Caught between Aristotle and Miss Marple ... – A Proposal for a Perceptual Prototype Approach to “ Estuary English ”

More than thirty years after the term was coined by David Rosewarne (1984), linguists have not come anywhere near to agreeing on a linguistically sound definition of the concept of ‘Estuary English’ (EE). Nevertheless, the term has come to stay. According to John Wells (2013), “we can now expect to be readily understood if we describe someone's speech as ‘estuarial’”. In this paper, I will argue that ‘Estuary English’ is the name for a heuristic conceived of and popularized by linguistic laypeople and therefore defying expert linguistic analysis in terms of Aristotelian categories, even sociolinguistically “enlightened” ones. Following Taylor (2003) and Kristiansen (2008), I will suggest describing the resulting folk-linguistic category in terms of the graded structure/prototype approach. In support of the prototype hypothesis, I will present data from an ongoing project in perceptual dialectology. It includes judgements of gradience of membership of 171 speakers from the South East of England, the Midlands and Scotland. Asked to rate the recordings of three young middle-class speakers from three Southeastern towns, these speaker-listeners are remarkably consistent in their responses and sometimes remarkably at odds with the analysis of expert linguists.


Introduction
More than thirty years after the term was coined by David Rosewarne (1984), linguists have not come anywhere near to agreeing on a linguistically sound definition of the concept of 'Estuary English' (henceforth referred to as EE).One could therefore argue that it was time to lay it to rest, together with other buzz words of the 1980s, such as 'Essex men' or 'street cred'.However, there are at least two reasons for not doing so.For one, EE has come to stay (see e.g.Deterding _____________ 1 Leibniz Universität Hannover E-mail: ulrike.altendorf@engsem.uni-hannover.de2005, Kerswill 2006, Eitler 2006, Hickey 2007, Kristiansen 2008, Bonness 2011, Crystal 2010, Wells 2013, Braber 2016).According to John Wells (2013), "we can now expect to be readily understood if we describe someone's speech as 'estuarial'".The second reason for not giving up on the concept yet is its rather "annoying" habit of raising theoretical questions which I consider more important than the concept itself.The most important of these questions is of epistemological nature and concerns the categorization of linguistic experience by interested, albeit non-expert language users.It is in this context that Aristotle and Miss Marple come into play.These two unlikely personalities represent two diametrically opposed ways of conceptualizing the world.The professional "academic" Aristotle demands clear-cut categories with well-defined necessary and sufficient criteria while amateur detective Jane Marple draws her conclusions on the basis of village parallels.In In this paper, I will argue that EE is the name of a heuristic conceived of and popularized by linguistic laypeople and therefore defying expert linguistic analysis in terms of Aristotelian categories, even sociolinguistically "enlightened" ones.Following Taylor (2003) and Kristiansen (2008), I will suggest describing the resulting folk-linguistic category in terms of the graded structure/prototype approach.

Not so humble beginnings: a new category
When David Rosewarne coined the term 'Estuary English' in 1984, he made two claims: (a) He located EE speakers at the centre of two interrelated accent continua, the social and the regional: "If one imagines a continuum with RP and London speech at either end, 'Estuary English' speakers are to be found grouped in the middle ground" (Rosewarne 1984: 29).(b) He classified the accent of this group of speakers as a linguistic "variety": "'Estuary English' is a variety of modified regional speech" (Rosewarne 1984: 29).
It is the latter claim that is the source of the ensuing controversy.From a sociolinguistic point of view, intermediate speakers on a socio-regional accent continuum are in themselves not problematic.They can be easily accommodated within the traditional Labovian paradigm.In his pioneering New York department store study, Labov himself identifies a linguistically intermediate group of speakers.However, he does not categorize their way of speaking as a variety in its own right and he does not coin a new name for it, such as "Macy's English".Rosewarne, on the other hand, does both and therefore invites trouble from the community of academic linguists.

Wells's early description and criticism of EE
The first expert linguist to tackle the notion of EE is John Wells (1992Wells ( , 1994a)).He is the first to draw up a list of phonetic EE features using expert phonetic terminology. 2His list consists of features in which EE "differs from Cockney" and in which "EE agrees with Cockney, but differs from RP" (Wells 1992).However, Wells has "issues" with delineating EE on the Southeastern accent continuum, which he summarizes as follows: Is EE a variety (lect, dialect) in its own right, or just the formal style for which Cockney is the informal?Does EE include stylistic variation?Does informal EE overlap with formal Cockney?Where is the boundary between EE and RP?Is localizability a workable criterion?(Wells 1992).
Wells's "issue" with the varietal status of EE as well as his strong interest in features and boundaries are very revealing.They show his underlying conceptualization of a variety as a category "in its own right", a conceptualization that is Aristotelian in nature.

The Aristotelian category within the context of Labovian linguistics
The notion of the Aristotelian category constitutes the classical approach to categorization and is "'classical' in two senses" (Taylor 2003: 20).It "goes back ultimately to Greek antiquity" (Taylor 2003: 20) and is "behind all of the theories of linguistic structure that have been presented in the twentieth century" (Labov 1973: 342, see also Taylor 2003: 20).In this paper, the definition of the Aristotelian category is based on a common core of characteristics shared by Taylor (2003: 21) and Rosch (1999: 63).Two of the three assumptions are also part of Labov's (1973: 342) concept of the "categorical view": 3 (1) Categories are defined in terms of a conjunction of necessary and sufficient features (Taylor 2003: 21, see also Rosch 1999: 64, Labov 1973: 342, "conjuntively defined").
(3) All members of a category have equal status (Taylor 2003: 21, see also Rosch 1999: 64)._____________ 2 Rosewarne and Coggle use partly folk-linguistic descriptions, which have to be explicitly "translated" into expert linguistic terminology to allow comparison with academic studies (Altendorf 2003: 14). 3 Labov's (1973: 342) "categorical view" is even stricter.It "includes the implicit assertions that all linguistic units are categories which are: (1) discrete, (2) invariant, (3) qualitatively distinct, (4) conjunctively defined, (5) composed of atomic primes".Taylor (2003: 21) adds a fourth requirement -"Features are binary"-which is at the heart of the Aristotelian conceptualization: "But we have now posited that it is impossible for anything at the same time to be and not to be […]" (Aristotle, Metaphysics, Part 4).It is also shared by Labov as a characteristic of the "categorical view", where it is termed "qualitatively distinct" (Labov 1973: 342).However, this requirement does not sit well with the principle of 'orderly heterogeneity' of (Labovian) sociolinguistics.Labov himself therefore suggests relaxing the categorical requirement: Instead of taking as problematical the existence of categories, we can turn to the nature of the boundaries between them.As linguistics then becomes a form of boundary theory rather than a category theory, we discover that not all linguistic material fits the categorical view: there is greater or lesser success in imposing categories upon the continuous substratum of reality (Labov 1973: 343).
In (conscious or unconscious) compliance with this suggestion, the whole bulk of expert linguistic literature on EE seems to be such an exercise in "boundary linguistics" starting with Wells's efforts of identifying the features that separate EE from both Cockney and RP.Wells (1992Wells ( , 1994a) has adopted the sociolinguistically 'enlightened' Labovian version of the Aristotelian category leaning, however, more towards the Aristotelian than towards the Labovian side.His lists of features consisting of variants which are present in one category but absent from the others follow the Aristotelian rationale of binary features.However, he also allows for quantitative variation across category boundaries by adding the comment "perhaps variably" to the following list of features: EE agrees with Cockney, but differs from RP, in having (perhaps variably) tense vowel in HAPPY   (Wells 1992).

Wells's early description and criticism of EE in the light of the sociolinguistically "enlightened" Aristotelian approach
At the time when this is list was drawn up, the first two, probably four variants, were already used by the then young generation of RP speakers (see Wells 1994b, Altendorf 2003).With so little external distinction, Wells (1994a) is not convinced that "EE is to be regarded as a variety (lect, dialect) in its own right" (261).Instead he can also imagine considering it as "simply the formal style/register for which Cockney is the informal one" (261).And instead of using a new name, he "would really prefer to call this variety simply London English" allowing for an "ambit much wider than the GLC area, covering at least most of the urban South East" (261).However, by doing so he would deny varietal status to EE and instead follow the route taken by Labov in New York.

Falling between two stools: folk-linguistic usage and expert linguistic concepts
In the 1990s, the term and concept of EE continued to develop along two lines, the folk-and the expert linguistic line.This split is not surprising given the intermediate position between folk and expert linguistics occupied by Rosewarne himself.The folk-linguistic line emerged as EE hit the headlines and became generally known. 4During this process, the semantics of the term were extended to an ever wider range of linguistic phenomena (for a detailed discussion, see Altendorf 2003: 16-26) and it became a shorthand for a number of different and partly divergent trends ranging regionally from London to Glasgow and socially from the working to the upper classes (see Altendorf e.g. 2003: 3).Puzzled by the success of the term and concept outside the trade, professional linguists unpacked their toolkit to get to the bottom of this new concept and found it lacking.In the following, only a selection of these studies, all of them focusing on the regional dimension of EE, will be discussed (for a methodological comparison, see Altendorf 2012).

EE in the South East of England
The South East of England is considered to be the "heartland of Estuary" (Coggle 1993: 28) or the "koine core" (Britain 2005(Britain : 1000) by both folk and expert linguists.The general claim made by proponents of the popular linguistics faction is that EE variants are spreading regionally at the expense of local variants.The following quotation from a popular novel serves to illustrate this point: "His wasn't a Suffolk voice, rather the accent dubbed in the eighties Estuary English (Vine 1998: 155).
In principle, most expert linguists do not deny that accents in the South East of England "are converging in both inventory and realizations" (e.g.Williams and Kerswill 1999: 149) and that "at least some of the phonological features" which are spreading are those "associated with 'Estuary English'" (Trudgill 2001: 10).What they dispute is that the process has evolved to such an extent that Southeastern English has emerged as a (1) homogeneous (e.g.Trudgill 2001: 10-11) and (2) distinct variety (e.g.Wells 1994a: 261).These are also the conclusions drawn by Joanna Przedlacka from the results of her empirical investigation of the speech of 16 teenage speakers from Essex, Kent, Buckinghamshire and Surrey: _____________ 4 An overview of articles published on this topic in the general media can be found on Wells's Estuary English website covering the period from 1998 to 2007 (Wells 1998(Wells -2007)).
(1) […] EE is in a sense a variety between RP and Cockney […].Such statement, however, is an oversimplification of the issue, since geographical differentiation must not be ignored (Przedlacka 2002: 94).
(2) At the same time, what is known as "Estuary English" appears to be part of more general changes (Przedlacka 2002: 97).
The geographical variation Przedlacka refers to is mostly based on quantitative frequency differences that are sometimes statistically significant and sometimes not (for a summary, see Przedlacka 2002: 71-72).In this respect, her varietal patterns bear considerable resemblance to those identified in my own study of the speech of female teenage middle-class speakers from London, Colchester and Canterbury (Altendorf 2003).In my opinion and only as far as my sample goes, my speakers are sufficiently homogeneous to be considered speakers of an overarching, of course internally variable and therefore heterogeneous, regional variety. 5What Przedlacka and I do not know is how much internal heterogeneity is permissible for this heterogeneity to still qualify as "orderly". 6In Altendorf (2013), I argue that the resulting feeling of unease is due to a langue-parole tension between the notion of the "variable" that Labov locates "within the system" (e.g.Weinreich, Labov and Herzog 1968: 167) and the locus of sociolinguistic research which Labov places at the level of "speech-language as it is used in everyday life" (Labov 1972: xix).In this article, I would like to take this line of argumentation further by analysing in more detail the categories behind the categories, or to be more precise the Aristotelian ideal that is interwoven into the notion of 'variety' and conflicts with the ideal of "(orderly) heterogeneity".

EE in the South East of England
Although the term 'variety' is a "cardinal term of sociolinguistics" (Berruto 1987: 264), few authors have reflected upon the basic, often tacit, assumptions underlying its conceptualization.Exceptions are, for example, Berruto (1987) and Wunderli (1992).Despite recognizing the problems inherent in the concept, Berruto ventures a cautious definition (1a) and sets up two requirements (1b) and ( 2): (1a) A linguistic variety is indeed characterized by certain realizational forms of the language system co-occurring in a predictable way with certain social and functional characteristics of language use (Berruto 1987: 264, my translation and my italics).
(1b) Within a variety, a certain degree of homogeneity and stability is to be required (even though sociolinguists assume internal variability for each linguistic variety) (Berruto 1987: 265, my translation and my italics).
(2) While one can certainly claim that the concept of 'variety' should imply 'discreteness', it may be more appropriate in view of the facts to understand The problem of external distinction will be discussed in section 3.1.2.

6
More recent work in dialectometry may be able to propose a solution to this question.It will, however, not be able to account for the use of the term for accents outside the South East.
'varieties' as points on a continuum (defined by convention and not easily distinguishable from one another) (Berruto 1987: 265, my translation).
These three extracts form a bridge between Labov's sociolinguistically "enlightened" Aristotelian category in general and Przedlacka's conclusions with regard to EE in particular.In (1a), Berruto's "certain realizational forms of the language system co-occurring in a predictable way" parallels the Aristotelian "conjunction of necessary and sufficient features" (Taylor 2003: 21).In (1b), Berruto's requirement of "a certain degree of homogeneity and stability" is an already watered-down version of Aristotle's binary-feature requirement that Berruto relaxes further by adding, rather reluctantly and in brackets, the tenet of "internal variability".In the following, these two requirements will be referred to as the "ideal of internal cohesion", a term coined by Haugen (1966: 928). 7It is this ideal that Przedlacka applies in her conclusion (1).In (2), Berruto addresses the external boundaries of a variety.His postulate of "discreteness" parallels Aristotle's "clear boundaries" (Taylor 2003: 21).Again, Berruto relaxes this requirement by adding, again reluctantly and partly in brackets, another Labovian tenet discussed above, i.e. that boundaries can be fuzzy because "there is greater or lesser success in imposing categories upon the continuous substratum of reality" (Labov 1973: 343).In the following, this requirement will be referred to as the "ideal of external distinction" (Haugen 1966: 928).It is this ideal that Przedlacka addresses in her conclusion (2).Both Przedlacka's and Berruto's conclusions follow a pattern that I will provisionally term the YES-BUT pattern.It is the pattern generally identified by researchers studying (the regional dimension of) EE empirically.Rosewarne (1984: 29) locates EE firmly in the South East of England.In a later publication, he sees its potential for future geographical expansion "westwards into Wales and northwards to the Scottish border" (1994a: 8) but points out that this development had not yet taken place in 1994.Paul Coggle (1993), on the other hand, goes further than Rosewarne in claiming that EE in addition to extending "south-east of London to the south Kent coast" (26) also "extends to the north-east of London as far as the north Norfolk coast [and] to the south-west of London as far as the Dorset coast" (26).For a sub-section of the accents of the area North-east of London, David Britain (2005) provides an empirical investigation, the summary of which fits the YES-BUT pattern to a T: YES Despite the fact that some features of the southeastern koine have diffused to the Fens, BUT enough local differentiation still survives for us to claim that the variety has not (yet) been fully swept up into the empire of "Estuary English" (Britain 2005(Britain : 1017))._____________ 7 Haugen (1966: 928) identifies "internal cohesion-external distinction" as the ideals for national standard varieties where the requirements are even stricter.

The YES-BUT pattern in the Fens
In the Fens, this "local differentiation" can be more easily established than in the South East. 8Local differences among the three localities in the Fens are not only quantitative but also clearly qualitative.Young speakers in Spalding, for example, still show signs of the "dialectologically northern" FOOT-STRUT Merger (Britain 2005(Britain : 1012) )  9 and young speakers in the Terringtons remnants of the East Anglian MOAN-MOWN Split (Britain 2005(Britain : 1012(Britain -1014)).In addition, these substratum differences lead to different outcomes "showing evidence of the interaction of the innovation with the traditional local form" (Britain 2005(Britain : 1017)).We can thus conclude that the requirement of internal cohesion is more clearly violated if one adds the Fens to the region allegedly covered by EE.On the other hand, we can also conclude that the features associated with EE are not confined to the South East and thus violate the requirement of external distinction.Both tendencies become even more obvious if we examine the claim made by some journalists that EE is now also spoken in Glasgow.

The YES-BUT pattern in Liverpool and Glasgow
The controversy surrounding the term 'Jockney' (blend of 'jock' for 'Scot' and 'Cockney', the latter used as a synonym of EE) has finally brought to light the seemingly irreconcilable ways in which journalists and linguists perceive the world of language variation and probably even the world in general.We owe it to Wells's excellent webpage on EE that this controversy has been made known to a wider public.According to some journalists, "Estuary English has taken the high road" (Corbidge 1998: 3).On its way to the north, it is accused of threatening to replace both Scouse in Liverpool (e.g.Marks 1999) and Glaswegian in Glasgow (e.g.Corbidge 1998: 3, Harris 1999).The journalists quote the work of expert linguists as their source of information."As is only to be expected" (Harmer 1999), both linguists feel misquoted and publish their corrections on Wells's webpage.Again, their corrected assessment results in a YES-BUT statement.
Andrew Harmer 1999 on Scouse: I am aware that my use of the term "South East" constitutes a simplification since the boundaries of this area are also not as clear-cut as the current use of the term seems to imply (for a more detailed discussion, see e.g.Britain 2005: 999-1000).9 Spalding, which is "dialectologically northern" (Britain 2005(Britain : 1004) ) with "[a] in the BATH lexical set and [ʊ] in the STRUT set" (1003), falls outside the area described by Coggle (1993: 27) who considers what he calls the "bath and love boundary" almost as "sturdy a barrier to contend with" as "geographic boundaries", such as "the sea".Such "subtleties" are typical of academic researchers but lost on journalists with a completely different agenda.What is interesting, however, is that in the case of Glasgow not even the "careless journalist" (Wells 1999) considers the urban accent to be part of an overarching EE 'variety'.The term 'Jockney', inexact as it may be, does take the issue of "internal cohesion", or rather the lack of it, into account.
Stuart-Smith's subsequent analysis of the "facts behind such reports In all cases, the YES-part of the YES-BUT pattern refers to a selection of the same group of variants associated with EE, such as T Glottalling, TH Fronting, L Vocalization, labio-dental (r), YOD Coalescence, S Retraction and GOOSE and FOOT Fronting and sometimes (parts of) the London Diphthong Shift (LDS) (Wells 1982).I hypothesize that it is this group that is the key to the notion of EE.In Altendorf (2003), I call this group "a pool of features".At the time the term was not taken from or related to Mufwene's (2001) "feature pool" although the concepts are similar.

Pool of features or the need for a new category
In 2003, my term "pool of features" simply described a group of features which were available to language users.What they have in common is that they can be interpreted as improvements in structural and/or articulatory terms (Altendorf 2003: 143-148) and that they allow speakers to perform acts of affiliation and distancing with or from other social groups within the same speech community (Altendorf 2003: 151-157).The disadvantage of the term "pool of features" was that it had no recognized theoretical status.It was chosen, however, exactly because of this.Alternative categories, such as 'variety' or 'lect' did not seem appropriate, for the reasons discussed above.When referring to these variants, other linguists also resort to terms which are outside the established nomenclature.For a subsection of this group, Trudgill (2003) chooses the term "linguistic changes in pan-world English", Milroy (2007) "off the shelf changes" and Williams and Kerswill (1999) "youth norms", to name but a few, all of them avoiding the terms 'variety' or 'dialect'.Lay speakers are faced with a similar problem.In their eyes, the terms known to them, such as dialect, Cockney, RP, Scouse and Glaswegian, do not seem to fit either.What they appear to be looking for is a term that reflects the perceived "recentness" and "pervasiveness" of the trend(s) in question.A new term is therefore very welcome and one that "kills two (or more) birds with one stone" comes in handy.And "Estuary English" is exactly this, new and handy.Everything that makes EE inappropriate in the eyes of linguists makes it a handy category in the eyes of lay speakers.The question we need to address after over 25 years of staring/glaring at each other in incomprehension is why this is the case.

A change of perspective
So far expert linguists have mostly looked at the object under categorization, i.e. they have analysed production data with the aim of establishing the linguistic characteristics of EE.However, this is not the major source of the continued controversy.The controversy is mainly between expert and non-expert linguists about the different ways in which they perceive and categorize the varietal patterns that they encounter.I therefore suggest a change of perspective by shifting the focus of investigation (a) from the object of categorization to the categorizing subjects, (b) from an analysis of speaker production to an analysis of listener perception and (c) from expert to lay categorization hypothesizing that it is at the level of lay hearer perception that the term is located.

Folk categories as prototype categories
Taylor ( 2003) claims that experts and non-experts categorize the world differently.He therefore distinguishes between "expert" and "folk categories".According to Taylor (2003), expert categories "have been specifically created, usually in conformity with Aristotelian principles" (75).Folk-categories, on the other hand, are more similar to prototype categories: Folk categories are structured around prototypical instances and are grounded in the way people normally perceive and interact with the things in their environment (Taylor 2003: 75, my italics).
In other words, when categorizing an everyday object or experience, people rely less on abstract features of categories than on a comparison of the given object or experience with what they consider to be the object or experience best representing the category.As a result, prototype categories are […] internally structured into a prototype (clearest case, best examples) of the category with nonprototype members tending towards an order from better to poorer examples (Rosch 1975: 544, my italics).
If this is the case, then laypeople, including naïve speakers of the language, apply a type of category that is almost diametrically opposed to an Aristotelian category, even to the type that is already "sociolinguistically enlightened".Apart from allowing graded membership (3), prototype categories also function without requiring "internal cohesion" (1) and "external distinction" (2):

Aristotelian Categoriesthe classical view
Prototype Categoriesthe graded structure view (1) "Categories are defined in terms of a conjunction of necessary and sufficient features".
(1) "Many categories have no, and no category need have any, necessary and sufficient attributes".
(2) "Graded structure categories do not have clear-cut boundaries".
(3) "All members of a category have equal status".
(3) "Items in a category are not equivalent with respect to membership but rather possess gradations of membership".Source : Taylor 2003: 21 Source: Rosch 1999: 68-69 Despite their greater flexibility, prototype categories also benefit from a certain amount of internal cohesion (1) and external distinction (2): The principle of family resemblance relationships can be restated in terms of cue validity since the attributes most distributed among members of a category and least distributed among members of contrasting categories are, by definition, the most valid cues to membership in the category in question (Rosch and Mervis 1975: 575-576, my italics).Taylor (2003: 58) concludes that "in this respect, the centre of a prototype category approaches the ideal of a classical category".High scores for cue validity are, however, not a necessary pre-requisite for prototype categories.As Rosch and Mervis (1975) have shown, "many categories show up with few or no attributes in common" (Rosch 1999: 66-67).The authors therefore prefer to "use the term family resemblance rather than cue validity" (Rosch and Mervis 1975: 576).At this point, it is important to keep in mind that we are NOT dealing with categories set up by experts in accordance with the requirements of methodological rigor but with categories formed by laypeople in everyday contexts.

Language varieties as prototype categories
The idea of applying the graded structure approach to language variation has been advanced by Gitte Kristiansen who suggests defining language varieties as prototype categories: If lectal varieties constitute prototype categories, some realizations will be more "typical" or "central" or "better examples" of a given variety than others (Kristiansen 2008: 59, my italics).
In support of her suggestion, Kristiansen (2008: 59-60) draws attention to Wells's description of Received Pronunciation: Some people deny that RP exists.This seems to be like denying that the colour red exists.[…] We may hesitate about a particular person's speech which might or might not be 'RP' or 'Near-RP' […] and define it more narrowly or more widely than I have done; but anyone who has grown up in England knows it when he hears a typical instance of it (Wells 1972: 301, my italics). 10  Wells is probably right when he is confident about native speakers of English English recognizing a high-profile accent such as RP.However, naïve speakers of a language are known to make mistakes when classifying language varieties as do laypeople in other areas when they are out of their depths: Humans have receptive competence of lectal varieties, but the images formed are not necessarily accurate, at least from the perspective of experts such as linguists (Kristiansen 2008: 61, my italics).
Experts, by the way, can be wrong, too, as shown by Köster et al. (2012) in their study of "identification of regionally marked speech in German telephone conversations by forensic phoneticians".

Are all varieties folk categories?
Among the first to apply the concept of varieties as prototype categories to her dialectological work is Elissa Pustka (2009). 11Pustka goes so far as to relegate the notion of 'variety' completely to the realm of folk linguistics: Since we can observe (continuous) variation on the 'objective' level of linguistic facts, varieties only exist in speakers' cognitive representations.In other words: varieties are per se 'subjective' and belong therefore to folk categories […] (Pustka 2009: 80).
Although this is certainly an idea worth considering, I do not suggest going as far as this.Linguistic terms, such as 'variety' are, after all, not the only terms employed by experts and laypeople alike.They are part of a group of terms to which Putnam's "hypothesis of the universality of the division of linguistic labor" 12 applies: Every linguistic community […] possesses at least some terms whose associated criteria are known only to a subset of the speakers who acquire the terms, and whose use by the other speakers depends upon a structured co-operation between them and the speakers in the relevant subsets (Putnam 1975: 146).It is interesting to note that both Wells and Trudgill cite RP as another example of a term and concept that they consider "equally unsatisfactory" as EE (Wells 1994a: 261) and "not particularly felicitous either" (Trudgill 2001: 10).
11 I am extremely grateful to Elton Prifti for an inspiring dinner conversation on prototypes and for drawing my attention to this study.
12 Strictly speaking, Putnam's hypothesis only applies to natural kind terms but in accordance with Geearerts (2008: 30) I suggest applying it to a wider range of terms, especially technical terms.
In this situation, one has to accept that "the 'average' speaker who acquires [the term] does not acquire anything that fixes its extension" (Putnam 1975: 146).It is then the expert's job to draw boundaries "around essentially fuzzy categories" and to set up "criteria on which membership is to be decided" (Taylor 2003: 75, following Wittgenstein).It is exactly this "boundary work" that Labov (1973: 343) suggests for cases in which the "categorical view" does not fit the facts.Linguists have been doing this when investigating the claims made by lay speakers about EE.What they have found are the fruits of Putnam's "division of linguistic labor".

EE as a folk-linguistic prototype theory
The first clue to this reading of the findings has been provided -probably inadvertently -by Przedlacka.Her "in a sense" in "Estuary English is in a sense a variety between RP and Cockney" (2002: 94, my italics) can be interpreted as belonging to the class of "hedges" that are known to be used to express the degree of membership in a prototype category.In terms of the prototype approach, she has shown that Southeastern middle-class accents lying between RP and Cockney are a good example of the Estuary English prototype category.The next clue is provided by the YES-BUT pattern.The YES-part, I would hypothesize, constitutes the prototype and the BUT-part describes the position in the periphery of the graded category.The YES-part is the "pool of features" containing T Glottalling, TH Fronting, L Vocalization, labio-dental (r), YOD Coalescence, S Retraction, GOOSE and FOOT Fronting and parts of the LDS, as described earlier on in this paper.These features occur in different combinations in the more and less prototypical members of the category and therefore bind these trends, seen by expert linguists as different and divergent, together.It is this pool that establishes Wittgenstein's "family resemblance" and causes laypeople to perceive striking similarity where experts see at best superficial resemblance.

Low cue validity
A weakness of the approach is that a prototype consisting of the variants listed above has very low cue validity.As pointed out before and described in more detail in Altendorf (2004), these variants are "off the shelf changes" (Milroy 2007) in "pan-world English" (Trudgill 2003) and "youth norms" observed in many parts of Britain (Williams and Kerswill 1999).Nevertheless, they seem to work for naïve speakers.So far not even a "careless journalist" (Wells 1999) has suggested considering American English as Estuary English although, say, T Glottalling (e.g.Eddington and Channer 2010), L Vocalization and GOOSE Fronting (for an overview, see Altendorf 2004: 93) can also be found in the United States.In a similar line, journalists use the term 'Jockney' to describe the outcome of the occurrence of EE features in Glaswegian thus suggesting a blending of two different categories and not the complete replacement of one by the other.This may reassure the expert that laypeople do not necessarily follow an "anything-goes policy".However, to understand their "policy" better, one would need to go deeper into the processes on the basis of which they perceive similarity.This, unfortunately, goes beyond the scope of the present paper.At this point, I would hypothesize that one central categorization criterion is that the EE prototype is typologically Southeastern.Rhoticity or the FOOT-STRUT merger may be too salient as features of other also prototypically known accent regions.Coggle (1993: 27) names these isoglosses as "sturdy barriers" to the spread of EE.Restated in perceptual terms, this means that most speakers would not categorize rhotic speakers (including Glaswegian speakers) or speakers who merge FOOT and STRUT as speakers of EE.However, it is conceivable that they think in terms of blended categories.

Production and Perception Data
From a methodological point of view, my own on-going work on EE bears a certain resemblance to Pustka's survey of Southern French in Aveyron.Like Pustka (e.g. 2009), I began by collecting and analyzing production data and proceeded later on to a perceptual study based on extracts from the original corpus.
In 1998, I interviewed altogether 40 speakers in three Southeastern towns, London, Colchester and Canterbury, applying traditional Labovian methodology.From this corpus, I chose three female teenage middle-class speakers from each of the three towns who read out the same text passage.An extract from these recordings was played to 171 listeners from all over England in 2015.

Speaker variables
Tables 3 and 4 show the participants' regional origin and age: Since the major data collection sessions took place at two universities, the majority of informants were students between 20 and 30 years of age.The universities were located in the South East of England (Brighton) and further north (Glasgow) in order to provide data from both in-group and out-group informants. 13In addition, I came across a group of regionally mobile speakers from other areas, especially from the Midlands.

Questionnaire
Informants were asked to fill in a questionnaire 14 in which the speech samples were embedded.Questions centred on the informants' socio-regional background and their perception of the accents of the three speakers.Results obtained until now from three of these questions will be presented and discussed in the following.

Estuary English
If participants indicated that they knew the term 'Estuary English', they were asked to categorize the three speakers: Table .5. Familiarity with the term EE.Table .6. Categorization.
The perceptual categorization results tally well with the results obtained from the articulatory analysis.Nadia, who displays the lowest percentage of "EE variants" (Altendorf 2003), is also the one least often categorized as a speaker of EE by lay informants.In this study and up to now, informants have proved to be quite coherent with regard to which Southeastern accents they subsume under the cover term of 'Estuary English'.Their judgements are not as "wrong" as experts might expect since they choose the speakers with the highest percentage of variants associated with EE.

Local differentiation
Before mentioning the term 'Estuary English', I had asked the informants to place the three speakers locally.No single informant, not even among the older speakers, was able to do this correctly for all three speakers at the level of counties.We can conclude that local differentiation within the South East does not seem marked enough for (young?) lay speakers to place South-eastern middle-class accents locally.Although this may be due to their lack of perceptual competence and life experience, it does explain why applied linguists and lay speakers are quite happy to classify EE as a supra-local Southeastern accent.Informants were, however, more successful in placing the speakers regionally.It is not surprising that the majority of informants from the South East as well as from Scotland classify the accents of the three speakers as Southeastern (Fig. 1).What is surprising is that quite a few speakers from the Midlands think that they hear the accent of their own region (Fig. 2).Unfortunately, I had not expected this outcome and had, in this version of my questionnaire, not subdivided the heterogeneous dialect region of "the Midlands" any further.Only one speaker added in the comment box where exactly in the Midlands he was from when commenting on the speaker from Essex: Near to me; rural Birmingham/Solihull We are upper-middle/middle-middle class.We are very mobile.
Before jumping to any conclusions, I have to admit that these results have to be viewed with caution.With only one statement of this kind and such a small sample of informants it is far too early to make a firm statement.Nevertheless, it is an interesting detail in the context of this study since a sociolinguistic survey of Birmingham English conducted around the same time as the other dialect surveys with an interest in EE cited in this study reports another YES-BUT pattern: YES The speaker's realization of get his (= geɁ ɪz]) suggests that the influence of Estuary English is continuing to build […], BUT but the majority of phonological variables in Brummagen articulation discussed above shows no signs of abating and do not appear to be any less clearly defined in connected speech (Thorne 2003: 140).Irrespectively of the problem of the representativeness of the Midland results, they can be interpreted as confirming Kristiansen's hypothesis that "the whole cluster of salient contrasts that compose a lect can be metonymically evoked by the use of just one [or in our case a few] of its components" (2008: 71).In the case of the speaker-commentator cited above, the lect in question is that of mobile middlemiddle and upper-middle class speakers in Solihull.In his case, it is very probable that local "Brummagen articulation", including the FOOT-STRUT merger mentioned under 4.1.2., is not part of the accents of the people "near to" him and the stereotypes evoked by the EE feature pool is a social rather than a regional one.

Conclusion: the structure of the EE prototype category
Based on the results from the perceptual dialect survey obtained so far and the comments on EE analysed in the first half of this paper, I suggest the following structure of the EE prototype category within the regional dimension (see Fig. 4): The prototype is an "abstract set of attributes" (Taylor 2003: 64) holding a selection of off the shelf (Milroy 2007) and Southeastern variants which are able to metonymically evoke EE, at least in the eyes of lay speakers.The best examples of this category are Southeastern accents, especially those spoken in and near London.
The more we distance ourselves from this core in regional and linguistic terms the more peripheral the category members become.The Midlands region constitutes a candidate for more research (c.f.m.r.), East Anglia a candidate for future research (c.f.f.r.) since a few informants also assign the sample speakers to this area."Jockney", although a journalistic artefact, may either be categorized as outside the EE category or as a peripheral member belonging to two categories at the same time.It resembles Labov's "funny cup with a stem" (1973: 355) that he included in his experiment on the categorization of cups (see Fig. 3).As a doubtful candidate for membership in the EE prototype category, Jockney exemplifies the flexibility, even playfulness but also lack of rigid designation which makes the category attractive for lay people but "raises the hackles" of serious linguists.

South East
London Canterbury _____________ 5 YES […] What I DID say, in reply to the question whether I thought Scouse was changing, was that one of the debatable changes was the substitution of *th* in *think* and *brother* by *f* and *v*.[…] BUT […] I certainly DIDN'T claim that Scouse is about to disappear into the estuarine slime.[…] (author's emphasis).Jane Stuart-Smith (in Wells 1999) on Glaswegian: YES […] OK, it may adopt variable TH Fronting […] _____________ 8 BUTbut it is very questionable whether we should call that fast-spreading feature "Estuary"[…].YESIn terms of consonants -they do show l-vocalization […], and glottalling […],BUT but no h-dropping[…].In terms of vowels […] these seem to be pretty Scottish[…]  with maintenance of the Scottish Vowel Length Rule.[…]

_____________ 13 I
am very grateful to Jane Stuart-Smith and Sandra Jansen for facilitating data collection at their respective universities and to Peter Bennet, Darren Foster, Matthew Emery and Pascal Hohaus for asking and encouraging their families and friends to participate in the pilot of this survey.

14I
would like to thank Peter Trudgill and Peter Bennett for their very helpful comments on earlier drafts of the questionnaire.Any remaining infelicities and errors are my own responsibility.

Figure 4 .
Figure 4. Structure of the EE prototype category: regional variation.