reading 2

how can we describe differences in accents?

There are two possible ways to set about describing differences between accents.

• We can compare their different historical developments, or we can compare their different synchronic states.

• Historically, we can describe the various changes which have occurred since, say, Middle English, or in general since the period when the ancestors of the accents in question first diverged.

• We can list for each accent the rules which have been added to (or lost from) the grammar, together with possible changes in rule order, rule inversions, and so on.

• We can tabulate the splits and mergers that have taken place, and analyse the restructurings which followed as rules ceased to be productive.

• In the alternative, synchronic, approach, we attempt to describe the existing accents as they are.

• We investigate the differences in phonetic detail.

• We examine possible differences in phonological structure (phonotactic distribution).

• We ask whether the phonemic systems of the various accents are isomorphic (i.e. whether there is a one-to-one relationship between the phonemes of accent A and those of accent B).

• We look for differences in the use of particular phonemes in particular words or morphemes.

What are the three kinds of phonic differences between dialects according to Trubetzkoy (1931)?

Phonic differences between two dialects may be of three kinds: they may concern

• the phonological system, or

• the phonetic realization of the various phonemes,

• or the etymological distribution of the phonemes in words.

Accordingly we shall speak of phonological, phonetic, and etymological differences between dialects.

What is the simplest kind of difference between accents, and what examples illustrate this regarding vowels?

• The simplest kind of difference between accents is a difference in phonetic detail - in the phonetic realization of a given phoneme.

• For example, the vowel in words such as coat, nose, snow (referred to in this work as GOAT words) shows considerable variation.

• It is monophthongal in some accents, but diphthongal in others; as a diphthong it may be narrow or wide, with a starting-point which is front, central, or back.

• Thus in RP it is generally [əu], but may also be more lip-rounded, [ou], or less so, [ʌʊ]; there are also old-fashioned back variants, [ou] etc., and variants with a front starting-point, [eʊ]; some variants have very little diphthongal movement.

• In the south-east of England wider diphthongs are common, thus [ʌʊ] etc. (Cockney)

• In Scotland and various other places a monophthongal realization is usual, [o:] or a shorter [o].

• Similarly, the vowel in words such as out, loud, now (the MOUTH words) shows a wide range of variation.

• Although usually diphthongal, it can also be monophthongal, as in the Cockney variants [æ:] and [a:].

• As a diphthong, it can have a starting-point ranging from back [ɒʊ] (South African) to front [ɛʊ] (provincial south of England), as well as the more widespread intermediate types [aʊ], [aʊ].

• There are also variants with less than open central starting-points, [əʊ] and so on.

What examples illustrate realizational differences involving consonants and segment duration?

• For example, there are accents in the north of England and in Scotland where /p, t, k/ are never aspirated, so that pin, tin, kin have [p, t, k], as against the [ph, th, kh] of other accents.

  • Indeed, accents that do have aspiration of voiceless plosives may differ in just how many milliseconds of aspiration (on average) they have in a given phonetic context.

• This brings us to the observation that phonetic details may vary in a context-sensitive manner; that is, one of the points of difference between two accents may be a matter of a particular positional allophone.

• The Canadian [aɪ] allophone of /aɪ/ ... occurs only in the environment of a following phonologically voiceless consonant, as in nice, write, lifelike.

  • In all other environments, as ride, tie, line, fire, the realization of /aɪ/ in this accent is much the same as in most other North American accents.

• In Irish English the consonant /l/ is generally clear in all environments.

  • In RP, GenAm, and many other accents two perceptibly different allophones may be distinguished, clear and dark.

  • Thus intervocalically, as in silly, RP uses clear /l/, thus ['sɪlɪ], while GenAm uses dark, thus ['sɪlɪ]

  • In this word, as in others where /l/ is intervocalic (valley, yellow, column), RP thus sides with Irish English against GenAm; compare words such as belt, milk, halt, where RP and GenAm agree in having dark /l/, but Irish English uses a clear variety.

  • Most accents use a post-alveolar or retroflex approximant for intervocalic /r/ in words such as very, sorry, arrow.

    • A certain kind of RP has free variation here between the approximant and an alveolar tap, [ɾ]; for some speakers the variation may be stylistic.

    • In the working-class accent of Liverpool, a tap in this phonetic environment is the norm, with [ɾ] thus being a positional allophone of the phoneme otherwise realized in the accent as an approximant.

• Accents clearly vary in the details of segment duration.

In the intervocalic /d/ of ready, for example, we face a range of durational possibilities extending from the very short tap of a typical American accent to the rather long plosive of a typical Welsh accent, with other accents occupying intermediate positions.

How may accents differ regarding the environments in which particular phonemes occur?

• Accents may differ in the environments in which particular phonemes do or do not occur.

• This may be looked at from either of two angles: as differences in the constraints on phonotactic distribution of given phonemes, or as differences in the phonological structures (syllable types, etc.) which are permitted.

• The phonotactic distribution of a given phoneme is the set of phonetic contexts in which it may occur.

  • It is not important, for instance, if one accent is adjudged to permit word-initial /vr/ (as in vroom!, vraic, vrille, where these items happen to be known and used), where another bars this initial cluster possibility.

• One fundamental division in English accent types depends upon a difference in phonotactic distribution of the consonant /r/.

  • In the rhotic accents /r/ can occur, with an overt phonetic realization, in a wide variety of phonetic contexts, including preconsonantal and absolute-final environments, thus farm [fɑrm], far || [fɑr].

  • In the non-rhotic accents /r/ is excluded from preconsonantal and absolute-final environments, thus farm [fɑ:m], far || [fɑ:].

  • The rhotic accents include those typical of Scotland, Ireland, Canada, Barbados, certain western parts of England, and most of the United States, including GenAm.

  • The non-rhotic accents include those typical of Australia, New Zealand, South Africa, Trinidad, certain eastern and southern parts of the United States, and most of England and Wales, including RP.

  • it is difficult to argue that an occurrence of /r/ is present underlying in the critical environment , but that it is obligatorily vocalised or deleted by rule so that it is not present phonetically.

  • There are also accents that can be classified as semi-rhotic, having lost preconsonantal /r/ but retaining it in certain word-final environments: thus Jamaican, with farm [fɑ:m] but far || [fɑ:r].

  • And one could call some southern speech (both British and American) hyper-rhotic, since it not only retains all historical /r/s but also has /r/ at the end of words such as comma, China.

• From a strict phonemic point of view, it can be argued that the quality of the vowel at the end of words such as happy is a difference in phonotactic distribution, since it is a question of the range of environments in which the phonemes /i:/ (FLEECE) and /ɪ/ (KIT) respectively occur.

How do accents differ in their phonemic systems, and what examples demonstrate this?

Accents may differ in their phonemic systems, i.e. in the number or identity of the phonemes they use.

• For instance, most accents of English have two distinct vowel phonemes in the close back area, a short /ʊ/ as in foot and a long /u:/ ... as in boot. But in Scottish English there is only a single phoneme, /u/, which corresponds to these two phonemes of other accents. Thus in a Scottish accent foot /fut/ is a perfect rhyme for boot /but/, and good /gud/ for mood /mud/.

→ Scottish English therefore has one fewer phoneme at this point in its vowel system than other accents: it differs from them SYSTEMICALLY

• All accents of England make a difference between pairs such as stock and stalk, RP /stɒk/ vs. /stɔ:k/. In some Scottish and Irish speech, though, as well as in Canada and in certain parts of the United States, stock and stalk, like don and dawn or cot and caught, are homophonous because the vowel system of these accents contains only a single phoneme in the open back area.

• An example of a possible systemic difference among the consonants arises in Scottish, Welsh and Irish English, with the possibility of there being a voiceless velar/uvular fricative /x/ in the consonant system of these accents — a phoneme missing from the consonant systems of RP and of most other accents. The opposition is illustrated by the minimal pair loch vs. lock.

• One part-system, part-system A (the 'short' vowels) has six members in some accents of English, as illustrated in the non-rhyming items pit, pet, pat, pot, cut, and put.

  • But a broad local accent of the north of England has only five members of the part-system, since cut and put have the same vowel and thus rhyme.

• In part-system B (long front-tending vowels or diphthongs) GenAm and RP have only one mid-height phoneme, that of face and similar words.

• But in some local accents of England and Wales there are two contrasting phonemes corresponding to the one of the standard accents: these accents distinguish a long monophthong /e:/ from a diphthong /ɛɪ/ so as to keep Dane distinct from deign and to cause late not to rhyme with eight

What is a subsystem, and how does neutralization relate to phonemic differences?

• It often happens that a given phonemic opposition is neutralized in certain phonetic environments.

• For example, the opposition of /p/ vs. /b/ is neutralised in the environment #s__

• The set of contrastive units operating in a position of neutralization may be referred to as a subsystem.

  • For example, in Belfast speech the opposition between short /ɒ/ and long /ɔ:/ may be illustrated by heterophonous pairs such as cot vs. caught, stock vs. stalk. But before voiced consonants the opposition is neutralized, so that for example don = dawn [dɔ:n], while cod and fraud rhyme.

• The American southern neutralization of /ɪ/ vs. /ɛ/ before nasals (pin=pen, although bit≠bet) is a subsystemic difference vis-à-vis accents that have no such neutralization.

How do accents vary in lexical distribution, and what are some examples?

• Accents (and indeed individual speakers) often vary in the phonemes they select for the lexical representation of particular words or morphemes: that is, they differ in the incidence of phonemes in a given lexical item or items.

• Thus every speaker has at his disposal both an /i:/ (as in beet) and an /aɪ/ (as in bite). In the word either some speakers use their /i:/, so that it rhymes with breather, but others use their /aɪ/, so that it rhymes with lither. Βy observing a given speaker’s vowel in these two words we can make no predictions about his pronunciation of any other words.

• Similarly with tomato (stressed vowel like that of father or fat or fate).

• Lexical-distributional differences are important where a substantial body of words behave in a parallel fashion.

• Most accents have contrasting vowel phonemes of the /æ/ and /ɑ:/ type, as in trap and palm (or father) respectively.

• Given these two vowels, though, some accents use /æ/ in bath, staff, grass, basket, while others use /ɑ:/.

• The choice between these two vowels in this lexical set is of indexical importance in distinguishing high vs. popular accents in the north of England, in distinguishing northern from southern accents in England as a whole and in distinguishing between the two standard accents RP and GenAm.

• A typical north-of-England accent and RP both contain the contrasting phonemes /ʊ/ and /u:/, as in put and boot respectively.

• But there are a few words in which the northern accent has /u:/ in spite of RP /ʊ/, thus hook, cook, look.

• Most popular controversies about 'right' and 'wrong' pronunciation concern issues of lexical distribution.

Should accomplish have the vowel of STRUT or the vowel of LOT?

Does Nazi rhyme with Benghazi, or should it have /-ts-/?

Should controversy be stressed on the first syllable or on the second?

What do the classifications of accent differences correspond to in phonology?

Differences in lexical incidence correspond to differences in the phonological shape of the representations stored in the speaker's mental lexicon.

Differences in phonotactic distribution and phonemic system in principle reflect varying conditions imposed on the phonological representations as a whole in the mental lexicon.

Differences in phonetic realization reflect differences in the rules which operate on stored forms to produce a phonetic output.

this four-way classification of accent differences is one springing naturally out of a taxonomic phonemic approach to phonology, and is subject to the possible shortcoming of this approach. Wherever phonemicization is in doubt (argument ab the nature of the stored forms), this classification is correspondingly uncertain.

Consider, for example, an accent such as Cockney where a vocoid [ʊ] is used in all environments where RP has a lateral non-vocoid [ɫ], thus milk [mɪʊk] (RP [mɪɫk]), shelf [ʃɛʊf] (RP [ʃɛɫf]), middle ['mɪdʊ] (RP ['mɪdɫ]).

Some Americans pronounce can't as [kæ̃t] (compare cat [kæt]).

If we treat this minimal pair as evidence for the presence of a phoneme /æ̃/ in the vowel system, then we shall have to say that this accent differs systemically from accents which preserve the historical /n/ in can’t as [n].

• Those who believe that RP start, [stɑ:t] and north [nɔ:θ] contain underlying /r/, would see the RP-GenAm difference at this point as being simply one of realization (compare GenAm [stɑrt, nɔrθ]). Those who agree with me in rejecting this analysis, and posit instead RP underlying, will see the difference as one of phonotactic distribution.

What are the consequences of differences between accents?

• Differences between accents, other than those differences which are merely realizational, typically result in words which rhyme in one accent not rhyming in another; in puns which work in one accent not working in another; and in utterances spoken with one accent being potentially misunderstood by a hearer who uses another.

• Words of one syllable are said to rhyme when they are phonologically identical from the vowel onwards, as face-place, rope-soap, saw-flaw.

• In polysyllabic words, or sequences of words, there must be phonological identity from the vowel of the stressed syllable up to the end, as happy-snappy, better-letter, edited-credited, stop it-drop it.

• The phonological identity on which rhyme depends does not necessarily imply phonetic identity.

• The pair lose-fuse is a good rhyme even in the kind of RP which has a clear allophonic difference between the [u:] of lose and the advanced [u:] of fuse.

What are settings according to Laver, and how are they divided?

• Settings, for Laver, are 'long-term muscular adjustments... of the speaker's larynx and supralaryngeal vocal tract'.

• They may have been acquired by social imitation, or idiosyncratically by individuals; but they quickly become unconscious and quasi-permanent.

• They fall into two types: those relating to the larynx, and those relating to the supralaryngeal vocal tract.

Larynx settings can be divided into three groups: phonation types, pitch ranges, and loudness ranges.

• Among phonation types (Catford 1964), as well as the obvious 'voiced' and 'voiceless' states of the glottis, important possibilities include 'creak' (which can be combined with voice as 'creaky voice') and 'murmur' (also termed 'breathy voice').

  • Thus Norwich working-class speech has been described by Trudgill (1974a: 186) as having a tendency to use creaky voice, while Norwich middle-class speech shows no corresponding tendency; on the other hand it has often been pointed out that RP-speaking men tend to go into creak towards the end of an utterance spoken with a low fall nuclear tone.

• Settings of the supralaryngeal vocal tract can be divided into four groups: longitudinal, latitudinal, those relating to tension, and those relating to nasalization.

  • Longitudinal modifications involve the raising or lowering of the larynx from a neutral position.

  • Latidunial modifications involve a constriction or expansion of the vocal tract at some particular point, and include various types of pharyngalization and labialization, as well as settings corresponding to segmental secondary articulations (velarisation, palatalisation, alveolarization).

  • Tension modifications involve the degree of overall muscular tension and its effect on the acoustic characteristics of the vocal tract.

  • Nasalisation and denasalization involve the operation of the soft palate and the consequent prevalence or scarcity of nasal resonance.

Within this framework one might, for instance, describe a typical Texan or Canadian male voice quality as 'lowered larynx voice', and that of working-class Norwich as 'raised larynx voice'.

• The characteristic voice quality of working-class Liverpool speech (Scouse) is velarized, involving a shift of the tongue's centre of gravity backwards and upwards.

Why do innovations in pronunciation tend to arise according to the principle of least effort?

• the fundamental reason why accents differ is that languages change. English pronunciation changes as time passes.

• The principle of least effort leads us to tend to pronounce words and sentences in a way which involves the minimum of articulatory effort consistent with the need to maintain intelligibility.

• If a simple articulatory gesture works just as well as a complex one, there is a natural tendency to prefer it, thus rendering the articulatory movements in speech simpler.

• For example, a voiceless [t] between vowels, as in better or atom, involves not only a tongue-tip movement up to the alveolar ridge and away again, but also a switching off and on again of the vibration of the vocal cords.

  • It is simpler, and requires fewer motor commands to the organs of speech, if the vibration of the vocal cords is continued throughout the alveolar articulation. The outcome is a 'voiced t', as in the typical American pronunciation of these words.

  • Another way of simplifying [t] is to abandon the alveolar component, concentrating all the articulatory modifications (switch-off of voicing, plosive occlusion) at the glottis. The outcome in this case is a glottal plosive, [ʔ], as in the typical Cockney (and wider British) pronunciation ['beʔə, 'æʔəm].

  • Or the alveolar articulation may not be abandoned, but merely carried out in a half-hearted way, so that no complete occlusion is effected: the outcome here is the alveolar slit fricative of Irish English.

• the principle of least effort provides a post hoc explanation, of each of these rival developments, but offers no guide as to why one rather than another should develop.

Maximal simplification of an articulatory gesture is achieved if we abolish it entirely; the maximal simplification of a segment is its deletion.

→ Hence we can reasonably attribute the loss of historical /r/ in specified environments in the non-rhotic accents to the principle of least effort: to pronounce start as ['stɑ:t] is simpler than to pronounce it as ['stɑrt].

How do naturalness, timing, sequence, assimilation, and infantilism play a role in sound changes?

• Some segment types are more natural than others: they are learnt earlier by children, they are found more widely in the languages of the world, and pronunciation changes tend to work towards them.

Naturalness is not quite the same thing as articulatory simplicity. For example, it is evident that the natural lip position for vowels is unrounded for open and front vowels (as [a, e, i]) but rounded for non-open back vowels (as [o, u]).

Now when English velarized [l] is simplified by the loss of its alveolar lateral component (L Vocalization, 3.4.4 below), the articulatory residue is a back close vocoid ... a vowel or semi-vowel of the [ɣ] type: closish, back, but unrounded.

In many areas where L Vocalization has occurred, e.g. in popular London speech, the resultant vocoid is in fact usually rounded, and thus similar to [o] or [u], thus middle ['mɪdo] etc. Although this roundedness necessitates an additional articulatory movement, namely lip rounding, it does result in a segment type which is more natural than it would have been otherwise.

• In a word such as mince ['mɪns], for example, three adjustments of the organs of speech are required in order to effect the transition from [n] to [s] ... Unless all three changes happen simultaneously, a transitional segment will result. For example, if the soft palate completes its rising movement before the tongue tip comes away from the alveolar ridge, an epenthetic plosive will come about, thus ['mɪnts].

• It appears that - other things being equal - a regular alternation of consonants (C) and vowels (V) is more natural than clusterings of one type or the other.

Thus the treatment of /r/ in non-rhotic accents results in more natural sequences than were present historically or persist in rhotic accents, since ['bɔ: ju:, bɔ:rɪŋ, bɔ:d] for bore you, boring, bored board are CVCV, CVCVC, CVC, while rhotic ['bɔ:r ju:, bɔ:rɪŋ, bɔ:rd] (etc.) involve two clusters of [r] plus another C.

• A way of reducing the articulatory complexity of strings of consonants is through assimilation, the process whereby a sound is made phonetically more similar to the sounds constituting its phonetic environment.

We know, for instance, that the word which is nowadays ant [ænt] had an Old English form æmete ... The change from [m] to [n] before a following [t] is an assimilation which results in an obvious articulatory simplification, namely the elimination of a labial movement.

• As children learn to speak, it can be observed that they tend to replace complex or less natural segments and sequences with simpler or more natural ones.

But this does not always happen: sometimes, instead, they retain an infantile makeshift into adult life. Where this happens among a whole community rather than just with individual speakers, we have a mechanism whereby sound change may come about.

Compared with them, the labiodental fricatives, [f] and [v], are more natural; and children, as is well known, readily substitute them for the difficult dentals, thus thing [fɪŋ], mouth [maʊf], mother ['mʌvə], etc. The prevalence of these pronunciations among adult working -class Londoners or American blacks can thus be regarded as a persistent infantilism.

The alternative development, found in Ireland, the West Indies, New York and elsewhere, replaces the dental fricatives with dental or alveolar plosives [t,d]

How does the necessity to preserve intelligibility act as a countervailing pressure to ease of articulation?

• Ease of articulation, if allowed free rein, would lead to nothing but a uniform grunt, perhaps [ə:], for every word in the vocabulary.

• It is held in check principally by the countervailing pressure of the necessity to preserve intelligibility and thus facilitate communication.

• Several of the innovations mentioned so far tend to lead to the loss of certain contrasts and thus to the creation of new homophones and new opportunities for misunderstanding.

Allowing [t] to become voiced risks making latter and ladder identical in sound, causing uncertainty between waiting in the river and wading in it, or between a lamb which is bleating and one which is bleeding.

Allowing an epenthetic [t] in mince risks making it homophonous with mints, so that 'Go and buy some [mɪnts]' is ambiguous.

Dropping the /r/ from bored/board may make this word homophonous with bawd or lead to confusion

Using /f/ for standard /θ/ will bring about confusion between thought/fought

• the obvious usefulness of preserving distinctions reinforces the natural human tendency towards conservatism in social institutions, language among them

• superficial appearances to the contrary notwithstanding, children to tend to model themselves upon their parents and upon societal norms. In pronunciation this implies the accurate imitation of existing pronunciation patterns, and hence resistance to the principle of least effort.

push-chain

Where sounds are rather close to one another in auditory space, a change in the phonetic realization of one may precipitate a change in another, adjacent, phoneme, so as to reduce possible confusion between them.

Given that a change takes place in /æ/ whereby it becomes closer, approaching [ɛ], there is going to be a risk of confusion between bad and bed, gnat and net, etc., unless this other vowel changes in its turn, by becoming closer (as has happened in southern-hemisphere English), or else by becoming somewhat centralized (as in the United States).

There may then be a knock-on effect on /ɪ/, making it too become closer (as in Australian) or central (as in New Zealand).

Such a chain of events is known as a push-chain.

drag-chain

A sound which undergoes a change may also leave behind it a vacuum, as it were, in the auditory space it formerly occupied.

An adjacent sound may then move in such a way as to fill this vacuum, or at least so as to restore the same auditory distance as previously obtained.

This process, too, many affect several members of a vowel or consonant system; it is known as a drag-chain

• The fronting of the [ɑ:] of father towards [a:], something which has happened in certain kinds of American English as well as in Australian and New Zealand, is perhaps to be explained as a drag-chain effect consequent on the raising of [æ].

What are splits and mergers in the system of oppositions?

• Some sound changes lead to alterations in the system of oppositions.

• In some cases a phoneme undergoes a split, so that what was previously a single phoneme now becomes two, as its erstwhile allophones achieve independent phonemic status.

• In other cases phonemes undergo a merger, so that what were previously contrasting phonemes now cease to contrast, becoming merged into a single phoneme.