4.2.1 Consonants

For consonants, we have a three-part technical description that addresses three questions, having to do with voicing, place of articulation, and manner of articulation. The first question addresses the issue of voicing, whether or not the vocal folds vibrate during articulation. Consonants can be described as either voiced (when the vocal folds vibrate) or voiceless (articulated without the vibration of the vocal folds). For a quick example of the difference between a voiced and voiceless consonant, you can channel your inner kid and make some animal noises: Place your hand on your throat and hiss like a snake and then buzz like a bee. In theory, you should be able to feel the vocal folds vibrate with the bee sound <zzzzzzzz>, which is voiced, but not with the snake sound <ssssssss>, which is voiceless. Most English consonants come in voiced–voiceless pairs, where the main difference between the two sounds is that one is voiced and one is voiceless. The sounds represented by the IPA symbols [s] and [z] are one of those pairs.

The two remaining questions we have to answer in order to describe consonant sounds have to do with the constriction that occurs along the articulatory path; place of articulation tells us where the constriction takes place, while manner of articulation tells us how much constriction is there. Sounds that are created with the constriction at the front of the mouth, like the initial sounds of the words ball and pop are bilabial sounds (bilabial meaning “two lips”), while sounds created with the top teeth and bottom lip, such as the first sounds in fox and van, are labiodentals. Next, we have interdentals, which are sounds made with the constriction occurring with the tongue behind or between the teeth, as in thank and the. Moving back in the mouth a bit further, we have alveolar sounds, which are articulated with the constriction at the alveolar ridge, the bony area behind the top teeth. There are a number of alveolar sounds in English, including the initial sounds in the words dog and top. Past the alveolar ridge are the post alveolar and palatal areas, points of constriction for the initial sounds of words such as cheese and jar. Back in the mouth further still is the velum, which raises and lowers as needed to block the opening of the nasal cavity. A few English sounds are articulated with constriction in this area; these sounds, called velar sounds, can be found in words such as kite and goat. At the back of the throat is the glottis; sounds articulated with constriction in this area are glottal sounds, such as the initial sound in the word horn.

In terms of how much constriction is taking place, we have a few options for characterizing the manner of articulation for consonants. There are some sounds that are produced with complete constriction at their point of articulation. These are called stops (or plosives), examples of which are at the beginning of the words pop, top, and goat. Other sounds are created by a good bit of constriction, when the tongue is very close to the teeth or alveolar ridge, for example, and these sounds are called fricatives, due to the “friction” that is created. Examples of fricatives are sounds such as those at the beginning of fox, thank, soda, and shop. Still other sounds are created when a stop and a fricative occur in close combination, and these are called affricates, including the first sounds in the words cheese and jar. For nasal consonants, there is a stoppage of the airflow through the mouth, but because the velum is lowered, the air is allowed to resonate in the nasal cavity, producing nasal sounds, such as those found at the beginning of money and note. Finally, there are a handful of consonants that are articulated with very little constriction, so little that they are referred to as semi-vowels (or approximants). These sounds can be liquids, such as the first sound in leg, or glides, as with the first sound in the word yarn. Table 4.1 is a chart containing all of the IPA symbols for English consonants, their technical descriptions, and some example words.

4.2.2 Vowels

In order to provide a technical description of a vowel, we have to take a different route. All vowels are voiced, so the issue of voicing does not apply and, because constriction isn’t an issue, we don’t talk about “manner” or “place” of vowel articulation. Instead, we answer four questions to arrive at the technical description of vowel sounds. The first two questions have to do with how we arch the tongue inside the mouth to create the different shapes that then produce the different vowel sounds: tongue height and tongue position. The tongue can arch close to the roof of the mouth, creating high vowels; it can be kept closer to the bottom of the mouth, creating low vowels, or in the middle for mid vowels. Thus, the terms “high,” “low,” and “mid” indicate an answer to the question of tongue height. In order to address tongue position, we look at whether the arch of the tongue is more toward the front (near the teeth), center, or back of the mouth, and we use the terms front vowel, central vowel, and back vowel to describe this. The third question deals with tongue tension, whether the sound is produced as a tense vowel or a lax vowel. Finally, we answer a question about lip roundedness to determine if the vowel sound is rounded or unrounded. Table 4.2 contains a chart that is laid out in such a way as to cover all four of the vowel characteristics along with example words for each vowel sound.

Table 4.2 IPA vowel chart

Note: the shaded symbols represent tense vowels (non-shaded vowels are lax) and the four bracketed back vowels are rounded (all other vowels are unrounded).

Because vowels are generally described using measures of tongue position (frontness/backness) and tongue height (high/mid/low), we can plot vowels two-dimensionally in what we refer to as vowel space. Figure 4.1 is a visualization of vowel space (note that this is simply a different way to present much of the same information found in Table 4.2).

Figure 4.1 Visualization of vowel space.

Individual speakers’ articulation of vowels and consonants can vary. In order to capture minor variations in articulation of speech sounds, linguists can use diacritics, which are small marks added to IPA symbols, in order to describe more specifically an individual speaker’s pronunciation. Diacritics for vowels can indicate how front or back (or high or low) the tongue is during articulation or whether the vowel is more/less rounded. Figure 4.2 contains some examples of diacritics frequently used to give a more narrow (vs. a broad, or more general) transcription of vowel sounds. Full IPA charts (with diacritics) can be found at www.internationalphoneticassociation.org (including an interactive chart that lets you listen to the sounds represented by each IPA symbol).

Figure 4.2 Diacritics used for narrow transcription of vowel sounds.

4.2.3 Wells’ Standard Lexical Set

In order to facilitate comparisons between varieties of English, John C. Wells came up with a list of keywords intended to highlight differing pronunciations of the same vowel sounds across different English varieties. The keyword system is a complement to the IPA; whereas in the IPA there is a one-to-one correspondence between symbol and meaning, so that it can be used for all languages (and therefore all varieties of a language), the keyword system indicates what IPA vowel symbol would be used in the same word across English varieties. Wells’ (Reference Wells1982) standard lexical set can be found in Table 4.3; each keyword is accompanied by its realization in Received Pronunciation (RP) and North American (NA), in order to offer a comparison between the so-called “Standards” of British and American Englishes.

Note that many of the differences between RP and NA have to do with vowel length, while others represent different articulations of the same vowel, such as BATH, which sounds like [bɑθ] in RP British English, but [bæθ] in American English. Wells’ [e] sound for DRESS is really the same as the American [ɛ] sound, but Wells uses a different symbol. The same thing can occur for STRUT, where many Americans would represent the sound as [ə]. British and American English also typically differ in rhoticity, the pronunciation of the <r> after a vowel, and Table 4.3 shows American English pronunciations with an [r] (IPA includes several different pronunciations for the <r> sound, of which [r] is just one).

4.2.4 Phonemes and Allophones

Besides the one-to-one correspondence between a speech sound and an IPA symbol, each IPA symbol can also represent a phoneme, a collection of phonetic segments that we perceive and distinguish as being the same sound in a particular language. When IPA symbols are used to represent phonemes they are enclosed in slant lines (also called virgules), so that /b/ is the phoneme for the <b> sound, and [b] represents the speech sound for a <b>. The important thing to note here is that, not only do different languages employ different subsets of speech sounds, speakers of different languages may have different perceptions of which sounds belong to the same phoneme. Different languages make different distinctions. For example, in English voicing is considered phonemic, because trading a voiced consonant for its voiceless counterpart would make a difference in meaning, for instance turning [bæt] into [pæt]. In Korean, however, voicing is not phonemic, which means that Korean speakers would perceive [bæt] and [pæt] as the same word. Likewise, in Korean, aspiration is phonemic. Aspiration occurs when there is a puff of air released in the articulation of a stop consonant, and is indicated in IPA by the addition of a small diacritic, in this case a small superscript ^h, to a transcription. What this means is that, in Korean, [b^hæt] and [bæt] are perceived as different words, while English speakers, if they perceived a difference at all, would simply hear them as different pronunciations of the same word. In English, the aspirated and unaspirated versions of [b] are allophones of the phoneme /b/. An allophone is a spoken variation of a phoneme. Allophones of the same phoneme are perceived as being the same sound, and they are not distinctive in the same way that phonemes are: They don’t make a difference in meaning. Take, for example, the different articulation of the middle sound in words such as letter and butter; in American English [ɾ] (called a flap) and [d] are allophones of /t/ in this environment, i.e. between vowels. In some varieties of British English [ʕ] (a glottal stop) is an allophone of /t/ in those same environments but [ɾ] and [d] are generally not. Each of these sounds can be an allophone of the phoneme /t/ in the different English varieties, and whether you say [bʌtər] or [bʌɾər] or [bʌdər] or [bʌʕər], the meaning is the same.

Allophones for a phoneme in any variety of a language follow the same A-curve frequency profile that you learned about in Chapter 3. In any environment for use of a phoneme – say the /t/ between vowels in a word like butter – the possible allophones will be used more or less often, one or a few very often and most of them rarely. If you happen to be in England, you will hear the [t] most often in butter, sometimes the [ʕ], and more rarely the [d] or [ɾ]; in the United States it is the other way around, so you will hear [bʌɾər] or [bʌdər] most often, and less often the [t] or [ʕ]. Complexity science tells us that the variation in allophones is actually much greater than we generally think it is, and survey data from the LAP confirm it.

How do we figure out which sounds are (or are not) considered phonemes in a particular language? We look to minimal pairs (or minimal sets), which are pairs or sets of words that differ by only one sound, a sound that then makes a difference in meaning. In English, pat [pæt] and bat [bæt] constitute a minimal pair because of the single distinctive difference. The contrasting sound must be in the same place in each word: It can be an initial sound (as in the pat, bat example), a sound in the middle (pat, pit), or a sound at the end (pat, pad). Once distinctive sounds have been identified, we use slant lines /b, p/ to indicate that we are talking about these sounds as phonemes for a given language or variety. We stick with brackets to talk about the phonetic segments found in speech.

In addition to making statements about what is or is not a phonemic difference between languages, minimal pairs can also give us a way to talk about differences between dialects or varieties of a single language. For example, within American English some speakers find the sounds [ɑ, ɔ] to be distinctive, and some speakers don’t; so for some American English speakers the words Don and dawn, or cot and caught, sound exactly alike (which means the vowel sounds are not distinctive), while other American English speakers consider each of these to be a minimal pair. What this means is that there are at least two different sound systems within American English, just based on this one distinction. British English also has multiple sound systems, as many speakers of British English do not pronounce the [r] after vowels in words like bar and barn, while others (usually from the North or West) do pronounce it. The possibility that a large variety like British English or American English can have more than one sound system is exactly what we predict from complex systems: It makes sense to talk about big varieties, but at the same time we predict that the systems at lower levels of scale can be somewhat different.

Despite internal variations, we can still talk about “American English” and “British English”; we simply need to remember that treating any language as a single, unified system is a generalization. In fact, we are making a generalization when we talk about the characteristics of a dialect of a language as well, as it is the case that even people who are considered to be speakers of the same dialect may not use vowels and consonants in the same way. Even individual speakers are not entirely consistent in how they pronounce their vowels and consonants: Just as for large groups and sub-groups, individuals have A-curve frequency profiles for the units of their pronunciation. Basically, any claim we make for a sound system, whether it’s about an individual speaker or a group of people, will be a generalization. It is fine to make generalizations about sound systems if we need to do so for particular purposes, but it is important to remember that the boundaries and categories we use are of our own making. There are no “natural classes.”

The IPA offers those who know it an effective and reliable way to talk about the sounds of a particular language and about the pronunciation of specific words, not to mention the fact that having an agreed-on set of categories for speech sounds gives linguists something to count. In reality, though, the IPA does not represent any sort of linguistic “truth.” Instead it functions as an international agreement on how we talk about what are actually arbitrary segments of continuous physiological movements. In the context of natural speech, individual sounds are indistinct; even individual words taken out of a speech stream context are difficult to identify.