flowing. Another possibly helpful idea is to keep one (side) edge of the tongue fixed against the palate while the other edge vibrates: strictly speaking, this is a lateral trill, but nobody can tell the difference, and in this position a lateral trill is easier to articulate than a frontal one.

[123] [R] (or more properly [ʀ]; [r"] in ASCII IPA): uvular trill

In theory, this is the ‘r’ of French, but in practice, it is nearly always degraded to a fricative, so that even native French speakers (except perhaps those with a markedly Parisian pronunciation) might find it difficult to produce a completely satisfactory uvular trill. For those who know neither French nor German, the first step is probably to get the fricative in question correct.

It is the uvula which should be vibrating on the tongue, rather than the tongue on the uvula: it is probably for this reason that this trill sounds more like a fricative than the two previous ones. If the fricative is pronounced strongly enough, as a kind of “growling” sound, some flutter is inevitable. Now the trick is to capture that flutter with no (or little) accompanying frication: the trill actually tends to be quiter than the fricative.

To finish on the subject of trills, I note that a palatal trill exists and is, in fact, relatively easy to pronounce. A retroflex trill is not too hard either (but I don't think it can ever be nearly as completely retroflex as the corresponding plosives). A velar trill is, apparently, not possible, because the tongue does not have enough freedom to vibrate at this point, and neither can the soft palate. A glottal trill is out of question. The Phonetic Association claims that labiodentals and pharyngeal trills are possible (though the Alphabet does not have symbols for them): for my part I was unable to produce such things.

Flaps

[124] [ɾ] ([*] in ASCII IPA): alveolar flap

This is the very shortest form of the alveolar trill, in other words the “short rolled ‘r’”. It is found in English in essentially two circumstances: in some British pronunciations of intervocalic ‘r’, e.g. in “very” ([vɛɾɪ]), and in the general American pronunciation of the letter ‘t’ in certain contexts, the most striking example being perhaps the word “butter” (pronounced as [bʌɾɚ]).

There is actually much variability in this segment; we have already observed this of the associated plosive.

The IPA symbol is a “fish-hook” ‘r’”. In ASCII IPA, it is an asterisk: [*].

[125] [ɽ] ([*.] in ASCII IPA): retroflex flap

This is the retroflex variant of the previous segment; see the retroflex plosives for general comments on retroflex sounds. As I pointed out on the subject of trills, however, your tongue can never be as completely curled back as it would be for the plosive.

This is usually used to transcribe the ‘r’ of Japanese (which they use to transliterate both European ‘r’ and ‘l’); but in fact, Japanese ‘r’ is more like a postalveolar than like a true retroflex.

The IPA symbol for this segment is an ‘r’ with a retroflex hook. ASCII IPA uses [*.]

Fricatives

Fricatives are in principle the easiest sounds to produce. However, they also have the greatest variability, so some of them are actually quite hard for speakers of most European languages to produce.

[126] [ɸ] ([F] in ASCII IPA): voiceless bilabial fricative

This is the sound made by blowing on something: you tighten your lips (with or without rounding) so that the flow of air will become more rapid, and hence more turbulent. It sounds rather like an ‘f’, but made without the teeth.

The IPA symbol is a phi. If we believe Unicode, it is a Latin small phi, whatever that means. ASCII IPA uses a capital F instead.

[127] [β] ([V] in ASCII IPA): voiced bilabial fricative

This is the voiced counterpart of the previous segment. It sounds much like a ‘v’ but made without the teeth. It has the tendency (like many other voiced fricatives, but this one particularly) to decay into an approximant.

The IPA symbol is a beta. ASCII IPA uses a capital V instead.

[128] [f]: voiceless labiodental fricative

This is the ordinary ‘f’ of many languages, such as English. It is made by lightly pressing the upper teeth against the lower lip, and blowing. Note that as English (for example) does not have the bilabial ‘f’, this sound is perceived as the fricative counterpart to the bilabial plosive.

[129] [v]: voiced labiodental fricative

This is the ordinary ‘v’ of many languages, such as English (but not Spanish). It is the voiced counterpart of the ‘f’.

[130] [θ] ([T] in ASCII IPA): voiceless dental fricative

This is the “hard” ‘th’ of English, found in the word “thin” for example. It is pronounced by putting the upper teeth against the tip (or blade) of the tongue, and blowing through. For this reason, it is somewhat similar to the sound [f], and some people confuse the two.

The IPA symbol for this sound is a Greek theta. ASCII IPA uses a capital T.

[131] [ð] ([D] in ASCII IPA): voiced dental fricative

This is the “soft” ‘th’ (symbolically, ‘dh’) of English, the voiced counterpart of the previous segment.

The IPA symbol is a Scandinavian “eth” (sometimes written “edh”): graphically it is like a rounded (cursive) ‘d’ with a bar through the top. The ASCII IPA uses a capital D to represent it.

[132] [s]: voiceless alveolar fricative

This is the ordinary ‘s’ of many languages, such as English. Compared to the dental sound, the tongue is further back, and further raised: the blade of the tongue and the alveolar ridge form a narrow passage for the flow of air, in which turbulence is heard.

[133] [z]: voiced alveolar fricative

This is the ordinary ‘z’ of many languages, such as English. It is the voiced analog of the previous sound.

[134] [ʃ] ([S] in ASCII IPA): voiceless postalveolar fricative

This sound is usually represented by ‘sh’ in English. Compared to [s], it is made with the tongue further back, and using the tip instead of the blade, curling it slightly to produce a little channel through which the air can flow.

The IPA symbol is an “esh”, that is, an elongated ‘s’, not unlike an integral sign in mathematics. ASCII IPA transcribes by a capital S.

[135] [ʒ] ([Z] in ASCII IPA): voiced postalveolar fricative

This sound is usually represented by ‘zh’ in transcription. It is not inexistent in English (it is, for example, the ‘z’ in “azure”) but it is mildly uncommon. In French (and Catalan) it is represented by the letter ‘j’. But most languages have this sound as part of the corresponding affricate (e.g. ‘j’ and ‘dg’ in English “judge”).

The IPA symbol is a tailed z, also called “ezh” or “yogh”. ASCII IPA uses a capital Z.

[136] [ʂ] ([s.] in ASCII IPA): voiceless retroflex fricative

This sound is like an ‘s’ but articulated with the tongue curled back (i.e. retroflex). In fact, it sounds very much like the postalveolar. See also the discussion on retroflex plosives.

The IPA symbol for this segment is a ‘s’ with a tail (or hook) to the right on the end. ASCII IPA uses a dot to indicate retroflex quality, so the representation of this character is [s.].

[137] [ʐ] ([z.] in ASCII IPA): voiced retroflex fricative

This is the voiced analog of the previous. It sounds very much like the postalveolar.

The IPA symbol for this segment is a ‘z’ with a tail (or hook) on the right.

[138] [ç] ([C] in ASCII IPA): voiceless palatal fricative

This is the soft ‘ch’ of German. To pronounce it, put your tongue in the position to pronounce [j], and blow. It is a kind of hissing sound, which could quite be used as an onomatopeia to illustrate a serpent (although of course [s] is also appropriate). Some people have trouble distinguishing this sound (‘ch’ in German) from the postalveolar (‘sch’ in German): in fact, although they sound somewhat similar, this one is much closer to an [s], in the way the tongue is placed, except that it is further back. The tongue does not touch the palate, it is parallel to it.

The IPA symbol is a ‘c’ with cedilla. ASCII IPA uses a capital C instead.

[139] [ʝ] ([C<vcd>] in ASCII IPA): voiced palatal fricative

This is the voiced counterpart of the previous segment. It is often used instead of the approximant [j], for example before an [i] (consider the English word “yeast”) or when whispering (in which case it is actually the voiceless version which is pronounced).

The IPA symbol is a ‘j’ with curly tail. ASCII IPA represents it as the voiceless counterpart of the previous.

[140] [x]: voiceless velar fricative

Being the fricative counterpart to the sound [k], this sound would be naturally transcribed as ‘kh’; it is the hard ‘ch’ of German; however, there are subtleties. Essentially, you pronounce it by putting your tongue in the position to say ‘k’, and then force the air through it (i.e. between the palate and the tongue), thus making a kind of rasping sound.

The difficulty is that there are various kinds of “rasping” sounds. And one tends quite naturally to pronounce a uvular fricative instead of this one; or at any rate to not distinguish the two: German, for example, will use [x] as “hard ‘ch’” after [u] and [o] (“Buch” and “hoch”), but the uvular sound is found elsewhere. The best way to make sure one is pronouncing a velar sound (rather than uvular) is to see whether one can make a trill out of it: if one succeeds in getting something like the French ‘R’, the sound was uvular (it is not possible to produce a velar trill).

[141] [ɣ] ([Q] in ASCII IPA): voiced velar fricative

This is the voiced counterpart of the previous segment. It is found, for example, in modern Greek, represented by the letter ‘γ’ (gamma).

The IPA symbol is a lowercase Latin letter gamma (whatever that is!), which should not be confused with the “ram's horn” symbol used for the close-mid back unrounded vowel (number 315) — and in practice it systematically does get confused except in very carefully typeset texts. ASCII IPA represents the sound as a capital ‘Q’.

[142] [χ] ([X] in ASCII IPA): voiceless uvular fricative

As we have mentioned, this is a kind of rasping sound, differing from the previous one in that it is uvular rather than velar. See also the voiced counterpart.

The IPA symbol for this segment is a (lowercase) Greek chi; ASCII IPA uses a capital ‘X’ instead.

[143] [ʁ] ([Q"] in ASCII IPA): voiced uvular fricative

This is the voiced counterpart of the previous segment. It is the normal ‘R’ of French, when not rolled (i.e. when not pronounced as a trill).

The IPA symbol is an inverted small Latin capital ‘R’. Evan Kirshenbaum proposes to represents the sound in ASCII IPA as a ‘g"’, but I find it more logical to use ‘Q"’.

[144] [ħ] ([H] in ASCII IPA): voiceless pharyngeal fricative

This sound is found in the so-called “Oriental” pronunciation of Hebrew, where it corresponds to the letter ‘ח’ (“het”, or, more accurately, “ḥêt”), transliterated ‘ḥ’ (‘h’ with dot below). (Non-Oriental Hebrew substitutes a uvular instead.) It also exists in (Classical) Arabic, corresponding to the letter ‘ح’ (“hah”, or, more accurately, “ḥā'”), transliterated in the same way.

It is somewhat difficult to pronounce for one who does not know one of the languages in question. It is a kind of ‘h’ but pronounced with a kind of “hoarse” voice: that is, with the root of the tongue pulled back so as to constrict the pharynx. (I would tend to say: if pronouncing this sound makes you want to throw up, or at least to cough, your tongue is probably in the right place. :-) The main problem is to not pronounce a velar or uvular sound; on the other hand, a normal (glottal) English [h] (even pronounced loudly) will not do either. Look at your throat in a mirror: your uvula should be clearly visible when you pronounce this sound (it should not be touching the tongue).

The IPA symbol for this segment is a (lowercase) Latin ‘h’ with a stroke; ASCII IPA uses a capital ‘H’ instead.

[145] [ʕ] ([H<vcd>] in ASCII IPA): voiced pharyngeal fricative

This sound is the voiced counterpart of the previous one, and it is even trickier. It is supposedly found in the “Oriental” pronunciation of Hebrew, where it corresponds to the letter ‘ע’ (“ayin”, or, more accurately, “aīn”), transliterated ‘`’ (reverse apostrophe). But in fact, Hebrew speakers pronounce an approximant rather than a true fricative. (Non-Oriental Hebrew substitutes a glottal stop instead.) It also exists in (Classical) Arabic, corresponding to the letter ‘ع’ (“ain”, or, more accurately, “`ayn”), transliterated in the same way. But there again, it is not truly the canonical sound which is used, rather some kind of pharyngealized glottal stop.

The reason why this sound is so difficult to pronounce as it should be is the following. To pronounce a voiced sound, the vocal cords must vibrate, which limits the intensity of the air flow. But then, to produce the necessary turbulence (“frication”), the vocal tract must be very strongly closed. However, it is almost impossible to constrict the pharynx enough for that (except by swallowing one's tongue, which is deadly). If you do it right, you get a kind of very hoarse ‘R’ sound. Otherwise, you get an approximant instead, which is what is generally meant by this segment, anyway.

The IPA symbol for this segment is a reversed glottal stop. ASCII IPA represents it as the voiceless counterpart of the previous.

[146] [h]: voiceless glottal fricative

This is the ‘h’ sound of English, German and other languages (in French, however, for example, the ‘h’ is mute and is never pronounced at all). It is the simplest sound of all: just breathe out through your mouth (without letting your vocal cords vibrate). In other words, it is a pure (voiceless) aspiration.

Although it is classified as a fricative, it would make just as much sense to consider it as an approximant (but a voiceless one, which is strange), because you don't really constrict your vocal cords beyond the normal amount of frication that goes on in the laryngeal region anyway.

[147] [ɦ] ([h<vcd>] in ASCII IPA): voiced glottal fricative

This sound is the voiced counterpart of the previous one. It is also a kind of “nothing sound”: this time you blow out with the vocal cords vibrating; or, if you will, you do a kind of humming sound. Only a great amount of hypocrisy can get this sound classified as a fricative (rather than an approximant): it physically doesn't make sense to produce at once a regular vibration (humming) and turbulence (frication) at the same level (the vocal cords). In fact, even classifying this sound as a consonant is an act of faith (or a bizarre way to split hair), because what you're really doing is pronouncing a vowel and wishfully think it's a consonant.

Yet through the kind of bizarre, unexplainable magic that only phonetics has, this sound still exists in a very definite way. For example, it is found in the most common way of pronouncing the English word “inherent” (corresponding to the letter ‘h’). Also, Hindi (and other Indic languages) has a series of voiced aspirated stops, which are clearly distinguished from the unaspirated by the presence of this elusive sound. The most probable solution to the mystery is that the segment probably isn't as distinctly voiced as it claims to be, thus allowing a flow of air greater than what would characterize a vowel.

The IPA symbol for this segment is a ‘h’ with hook on top (reminiscent of the symbol for the pharyngeal sound). ASCII IPA represents it as the voiceless counterpart of the previous.


David Madore

Last modified: $Date: 2002/07/29 20:09:07 $