The forms adopted for use in the IPA are purposefully designed to follow the Latin alphabet, with secondary recourse to the Greek alphabet. Sometimes an IPA character will be a modification of a letter from these two main sources. Finally, a fourth category of invented characters was added. For example, the "gelded question mark" (the punctuation mark minus the ball) was adopted for the glottal stop (like the sound made in the back of the throat in between the syllables of an American's pronunciation of "uh-oh"). In reverse silhouette this character is designated as the pharyngeal approximant.(like the "ch" in the Scottish word "loch"). It is said to have come to be this representation because of its similarities to the letter "ain" in Arabic.
Several of the IPA symbols did not derive from human letters or punctuation, and were simply accepted because they reflected the common notational usage employed by linguists at the time.
The easiest forms for speakers of English are the "simple" consonants: , , , (hard), (aspirated), (not aspirated), (ell), , , (not aspirated), (not voiced), (not aspirated), , and . They are the same in the IPA and in English.
The "simple" vowels of ,,,, are pronounced in the traditional manner of Latin, that is as in father, as the "a" in savor, as in machine, as in so, and as in rule.
Other letters -- like , and - do not follow the English sounds; rather, they follow the conventions of other languages spoken in Europe.
To this one must then add sounds not represented by "simple" consonants or vowels, like "sh" or "dgee" (soft "g"). So some capitals and cursives were adopted to distinguish shades of sound.
Overlaid on the letters are the diacritics. Rotation was also used as a means of distinguishing sound. The most well-known in English is -- -- the upside-down "e" or "schwa" (pronounced like the "a" in casino.) A rotated "a" -- -- is used for the central open vowel "ah" like the "o" in "pot." A rotated "r" -- -- is the hard, Midwestern "r" of American English, which can sound like a Hollywood pirate's "Arrr" in exaggerated form.
As sounds are modified, diacritical marks indicate the shift. A little hook on the bottom of a consonant, facing to the right, may indicate retroflexion, which is the pronunciation of the consonant with the tongue placed towards the roof of the mouth (corona), behind the inside of the gums of the upper teeth (alveolar ridge). English does not use this sound. It would be the full and round sound made for the "or" by a person with a Swedish accent who pronounces the word "north." A leftward-facing hook indicates uvular consonants, like the simple (not trilled) Gallic "r" or the "ng" sound in the Japanese pronunciation of Nihon.
The IPA diacriticals do not necessarily correspond to those employed in a specific language. For example, the closed "Ã´" in Portuguese can be made to sound more open, like the "aw" is "saw" (not prolonged) by adding an acute accent over it ("Ã³"). In the IPA, the near-front, close-mid "o" is Ã¸, and the near-back, close-mid "o" is ɣ (the small ram's horn or lower case gamma).
An example of the application of an IPA diacritical mark is the unreleased stop indicator, which is an apostrophe, as in
Because of the comprehensive nature of the IPA, a word or sound can be represented in many different ways, with different levels of precision. Speech therapists, for example, may want much more detail and resolution for a sound than would a lexicographer. In the world of distinctions between concept and sound, they want the real sound, not the speaker's idea of what the sound is. They want the phone and not the phoneme. This is called "narrow transcription" and it should be reported between square brackets.
On the other end of the spectrum, dictionaries have no concern for how someone actually pronounces a word. They only want to describe how it is supposed to be pronounced in the concept of some sort of Platonic form - the phoneme. They will be able to describe the phoneme (between slashes) with fewer symbols and diacritical marks from the IPA than those engaged in "narrow transcription." This is "broad transcription." Dictionaries and text books, for example, will try to avoid the nuances of allophones as they appear in regional accents and slang terms of a language. The purpose of these texts is just to report what the phonemes are like. This makes life much simpler for someone not profoundly familiar with the IPA.
Another example of the difference between narrow and broad transcriptions is "little." The dictionary should report the phoneme as /lItl/. Suppose it was important to record how this word was actually pronounced in a cockney accent. The focus is on the phone, the sound, and not the phoneme or the model form. In cockney, there would be no "t" behind the teeth, but rather a glottal stop (like the middle of the word "oh-oh"). It would then look like this: [ɪʔɫ] with the glottal stop and the tilde stroke on the last letter, indicating that it is pronounced in the back of the mouth, near the velum or pharynx.
Good linguists, who are picky, perfectionist professionals, are always trying to discover situations not covered by the IPA, and several have been identified. But for professional interpreters, translators and other wordsmiths, the IPA serves them fully and well. The only more precise method of reproducing the sound of speech would be to make an audio recording.