Rapport de Stage - ENGLISH IN NEWPORT: A PHONOLOGICAL APPROACH TO H-DROPPING
Rapport de Stage - ENGLISH IN NEWPORT: A PHONOLOGICAL APPROACH TO H-DROPPING
Rapport de Stage - ENGLISH IN NEWPORT: A PHONOLOGICAL APPROACH TO H-DROPPING
Je tiens à remercier toutes les personnes qui m’ont aidé à réaliser ce rapport de stage.
Je souhaite adresser un grand merci en particulier :
• À Nicolas VIDEAU et à Sylvie HANOTE, pour toute la lecture qu’ils m’ont prêtée pour
aider avec la recherche lors de mon stage, pour tous les précieux conseils, et pour tous les
encouragements du début jusqu’à la fin qui m’ont permis de persévérer.
5 Conclusion 21
References 23
A Appendices i
A.1 Vowels and Lexical Sets of Welsh English . . . . . . . . . . . . . . . . . . . . i
A.1.1 Paulasto’s Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
A.1.2 Wells’ Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
A.1.3 Ferragne’s Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
A.2 Ramisch’s Map 2: Initial [h-] in the SED (25 items . . . . . . . . . . . . . . . ii
A.3 Text from the corpus: A Christmas Interview . . . . . . . . . . . . . . . . . . iii
A.4 Example of Annotations under Praat . . . . . . . . . . . . . . . . . . . . . . . iv
A.5 IPA transcriptions per informant (11 items) . . . . . . . . . . . . . . . . . . . iv
Introduction
For the third year of my degree, I am to intern in the field of my choice for a total
of forty-eight hours of work. As passionate as I am toward phonology and phonetics, it was
naturally an easy decision on where and on what I wanted to work. I got in contact with my
tutors (Mme Sylvie HANOTE, professor of English linguistics; Mr Nicolas VIDEAU, lecturer in
English linguistics as well, both of them oralists and both members of the FoReLLIS laboratory,
Team A - linguistics) and we decided to do my internship on English in Wales, as I have a
particular passion for variational linguistics, in particular on a phonetic and phonological level.
We decided this for three reasons. In my second year I had already done research on Shetlandic
English (Scots and Scottish English on the island of Shetland), so it felt right to continue with
this way of working. Since I am English myself, and my tutors research English, it wasn’t even
requestioned whether we would work on the English language or not. Finally, as I had done
research on a dialect of English with a Celtic influence, we thought it interesting to work on
the same kind of influence, and Wales came to mind–south east Wales in particular as it has
quite a bit of contact with English English. After a few hours of research reading, with the
need to look for an analysable corpus, Hanote got in contact with a researcher with whom she
was acquainted (Prof. Olivier GLAIN, Jean Monnet University of Saint Etienne, specialist in
sociolinguistics and variation) who had sent one of his Masters’ students (Julie PEYRACHE,
M2 student in English Studies of the same university as Prof. Olivier GLAIN) to Newport for
research on English in Newport for the PAC Project (Phonologie de l’Anglais Contemporain), a
project of which I will go in much more depth in 2.4. Thanks to this student’s audio recordings
on Newportonian English, I had the material I needed to start research. With the reading, I was
able to look for phenomena that occurred in the South West of Wales. H dropping was one of
the most interesting ones to research, as I have been interested in the phenomenon for a while as
I often picked up on it as an English Native speaker–in particular in close friend’s and family.
My mother, being one of them, would often tell me she was put down by her own mother for
dropping her aitches. We hoped to have enough time to research a second phenomenon but time
was precious and my research took more time than I had anticipated.
I wouldd like to mention, in case it was not clear, that, for this internship, instead of
helping my tutor’s with their personal research, we opted for a mini-thesis type work whereby
I do my own research on a matter that interests me and I am guided through my work with help
from my tutors. Naturally, I will be writing more on the matter in the following chapters.
1
LIS link (Laboratoire FoReLLIS 2024)1 . If the reader wishes to consult this website, I have put
it down as a footnote and, naturally, it is retrievable in the bibliography provided at the end of
the rapport. I also use information provided to me from the laboratory’s HCERES report, in
particular, the one led by S. Hanote in 2020 which provides information about the budget, the
housing of the laboratory, doctorates and more). This research was conducted at the FoReL-
LIS (Formes et Représentations en Linguistique, Littérature et dans les arts de l’Image et de la
Scène)2 Laboratory of Poitiers which is housed at the MSHS (Maison des Sciences de l’Homme
et de la Société).
The MSHS plays a crucial role by providing technical and logistical support to the re-
searchers of FoReLLIS. It also facilitates the search for partnerships and project submissions,
thus fostering research development. Furthermore, the MSHS contributes to structuring pro-
grams within the State-Region Planning Contract (CPER) by providing expertise and promot-
ing scientific cooperation among the Humanities, Economic and Social Sciences (SHES) labo-
ratories. This involves investments from laboratories in the structure, co-animation of priority
research axes, and co-organisation of interdisciplinary scientific events. By closely collaborat-
ing with researchers, the MSHS strengthens the connections between various stakeholders in
humanities and social sciences research.
In terms of budget execution, allocations are distributed across various areas. Thirty
percent of the budget is allocated to supporting research for both teaching staff and doctoral
students. Another 30% is designated for facilitating scientific activities within teams and or-
ganising scientific events. Fifteen percent of the budget is directed towards supporting funded
projects, maintaining the documentary fund, and assisting in the publication endeavours of lab-
oratory members, including contributions to Cahiers FoReLLIS, essentially an archive of some
of FoReLLIS’ projets3 , events and studies. Additionally, 15% of the budget is dedicated to or-
ganising thesis and HDR (Habilitation à Diriger des Recherches) defences. Finally, 10% of the
budget is reserved for the renewal of the computer infrastructure and the purchase of equipment
based on the needs of the teams.
The laboratory notably places significant emphasis on doctoral training and research,
with over 50 doctoral students enrolled annually, a figure that remains stable from year to year.
Various funding sources support these doctoral endeavours, including regional projects and gov-
ernment scholarships for international doctoral candidates. Doctoral candidates play a vital role
within the laboratory, actively participating in its activities, holding elected positions on the
council, being integral members of research teams, and collaborating by organising scientific
events and publications.
1
https://forellis.labo.univ-poitiers.fr
2
Translation: Forms and Representations in Linguistics, Literature, and Visual and Performing Arts
3
URL: https://cahiersforell.edel.univ-poitiers.fr/
2
1.2 Researchers and Research at FoReLLIS
The laboratory is comprised of two groups: Group A which consists of research on lin-
guistics and language didactics and Group B which consists of research on Literature and Visual
and Performing Arts. Inherently, my tutors (and so I) are part of Group A. As I did not work with
Group B, I will not be discussing it further. The manager of Group A is Mme Stéphanie Gobet, a
linguist who specialises in language acquisition and FSL (French Sign Language) and a lecturer
at the UFR LL (Faculty of Literature and Language of Poitiers). She is the person in charge of
overseeing the projects conducted in the Linguistics department of the laboratory. The linguists
of Group A are welcomed into the laboratory from the Faculty of Literature and Language of
Poitiers and the INSPÉ (Institut National Supérieur du Professorat et de l’Éducation)4 .
There are four axes of research in this linguistic branch, which are the following:
The Inter-language Contrastivity branch, supervised by Mme Raluca NITA, has a focus
on corpora-based contrastive linguistics, comparing between languages and/or various textual
genres. The researchers of this axis analyse different disciplines of language from graphemes
to co-text in the many disciplines of linguistics (i.e. Syntax, Discourse Analysis, etc.). The axis
benefits from four open bidirectional multilingual corpora resulting from national and interna-
tional institutional collaborations. These corpora are linked to the exploration of three genres
(journalistic, literary, scientific) in the PLECI and GRAFE corpora, and, since 2019, they have
been developing a research project focusing on the contribution of the inter-language and inter-
genre comparative approach to the study of the French pronoun on.
3
its linguistic realisation; Syntax and Prosody, which addresses a multidimensional approach
to linguistics, integrating syntax, semantics, prosody, and textuality, while analysing linguistic
markers in their context and their functioning orally; and finally Multimodality, in which a link
is made between visual media and linguistics. This includes sign language, gestures and the
visual aspect of writing systems. Four main corpora have also been created, one of which is
“Genre, Contraintes et Variations”, a collaboration with the University of Bordeaux-Montaigne.
Researchers often subscribe to different axes depending on the research they do, and
research can sometimes subscribe to two different axes. This is often the case with the Vari-
ational Axis whose work often intersects with the three other axes. Otherwise, projects may
be worked on by the whole team. These unifying projects enable research to be looked under
the scopes of different linguistic domains thus mobilising various language approaches and data
together. Recently, team A have begun two new projects together. The first pursuing a better
understanding of the processes of hesitation and adjustment in speech and discourse. Hesitation
being primarily a speech exclusive phenomenon while adjustment, according to FoReLLIS in
their website, “holds significance within the framework of the Predicative and Enunciative Op-
erations Theory, referring both to 'inter-subjective accommodation' and the malleability of forms
in context and co-text”. This projects aims to unify data from different linguistic perspectives
and continue one of the earlier main projects “Cohésion et Cohérence textuelle”. The second
project of the 2021-2025 period focuses on the access to and accessibility of data in a Dynamic
Corpus. The aim of this project is to render all their data and research Findable, Accessible,
Interoperable, and Reusable (F.A.I.R.). in accordance with the “Plan National pour la Science
Ouverte et la Recherche Reproductible5 ”.
4
logical inventories of south-east Wales in this subsection to bring my reading closer to Newport's
English. Naturally, I will be mentioning contrasts between different regions, in particular with
the vowels, as they have the most possible variations. That is to say, variation does exist with
consonants such as H dropping, consonant lengthening, (absence of) glottalisation and syllabic
consonants. For this section, I will be focusing on the inventories classed by Wells (1982b),
Paulasto et al. (2021), and Ferragne (2015). All the tables talked about in this section can be
found in the Appendices.
In the South (East) of Wales, four main differences arise compared to other regions of
Wales: the strut–Schwa Merger, the roundness of the nurse vowel, a socially varied quality of
the start vowel along with the bath vowel and the maintenance of diphthongs in face and goat
realisations.
While the strut–Schwa Merger is not exclusively found in the South East, it is marginally
more present (Paulasto et al. 2021). Paulasto transcribes the realisation of this vowel as [ʌ̈]; a
more central variation of /ʌ/, though Wells (1982b) maintains the /ә/ phoneme for both, as does
Ferragne (2015) (who cites Wells' transcription of such). This means that syntagms such as an
easy and unease become homophonous.
start vowels and bath vowels vary widely across Wales with a contrast between length
and front/back variants and, to some degree, openness [a ∼ aː ∼ ɑ ∼ ɑː ∼ æ ∼ ɛ] (Paulasto et
al. 2021). They are often intertwined together and with the trap and palm sets. In Cardiff,
Newport, and the surrounding areas, there seems to be a higher preference for /a/ and /ɑː/ in
these vowels, yet there exists an often stigmatised contrast between the worker class's set; often
preferring [a] or [æ] even sometimes [ɛ] in Cardiff, and the higher class's set; preferring an “RP
style” pronunciation with [ɑː].
What's more, the face and goat sets with monophthongal [eː] and [oː] as a standard
in Wales take on an RP-influenced diphthongal variation with [eɪ] and [ou] respectively in the
South East. Paulasto et al. (2021) notes that there can be a contrast depending on the spelling of
a word. Words with the combinations of letters <ai>, <ay>, <ei>, and <ey> fall under the stay
set with the diphthong [ei] with the monophthong falling under the face set, and spellings such
6
I will prefer using this term so as to not confuse readers between the Celtic language Welsh and the various
dialects of English spoken in Wales
5
as <ou>, <ow> and <ol> will fall under the snow set with the monophthong being pronounced
with the goat vowels. Compare daze /deːz/ with days /deɪz/, and no /noː/ with know /nou/. This
is why Paulasto et al. add these extra sets to their table.
The consonants of Welsh English are relatively identical to those of RP, but with three
added sounds borrowed from Cymraeg that are mainly only used in proper nouns such as town
names: /x/ as in Pentyrch, /ɬ/ as in Llangollen and voiceless /r̥/ as in Rhonda. These consonants
are sometimes even hard to pronounce for the Welsh who grew up in fully Anglicised regions
of Wales and only learnt Cymraeg in school (Wells 1982b), which means that these sounds are
often replaced by ones standard to English; /x/ will be swapped for /k/ ( /pɛntøːk/ ), /ɬ/ is swapped
for either /l/, /θl/, /fl/ or /kl/ ( /θlaŋgoθlәn/ ) and /r̥/ for voiced /r/ [ɾ ∼ ɹ] ( /rɒndә/ ). Apart from
these three consonants, we have: /p/, /b/, /t/, /d/, /k/, /g/, /tʃ/, /dʒ/, /f/, /v/, /θ/, /ð/, /s/, /z/, /ʃ/, /ʒ/,
/h/, /m/, /n/, /ŋ/, /l/ [l ∼ ɫ], /r/ [ɾ ∼ ɹ], /j/ and /w/.
The biggest difference between English English consonants and Welsh English conso-
nants is, as Wells (1982b) states, the length. Consonant gemination, while it is not commonplace
in South Wales, does appear in other regions more commonly where Anglicisation isn't as ap-
parent. For example, in the word nothing, the consonant (/θ/) in between two vowels (here, /ә/
and /ɪ/) becomes longer: /nәθːɪn/. Variation of this phenomenon has been attested where the
lengthening is realised as an unreleased stop followed by a glottal stop, such as in the word
meeting [miːt̚ʔɪn].
Outside of Cardiff and Newport, plosives in the Welsh accent also rarely have glottalised
variants, so fat will never be pronounced /faʔ/. Plosives also always release with as much aspi-
ration in the coda as in the onset. Words such as clock will consistently be pronounced [kʰlɒkʰ]
and not [kʰlɒk] or [kʰlɒk̚], like in Newport or Cardiff. The same happens with the word captain,
where the /p/ in the coda of the first syllable is fully aspirated. Compare Welsh [kʰapʰtʰɪn] and
Newport [kʰaptʰɪn].
In Cardiff and carried into Newport, glottalisation of /t/ appears before syllabic /l̩ / ( little
/lɪʔl̩ / ), and /t/ between vowels is sometimes realised as a tap /ɾ/ or voiced [t̬ ]. Speaking of /l/,
in Cardiff and Newport, there is a tendency to contrast clear [l] in the onset with dark [ɫ] in the
coda, similarly to RP. However, this is not the case with other areas of the South who often prefer
clear [l] in all positions or in the North where dark [ɫ] is preferred in all positions. (Wells 1982b)
Wells (1982b) also notes one of the most interesting phenomena for my research, H
dropping. H dropping mainly occurs in South East Wales and shares this with England. The
phenomenon has been mainly seen in the worker class and becomes less relevant in middle
class or careful speech. Of course, more will be said on this phenomenon in 2.3.
6
2.2 Sociohistorics of English in Wales
The sociohistorical and sociolinguistic dynamics of Wales and the influence of the
British culture are quite complex due to many factors that came into play, including (among
other subjects) the sharing of a boarder with England, colonisation periods, and century-deep
roots of interaction between the Welsh and English cultures. In English in Wales: Diversity,
Conflict, and Change, Colin H. Williams (1990) contributes with a chapter named "Anglicisation
in Wales", a chapter I will be referencing a lot. Wales has a rich history dating back thousands
of years, with its own distinct language, culture, and traditions. The Welsh language, Cymraeg,
is a Celtic language, and it has been spoken in Wales for over 1,500 years, making it an integral
part of Welsh identity and heritage. It is theorised that the British Isles first adopted Celtic cul-
ture and language from the continent due to its "popularity" and made it easier to trade tin in the
deposits in today's Brittany and the surroundings of today's Cornwall. In the first century and
up to the fifth, the Roman Empire settled in Great Britain (excluding Scotland) but only slightly
left a mark on the island's culture and language and began leaving between the fifth and around
the seventh century. During the same time, the Angles, Saxons and Jutes migrated to the east
of the island from Northern Europe, settling in today's England, while the Irish Celts migrated
from the west into Wales. This period of migration gave way to a very diverse early-medieval
Britain. Meanwhile, Wales (and Cornwall) stayed relatively Celtic in its demography. While
the native British accepted their new Germanic neighbours, they began speaking their language
also. The separation between Britain and its Brythonic neighbours of Wales was underway.
While this separation aided in the Welsh culture's vitality, it did somewhat cut Wales' access to
the rest of the world as an unhealthy relationship built against the border. I think most notably
of Offa's Dyke, which is theorised to have most likely been an attempt at separating the two
nations even further. However, a lot changed after the 11th century Norman invasion, whereby
Britain was conquered, and the Normans attempted invasion into Wales (Williams 1990). This
is when Anglicisation would begin. With Wales now part of England, (dubbed by Williams as
“the first colony of an expanding English state”), Henry VII passed the Laws in Wales Act in
the mid 16th century that included the Acts of Union which, as the name would imply, helped
“harmonise” the relations between Wales and England. The major issues with this act showed
the reality of unifying a more powerful civilisation with another. Williams writes it well with
the following paragraph:
Subsequent relations between the two peoples, the English and the Welsh, might be
categorised as a classic illustration of both the tensions and the harmony generated
when a dominant majority seeks to incorporate a subservient minority into its sphere
of influence, though of course the precise explanation for such a relationship has
as much to do with political interpretation as it has with the analysis of 'objective'
historical facts.
The process of Anglicisation, or the influence of the English language and culture, began
with the English nation beginning to exert its influence on the Welsh population. Over time,
English became the language of administration, law, commerce, education, social mobility, and
7
progress, particularly in urban areas and along the border regions; whereas Cymraeg (Welsh)
became associated with rural, traditional ways of life. This contrast between the two languages
naturally made English the preferred language in a progressing world, where most things were
treated with the English language. This linguistic shift from Welsh to English was not simply a
matter of linguistic dominance; it also had profound socio-cultural implications. For example,
Newport had a port with very easy access to the sea which favoured overseas commerce, and
a coal deposit that was essential for the era's advancement, and rapidly developed in the 19th
century.
Several factors have been attested to have significantly affected the linguistic landscape
of Wales and to have endangered Welsh as a language. A controversial Education Act of 1870
enforced English as the sole language of instruction and completely neglecting Welsh as a useful
language (Williams 1990). During the World Wars, English was promoted as the language of
unity and encouraged parents to not pass down Welsh to their children, completely losing Welsh
after a single generation. Finally, several debates have taken place about the importance of
in- and out-migration from rural to urban and vice versa, claiming that in-migration promoted,
in coalfields, many micro-cultures of Welsh-speaking communities, while out-migration from
urban areas “devastated the socio-cultural dynamic of rural communities” (Paulasto et al. 2021).
The substratum and superstratum languages play a significant role in understanding this
process. The Welsh language, as the original language of Wales, represents the substratum, the
linguistic foundation upon which subsequent layers of language influence are added. English, as
the dominant language of administration, commerce, etc., represents the superstratum, exerting
influence over the Welsh language and culture. The dominance of the superstratum does vary
from region to region, and, Pryce (1975) notes that, in and around the 18th, 19th and 20th
century, there was a division of the land into Welsh Wales (in the north and west), and Anglicised
Wales (in the east and south), mentioning also that there is a continuum of Welsh, to bilingual,
to English, for the Anglicisation process. The heartland or Inner Wales is how he names the
region in which Welsh was the main language spoken on a daily basis; this would include, for
example, Gwynedd. Then is the bilingual zone in which both Welsh and English were used
daily, but as two micro-cultures, of Welsh monoglots and English monoglots who coexisted in
the same region. Of course, the last were regions majorly "colonised" by the English, often of
the southeast border, where English was the majority, spoken daily.
Despite the pressures of the English superstratum, the Welsh language has persisted,
albeit with fluctuations in its vitality over the centuries. Considered by most as the biggest and
most effective effort, the translation of the Bible into Welsh in 1588 helped preserve the language
and culture, even though the initial intention of such a translation was to spread the Church across
Wales. In the 20th century, the Welsh Language Act of 1993 represented a significant milestone
in the recognition and promotion of the Welsh language, aiming to safeguard its future and ensure
its continued use and vitality in Wales. It was followed by further legislation, including the
Welsh Language Act 1997 and the Welsh Language (Wales) Measure of 2011, which continued
to strengthen the language's status and protection. Today, Wales is officially bilingual, with
8
both English and Welsh recognised as official languages. There is a renewed sense of pride and
identity associated with the Welsh language, reflected in its use in education, media, and public
life. However, the process of Anglicisation continues to shape Welsh society, albeit in more
subtle ways, reflecting the ongoing interplay between Welsh and English cultures in Wales.
H dropping occurs at different rates in Great Britain as not all dialects have this phe-
nomenon and some speakers will do it more consistently than others. There can also be variation
in production depending on age, gender and working class. Based on Ramisch (2010)'s Map 2
(at the end of the report as appendix A.2), East Anglia, the North and scattered around the
Southwest counties, there is a clear tendency to consistently produce [h-]. The omission of [h-]
is seemingly the most common geographically.
Wells (1982a) states, “H Dropping does appear to be the single most powerful pronun-
ciation shibboleth in England”; an interesting statement in and of itself. As he later describes, H
dropping appears to be a very powerful and stigmatised pronunciation, and I'm sure my mother
isn't one of just a few English who have been told to stop dropping their aitches. Whether it be
from school teachers or well-to-do parents, H dropping is the easiest way for a "high-profile"
person to put down a working-class other. This has been attested in sociolinguistic research, as
Wells writes about Trudgill's (1977a) and Hudson & Holloway's (1977) research, which indi-
cates a high percentage of H Dropping in the working class compared to the middle class.
When we talk about this phenomenon, it is important to note that only lexical verbs
9
Figure 1: Ramisch’s Map 2: Initial [h-] in the SED (25 items)
are counted, as the pronouns he, him, her, his; who; and the auxiliary verbs have, has, had do
exhibit aitch-less pronunciations even in Received Pronunciation. There is however a tendency
for a written <h> is more often pronounced due to anxiety-induced forced pronunciation and
hypercorrection7 .
In South-east Wales, H Dropping does occur (Wells 1982b) and as Ramisch (2010)
shows on his map, there seems to be a lesser percentage compared to other areas that they have
more contact with, for example, Bristol with a very low percentage of H-retaining. Nevertheless,
on the other side, Wales in general does not exhibit H Dropping.
10
of most (if not all) varieties of English across the world in an effort to obtain a more accurate
and full representation of the variation on a geographical, social and stylistic level. As Durand
et al. (2004) explain, the project also aims, among other objectives, at gaining better data to help
with teaching English, and testing different phonological and phonetic models from both syn-
chronic and diachronic points of view. The project involves gathering spoken English data from
different parts of the English-speaking world, using a standardised method: Researchers are re-
quired to give (but are not limited to) 4 reading tasks, which consist of a wordlist prioritising
vowel analysis in isolated environments, a wordlist prioritising consonant analysis in isolated
environments, a list of sentences, and a two-page text called A Christmas Interview configured
to showcase the ways that words link together in continuous speech, all to be read aloud. There
are also 2 spontaneous speech conversations; the first being between the informant and the re-
searcher as a (semi-guided) interview whereby the researcher asks questions often relating to
work, civil identity, and language, and the second being an informal "chit-chat" between two or
three informants. Durand likes to point out that, past the three-person maximum, it is near im-
possible to unravel or discern any worthy material. While he does not give a descriptive reason,
I like to believe groups would naturally form and two conversations would be recorded at the
same time and, in so being, would be impossible to analyse.
Since the programme's main goal is to homogenise analysis procedures and have the
same type of material for all dialects and accents analysed, it is natural for there to be many
transcription conventions put in place to have a more cohesive analysis procedure for all re-
search. I will not list all the conventions, but they are all available on the PAC Programme
website mentioned above and listed at the end, in the bibliography. These vary from the use of
punctuation markers (using the comma for small pauses, the final stop for long pauses, the ques-
tion mark for questions), the transcription of truncation and repetition, to how to mark laughter
and hesitation, and even how to transcribe acronyms and dialectal expressions. All speech an-
notations and analyses are done under Praat; a computer software package for speech analysis
in phonetics. I have a subsection dedicated to my own experience with Praat, so I will write
more about it in 3.2.
As the project grew, researchers and students alike from varying fields of expertise in
linguistics got to know the programme, which developed further into five different research
groups. PAC-LVTI is a protocol developed in 2011 that aims at researching the sociolinguistics
of urban and post-urban contexts. This is now the mainline protocol to be followed by all re-
search done within PAC due to its concise questionnaire relating to the many analysable factors
affecting a dialect and/or accent (i.e. language, city, work and identity). Other protocols will
most often add an extension of material if needed. IPCE-IPAC is a protocol coordinated at the
University of Lille by Paolo Mairano and Caroline Bouzon whose aim is to gather and provide
data on the acquisition and learning of English as a foreign or second language and to compare
it with L1 speakers with different varieties of English. This data can provide better analysis ma-
terial for psychologists, phonologists and didacticians alike. The protocol is inspired by PAC's
protocol and that of IPFC (Interphonology of Contemporary French; PFC being the French vari-
11
ety of the programme by which PAC is inspired), and consists of two wordlists to be read aloud,
the Christmas Interview text, a read-aloud task of sentences explicitly targeted towards eliciting
different intonation patterns, and a formal conversation with a teacher. The protocol is inter-
twined with the PAC-ToE (Teaching of English) project, which “aims at linking theoretical and
methodological research interests with pedagogical commitments and didactic implementations
in the field of spoken English Teaching & Learning” (PAC Programme Website 2024). The
PAC-ToE protocol has, among other research projects, developed PICL! (Phonologie Incarnée
de l'Anglais au Collège); a project that aims at 1) analysing the production and perception of
spoken English in French non-native learners, and 2) evaluating how confident a student feels in
their ability to speak English, what motivates them when it comes to speaking and pronouncing
correctly, and planning educational materials or lessons related to such data. The corpus was
created with 120 pupils in a French Secondary School from 2019 to 2023. The PAC-Prosody
Protocol adds an extension of reading material with a task where informants are to describe
an image. The goal of these tasks in this protocol is to analyse the rhythm and intonation of
varieties of English at the interface of with acoustics and syntax. The project has also begun
building tools for the semi-automatic annotation of prosody, a feat that is in and of itself very
important and fascinating. The PAC-Syntax is linked to the latter protocol as its aim is to study
the structure of spontaneous oral speech through syntax, morphology, semantics and prosody.
The reason I am writing so much about the PAC Project is because my research benefits
a lot from the protocols of such a programme. H dropping, as I have written in 2.3, is widespread
in England, but little research that I could find and was accessible to me could help me with the
analysis of the morphophonological reasons and sociolinguistic reasons as to why H dropping
occurs or not. The PAC protocol has helped me greatly, thanks to a standardised method of
research that works almost perfectly in my favour with my analysis. All my research is thanks to
the corpus built by Mme Julie PEYRACHE, the student under Prof. Olivier GLAIN's direction,
a corpus following the PAC-LVTI protocol with Newportonian informants varying in age and
gender. There is a better place for all this in 3.1, so the reader may skip to said section for more
development on the corpus I used and how I used it.
12
ing why. Mme CARATINI also gave very insightful classes on phonetics and to some extent
phonology. I found that my prior familiarity with phonetics and the International Phonetic Al-
phabet (IPA) from my former hobbies gave me a head start in my phonetics classes. However, it
was when we delved into acoustic phonetics and began exploring spectrograms that I truly felt
a sense of fascination and encountered something entirely new. Being here now, applying the
knowledge I have accumulated over the years, I finally feel that what I have been learning and
what I have been passionate about for a while is truly useful.
After doing the literature reading, I found that Southeast Wales was the only area that
attests for H dropping in Wales. As a self-diagnosed H dropper, I really wanted to look more
into it. So, Mme HANOTE, got in contact with M GLAIN, and I was able to acquire a corpus
on the Newport accent. I initially had just one question: how often does H dropping actually
occur in Newport? But as I followed through with the several tasks, I began asking myself many
more questions and coming up with a few hypotheses. Does anything make H dropping more
common? Does anything force H retention? Maybe the /h/ could assimilate with preceding
consonants, giving the impression of H dropping. How can reading affect the pronunciation of
/h/? I also began asking myself why many of the informants had such different accents, and,
before listening to the interviews, I assumed it had to do with age, where the younger speakers
would have an accent closer to English English, while the older speakers would have more
standard or stereotypical Welsh English as an accent. All this begun thanks to the provided
corpus.
3.1 Corpus
Julie PEYRACHE, M2 student in English Studies at Jean Monnet University of Saint
Etienne, under the supervision of Prof. Olivier GLAIN, travelled to Newport, Wales, to collect
a corpus of the Newport accent under the PAC-LVTI protocol. She chose 11 Newport citizens, 5
male informants and 6 female informants of varying ages. In total, informants had to read aloud
2 wordlists, one text (A Christmas Interview), a collection of sentences and answer questions in
a formal interview where Peyrache asked about work, citizenship and language. Peyrache was
kind enough to send me her corpus, and, if not for her, I would most likely still be looking for a
usable corpus on the Newport accent to this day.
Next, I intend to outline the various advantages and disadvantages inherent within the
corpus, giving further explanation as to why some things benefitted my research, and why some
others did not. As for the main advantage of using this corpus. The PAC protocol benefits the
corpus by getting consistent phonological data from each informant who has produced the same
amount of analysable data from the reading tasks (of course, in a perfect world where nobody
stutters, misreads, skips or repeats words/sentences). The formal interview task is also very use-
ful as it helps with getting a better sociolinguistic image of the informant's speech. For example,
a few of the informants had previously lived in a certain area other than Newport, or worked
with many people from outside of South Wales, which are factors that can alter the informant's
accent. The questions on language also really helped to assess whether the speaker was proud
13
of the Newport accent, or (un)consciously stigmatises the Newport accent. The interview was
absolutely crucial for my sociolinguistic approach to the analyses.
As for the main disadvantages of the corpus, the reading tasks can unfortunately back-
fire, especially due to the written side of things. This is particularly troublesome for my re-
search on H dropping, as the presence of a written 'H' may compel speakers to pronounce the
/h/ phoneme. This is particularly true for phrases such as a high-profile personality in which
the determiner a preceding the <h> has a very high probability of forcing the pronunciation of
/h/ as the English language generally does not like vowel contact of non-high vowels (/i/ or /u/).
Consequently, many speakers who do drop their aitches will often times turn a into an (as I have
explained in 2.3). Reading is also variably easier or more difficult depending on the person. An
informant who has more difficulty with reading will stutter much more, and repeat sentences or
words. Their confidence in their reading capabilities can also stress them out and make reading
even more difficult. On the other hand, informant's with a lot of confidence can completely
misread or skip lines without even realising it, as I've noticed in the tasks after annotating them.
Another downside is due to Newport's “downfall”. As jobs become more and more difficult to
find in Newport, many Newportonians look for jobs elsewhere and never stick around other cit-
izens who share their accent. This means that finding “true Newportonians” would be a difficult
task to pull off.
Unfortunately, when I received the corpus, the .wav files of 3 informants were corrupted,
and I was unable to analyse them with Praat. I could have passed the files through Audacity and
filtered out the background noise, but this would have taken too much of the time I had for my
internship. Instead, I preferred to continue with the files I had of the 8 remaining informants,
and, while this made the amount of informants an equal number, 3 were male, the other 5 were
female, which might have been an altering factor in my results. Nevertheless, I pressed on.
For my research, I needed to locate all the words in the corpus that were pronounced with /h/,
whether it be in initial position of the word or medial position. There were 8 words in total
in the wordlists and 52 words in the text. I thought it best not to analyse the interview, as the
interviewer's words would sometimes overlap the interviewee's and the data was going to be
inconsistently disproportionate anyway. The /h/ words in the wordlists were all pronounced
with an /h/ by all the informants, so I didn't even find it necessary to annotate them, especially
if they were isolated words, which most likely induced the intention to pronounce the /h/ even
if the informant had strong H dropping. In the 52 words of the text, there were a huge amount
of non-lexical words (i.e. neither nouns, verbs, adjectives nor adverbs). 24 words were he or
any derived words (him, himself, etc.), 12 were the auxiliary verb have in its possible forms
(has, had, have) and there were two instances of who. All these words can be pronounced by
dropping the H even in RP (Wells 1982b), which is why Mme HANOTE and I chose to remove
them from the analysis. The words I did end up analysing were the following:
• behaviour in “He didn't mean: 'be on your best behaviour and be happy all the time'.”
14
• happy in “He didn't mean: 'be on your best behaviour and be happy all the time'.”
• high(-profile) in “Peter Smith looks exactly how you'd expect a high-profile television
personality to look [...]”
• whole in “During the whole interview, Peter Smith stressed the need to be on the side of
the poor and the needy.”
• hundred in “[...] and which it is generally agreed employs eight hundred staff and runs
schools, hospitals and hostels around the world.”
• hospitals in “[...] and which it is generally agreed employs eight hundred staff and runs
schools, hospitals and hostels around the world.”
• hostels in “[...] and which it is generally agreed employs eight hundred staff and runs
schools, hospitals and hostels around the world.”
• human in “He just sighed in answer to my remarks and said: 'I'm only human, John'.”
It was especially exciting to find out how the informants would pronounce these words
for particular reasons. For one, it was going to be interesting to see how the H in behave and
behaviour would be pronounced, whether they be both pronounced, neither, or one or the other.
Some lesser exciting results, for me, were going to be handsome, hospitals and hostels, for I hy-
pothesised that enumerations would give way to a stronger chance of the H being pronounced, as
the words can easily be isolated between pauses, especially in reading. However, my hypotheses
had yet to wait for results, as I was first tasked to annotate under Praat.
I had learned to use Praat in one of my Phonetics classes at the university, so I was
already a little familiar with it, but after two years of not using the software, I felt my knowledge
was a little rusty. So, I decided to have a little fiddle about for an afternoon in an office at
15
FoReLLIS, and I quickly realised that, after what felt like an hour, three had already gone by.
This is what made me realise just how time-consuming Praat annotations can be, especially if
you are new to the program and/or are a perfectionist. However time-consuming it may be, it
is actually quite entertaining and was probably one of the easiest tasks I had to do during my
internship.
The annotations were done, as is possible with Praat, with three floors of annotation. On
the first, I annotated the sentence or phrase with the same conventions used in the PAC Project
but with a few liberties in the usage of the comma and final stop (i.e. hardly ever using the stop as
I only annotated certain sentences, rather than the entirety of the file). On the second, I sectioned
the words, and on the third, I wrote the sounds with the International Phonetic Alphabet. I
transcribed the sounds to the best of my abilities to the actual phonetic pronunciation, though I
had sometimes used the equivalent phoneme if I was unsure of the correctness of my phonetic
transcriptions. This was especially the case with vowels, where I found it difficult to discern
whether a close front vowel was /ɪ/ or /i/, or whether a glide was produced or not between vowels.
I chose not to delve into too much precision vis-à-vis the realisation of consonants or vowels
(for example whether a consonant was fully released or not, or amount of aspiration), as it was
not important for my research, would have cluttered the annotations and I would not have had
time anyway. So, fortis stops are written without the aspiration. For a better visualisation of the
transcriptions and annotations, here is an example that the reader can also find in Appendix A.4.
In this audio, the phrase is: “in which he describes himself”, the /n/ of in assimilates with the
following labial-velar glide [w], making the realisation [m]. Further on, we find that himself is
completely reduced to a single syllabic consonant [m̩]. We also notice that /h/ is not pronounced.
I believe I struggled the most with getting to know the system, as the software often
crashed when I least expected it to, and often forgot to save the TextGrid (the file with the
annotations on it) to the same place on my computer. This would mean that I would often find
different annotations on different files and would have to copy manually the annotations back
16
onto the original file. There may have been an easier automated way of doing this, but I did not
know (and still do not). My tutors helped quite a bit with my annotations in moments when I
had no clue how to correctly transcribe an utterance. The most prominent example would be my
problem with separating vowel sounds, as they would often glide together, and I could not find
the "perfect" division point. Looking back, I realise I was picking at hairs for no good reason,
as my research was not going to focus on said division anyway.
What I tried to focus the most on was the realisation of /h/, where I found that more than
two realisations were in fact possible. [h] and zero were the most obvious; however, I found a
palatalised realisation [ç] in human quite often, and a voiced variant of the glottal fricative: [ɦ].
This one was much more difficult to stop, as, most of the time, it was uncertain whether the pitch
line showed up due to the actual utterance or due to background noise (a fan in the background,
for example). Even after analysis, I am uncertain of a few being [h] or [ɦ]. In any case, the H
would not be considered to be dropped; just weakened. For this reason, I kept the transcription
of [h] for both.
17
4 Results and Analyses
4.1 Results
The amount of aitches being pronounced is actually much lower than I initially expected,
with only around 13.6% of H dropping across all the words produced. The average of H dropping
per person is merely 1375%, noting that only 50% of speakers dropped at least one H, and the
average of words pronounced without an H from H droppers is approximately 28%, with the
lowest rate at 17% and the highest at 36%. The informants produced at least one instance of H
dropping are BC, DF, GS and HP. HP is the only female informant who drops H. The informants
who always pronounce H are CR, KW, LH and ML, who, interestingly, are all female. BC
pronounces the least amount of /h/, with 4 instances of H dropping., in the words behaviour,
whole, hundred and human.
Interestingly, behaviour, the word with the most H dropping (with four instances), seems
to be the word that all H droppers produce without an /h/. Behave, happy, whole, hundred and
human are all the other words that have at least one instance of H dropping; behave, happy and
whole having only one instance (12.5%), hundred having three (37.5%), and human having two
(25%).
In table 1, the realisations with H dropping are highlighted in green, while the realisa-
tions with [h] are highlighted in orange.
There were a few hiccoughs in the analyses as there were many factors that affected the
final results. Firstly, informants would sometimes struggle with the reading task. This meant
that they would occasionally misread words, skip words or entire lines, or stutter on the /h/ words
I was analysing. HP accidentally misread hostels as hotels, and, while I still had a /h/ word, the
following vowel was not the same and stress was on a different syllable. There were times
when words would be skipped, like DF skipping happy, or ML who forgot hostels. Meanwhile,
some informants skipped a line and the sentence ended up making no sense, so they repeated the
sentence in a whole. This meant that I had extra data from a few informants. In particular, CR
repeated whole, which posed no problem, it was when GS pronounced once as [hәndɹәd] and
once without the [h]: [әndɹә]. While at first, I found it troublesome for the statistics, I realised
18
how interesting it was for him to pronounce the second time with H dropping. I would have
expected more for it to be the other way around, with the correction/repetition to be pronounce
with /h/ as though to pronounce it "correctly". In actual fact, once he understood the sentence,
he flowed through the words as usual. This could indicate that the fact that he was reading made
it, for him, less natural and this might have forced the /h/ the first time. As the reader may notice
in table 1, the realisation of behaviour was occasionally realised without the glide in the third
syllable: [bihɛɪvә(ɹ)]. I initially thought it was just a mispronunciation due to slight stuttering,
but it happened with 3 informants, which means there's a high chance it is not just a coincidence
and is potentially part of the Newport dialect just like how programme is pronounced with a
schwa in the second syllable in many dialects of Wales. However, I cannot find any evidence of
such a phenomenon in any of my reading.
As I have already mentioned, about half of the informants have had a relatively heavy
linguistic influence from other areas, whether it be working with a lot of people from the Valleys
like BC, living for a time in Bristol like LH or moving to Manchester for university like CR,
most informants had something that made their linguistic background different. That is to say,
living in another area does not obligatorily mean someone's will change. It is, however, still
something to keep under consideration during the analyses.
There also seem to be phrases in the text that could potentially have forced H retention.
For example, high follows a and the English language often dislikes the contact between vowels
if one of the is not a high vowel /i/ or /u/.
4.2 Analyses
4.2.1 Morphophonological Analysis
The phonotactics of words can often reveal quite a lot when dealing with odd phenomena
that do not appear consistently such as H dropping. I intend to analyse the morphophonological
components of the 11 words and study how sounds and morphemes articulate together with /h/
through various rules. These include how vowels react when in contact with [h]; whether pauses
can force H retention; and whether or not stress has anything to do with the pronunciation of [h].
This will prove to be somewhat complicated as I have only 12 instances of H dropping which is
a measly number of occurrences to be able to accurately analyse.
There is also something interesting that happens with all the word that have been H
dropped, apart from hundred. In medial position, behave becomes [bieɪv] and behaviour be-
comes [bieɪvjә(r)], and in initial position, be happy was realised as [biˈaːpi], the whole was
realised as [ðiˈɔɫ], and only human was realised as [әʊnliˈjumәn] or [әʊnliˈjumn̩]. In each case,
when the [h] was dropped, either a preceding vowel was a high front [i] or a schwa reverted
back to [i] (/ðәˈhɔʊɫ/ → [ðiˈɔɫ]). Interestingly, as Ramisch (2010) also points out, H dropping
has also been found to replace [h] with a high front glide [j]. This much is obvious in the typical
Newport slang sentence “come by yer” (yer = here). At first, I analysed this as being a higher
chance of H dropping between a preceding vowel and following long vowel or diphthong. As
19
I've mentioned above, the English language does not like hiatus9 except if the first vowel is a
high one, most likely due to their easier gliding capabilities. While it has been proved that H
dropping does occur even when exposing vowels to hiatus (see Wells (1982a), chapter on H
Dropping), it could be an oddity to some H droppers who do not constantly drop H, making H
retention prominent in vowels that are not or cannot revert to a high vowel.
The glottal fricative seems to also be consistently retained after a pause. This is seen
with handsome, which is in an enumeration, which, especially while reading, isolates the word
between two pauses, and, as I have noticed with the wordlists provided in the corpus, isolated
/h/ words seem to consistently retain /h/. The same can be said for hospitals and hostels.
Frequency of usage of words could explain some higher rates of H dropping in some
words compared to others. Hundred is a good example for this compared to hostile which are
both exposed to a preceding plosive. In “employs eight hundred staff”, hundred's H could easily
be lost before the eight's final consonant due to the very common usage of such terms together.
That being said, emphasis could also be a factor, as emphasis is but on eight rather than hun-
dred, and, in “tried not to sound hostile”, hostile is given more emphasis in the word. With so
little instances, it is hard to come up with a definite theory. While stress would seem to be a
solid hypothesis at first, it is quickly debunked, once one notices that every word analysed and
annotated has stress of the syllable with /h/.
While I have spoken a lot about the linguistic environment that can interfere with one's
accent, I have not gone into too much detail because I intended to elaborate in this section.
H dropping, as demonstrated by, appears in almost all of England, as well as in Cardiff and
Newport. However, it is not a phenomenon common to the rest of Wales. For someone who
worked all his life around Valleys people, BC still drops his aitches quite a lot. In his interview,
he explained that he felt more welsh than Newportonian, believed his accent was closer to the
Valleys accent than the Newport accent and believed that learning Welsh had also influenced
his accent as well. This may be true to some extent as he does lengthen his consonants between
vowels for example, but none of these factors removed his H dropping. BC does, however, seem
to be carefree about his accent. When asked about examples of how the Newport accent sounds
to him, he is unable to give any and says that the accent is simply natural to him as he has never
had to compare it with anything. This could indicate that he has never been one to "correct" the
way he speaks, so H dropping could be a feature of his accent that he never realised was there,
even as he worked with people having a different accent where H dropping is not present. After
all, I doubt H dropping is an easy-to-spot difference if you have never been stigmatised by it. In
accordance with what Wells (1982a) notes, /h/ could be an optional sound to make in his mental
lexicon.
CR often claimed in her interview that she often spoke “properly” when not with her
9
hiatus or diaeresis refers to two vowel sounds occurring in adjacent syllables, with no intervening consonant.
20
family or friends, stating that she has different accents depending on who she is around and where
she is (home, work, professional context). This much was clear when she read the text with a
close-to-standard English English accent. She often spoke somewhat derogatorily about her
actual Newport accent, often saying that she sounds very “chavvy”, a harsh slang term meaning
a young person of a type characterised by coarse and brash behaviour (with connotations of low
social status). This would indicate the H retention in her reading task. She had also moved to
Manchester for university, but other than that, she has remained in Newport ever since she was
born there.
DF is retired, residing in Riscar but had extensively traveled throughout Britain and
the continent for work. Despite his diverse experiences, David's urban accent often prompted
remarks from locals in Riscar, who perceived it as characteristic of a "townie." Notably, DF con-
trasts the Newport accent, exemplified by the phrase "I talks like this, don'I," which he identifies
as distinct from his Riscar accent. He stated that the Newport accent is regarded as "weird" and
faces social stigma, with British society often looking down upon it. What's more, DF noted
a linguistic shift when speaking Welsh, wherein his accent becomes more Welsh than English,
highlighting the complex interplay between regional identity and language.
GS, who lives in the vicinity of Newport but not directly in the city itself identified
his accent with what he terms "south Welsh English,". He illustrated the Newport accent with
phrases like "mornin's a good day, innit," highlighting distinctive linguistic features associated
with the area. Notably, he brought up the fact that the Welsh language had no interplay with
English in Southeast Wales, which demonstrates the dominance of English in the area and why
H dropping seeped in from England. He also talked about his father's disdain for expressions
like "I talks, I wants," indicating the attitude some people may have toward language variation.
Interestingly, GS also emphasised that he preferred when people understood so he would rather
"tone down" his accent, so to speak, for a better communication. He wanted to emphasise that
he does not do this because he is ashamed of his accent.
HP, most notably, stated that the Newport accent was one in which you spoke very fast,
and stumbled into words and later on says she finds the Newport accent to be a bit "stupid"
and believed it made people underestimate Newportonians for that very reason. LH gave an
imitation of the Newport accent in which no Hs were pronounced: “alrigh', 'iya, 'owzi' goin'”.
This example shows that H dropping is a stereotype of the Newport accent, or at the very least
it is one for LH. She would go on the stigmatise the accent with many remarks which clearly
showed her view on the accent. For the most part, the Newport accent is heavily stigmatised, at
least from what is reported from the 8 informants.
5 Conclusion
In conclusion, H dropping does occur in Newport, but it is not so common or consistent
read aloud. The Newport accent is seemingly viewed with a lot of stigma with H dropping being
21
one of the phenomena avoided by Newportonians. H dropping, while it is common in almost all
of England and extensively researched in sociolinguistics, it is a phenomenon often looked over
on a morphophonological level. While its inconsistency in its production may feel hazardous, it
would be great to see the phenomenon analysed with a much larger corpus that expands across
multiple dialects.
This internship has truly helped me in so many ways, and I did not think it was possible.
With a plan to become a phonetician after my studies, I was able to fully immerse myself in a
project that felt uniquely personal to me and for my future. I realised how tedious, knit-picky
and time-consuming research can actually be, especially in phonetics and phonology, but also
realised just how passionately interesting it can be and feed one's curiosity.
If I had double the time for my research, I would have loved to also focus on the realisa-
tion of syllabic consonants. This interest was especially sparked by CR's realisation of himself
as [m̩seɫf]. The elision of the unstressed vowel really peaked my interest and had me thinking
about the elision of unstressed vowels and the epenthesis of schwa with the syllabic consonants
/m̩/, /n̩/ and /l̩ /. I highly recommend looking into research done on said phenomenon.
22
References
Boersma, Paul and David Weenink (Nov. 21, 2022). Praat: doing phonetics by computer [Com-
puter program]. Version 6.3.01. URL: http://www.praat.org/.
Durand, Jacques, Philip Carr, and Monika Pukli (2004). “The PAC project : Principles and Meth-
ods”. In: La Tribune Internationale des Langues Vivantes (TILV), English Pronunciation
: Accents and Variation 36, pp. 24–35.
Ferragne, Emmanuel (2015). “Le Pays de Galles”. In: La Prononciation de l’Anglais Contem-
porain dans le Monde, pp. 215–227.
Laboratoire FoReLLIS (2024). FoReLLIS. URL: https://forellis.labo.univ-poitiers.
fr/ (visited on 04/11/2024).
PAC Programme Website (2024). URL: https : / / www . pacprogramme . net/ (visited on
04/11/2024).
Paulasto, Heli, Robert J. Penhallurick, and Benjamin A. Jones (2021). Welsh English. Dialects
of English volume 12. OCLC: on1062274855. Boston: De Gruyter Mouton. 263 pp.
ISBN: 978-1-61451-381-0.
Pryce, W. T. R. (1975). “Migration and the Evolution of Culture Areas: Cultural and Linguis-
tic Frontiers in North-East Wales, 1750 and 1851”. In: Transactions of the Institute of
British Geographers 65, pp. 79–107.
Ramisch, Heinrich (2010). “Analysing Linguistic Atlas Data: The (Socio-) Linguistic Context
of H-Dropping”. In: Dialectologia (Special Issue), pp. 175–184. ISSN: 2013-22477.
Wells, J. C. (1982a). Accents of English 1 An Introduction. Cambridge University Press.
— (1982b). “Accents of English 2 The British Isles”. In: Cambridge University Press,
pp. 377–393.
Williams, Colin H. (1990). “Anglicisation in Wales”. In: English in Wales: Diversity, Conflict
and Change. Multilingual Matters, pp. 19–47.
23
A Appendices
Table 2: Paulasto’s et al. (2021) distribution of Welsh English vowel phonemes in the standard
lexical sets
keyword realisation keyword realisation keyword realisation
kit ɪ goat oː cure ɪuwә ∼ uә >ɔː
dress ɛ snow ou power auwә ∼ auә
trap a palm aː fire aijә
lot ɒ thought ɔː near iә
strut ʌ̈ goose uː ears iә ∼ œː
one ʌ̈ ∼ ɒ price ai tuesday ɪu
foot ʊ choice ɔi happy i
bath a ∼ aː mouth au letter ә
cloth ɒ start aː horses ɪ
nurse œː north ɔː comma ә
fleece iː force ɔː started ɪ
face eː boar ɔː morning ɪ
stay ei square ɛː
Table 3: Wells’ (1982b) distribution of South East Welsh English vowel phonemes in the stan-
dard lexical sets
keyword realisation keyword realisation keyword realisation
kit ɪ goat ou cure uːә
dress ɛ
trap a palm aː
lot ɒ thought ɔː near jɜː ∼ iːә
strut ә goose uː
price әi
foot ʊ choice ɔi happy i
bath a ∼ aː mouth әu letter ә
cloth ɒ ∼ ɔː start aː
nurse ɜː north ɔː comma ә
fleece iː force ɔː
face ei
square ɛː
i
Table 4: Ferragne’s (2015) distribution of Welsh English vowel phonemes in the standard lexical
sets
keyword realisation keyword realisation keyword realisation
kit ɪ goat ou ∼ ɤu ∼ oː cure juːә ∼ jɔː ∼ jʌː
dress ɛ
trap a∼æ∼ɛ palm aː ∼ ɑː
thought ɔː ∼ ʌː near jɜː ∼ iːә
strut ә goose uː
price әi
foot ʊ∼ɤ choice ɔi ∼ ʌi happy i
bath a mouth әu ∼ ʌu letter ә
cloth ɒ∼ɑ start aː ∼ æː ∼ ɛː ∼ ɑː horses ɪ
nurse øː ∼ œː ∼ әː north ɔː ∼ ʌː comma ә
fleece iː force ɔː ∼ ʌː
face ei ∼ eː
square ɛː
ii
A.3 Text from the corpus: A Christmas Interview
© PAC Project
If television evangelists are anything like the rest of us, all they really want to do in
Christmas week is snap at their families, criticize their friends and make their neighbours' chil-
dren cry by glaring at them over the garden fence. Yet society expects them to be as jovial and
beaming as they are for the other fifty-one weeks of the year. If anything, more so.
Take the Reverend Peter 'Pete' Smith, the 'TV vicar' who sends out press releases in
which he describes himself as 'the man who has captured the spirit of the age'. Before our 9
a.m. meeting at his 'media office' on Crawshaw Avenue, South London, he faced, he says, a real
dilemma. Should he make an effort 'to behave like a Christian' – throw his door open, offer me
a cup of tea – or should he just play it cool, study his fingernails in a manner that showed bored
indifference and get rid of me as quickly as possible? In the end, he did neither.
As a matter of fact, John,' he says in a loud Estuary English twang, 'St Francis said, “At
all times preach the gospel and speak whenever you have to.” But hey, he didn't mean “Be on
your best behaviour and be happy all the time.” I could have been extra-polite to you, but the
real me would have come out as I was talking. You cannot disguise what you are.'
'Well, I'm a Christian, John. I've been one since I was 14. And I know for sure that
Christianity will be judged more on who you are rather than what you have to say about it.
Many church leaders don't appear to understand this. They think we can only be really Christian
when we are ramming the doctrine of the Creation down people's throats. But if you try to force-
feed people they get sick of it and think you're a pain. It's seen as the job of a Christian leader to
wear a dog-collar and dress in purple and always be talking about the real meaning of the New
Testament. In reality, that turns people right off!'
In many ways, 'Pete' Smith looks exactly how you'd expect a high- profile, born-again
Christian to look: tall, handsome, clean-cut and evenly sun-tanned. He has those scarily white
teeth that TV evangelists tend to have, and he doesn't wear a dog-collar. In fact, when doing his
various religious programmes on Sunday mornings, he has been known to wear a black leather
jacket instead, in casual mode. Today, the look is more business-like: metal-rimmed glasses, a
grey suit, a blue open-neck shirt, and fashionable black shoes with large buckles. Smith is 44
but he looks a mere 24.
During the whole interview, there wasn't any talk of the poor or the needy but only of his
forthcoming trip to China in February and the masses waiting for his message there. I ventured a
few questions relating to the charity trust he founded some ten years ago and which, it is generally
agreed, employs eight hundred staff and runs schools, hospitals and hostels around the world.
And what about the gambling organization he has been willing to advise? Is that a temporary
activity or might it be true that he has accepted to be paid to sit on its Board of Directors? Which
iii
side is religion on these days? Does money matter? It was as if I had launched a few missiles
in his direction. He just sighed in answer: 'I'm only human, John. God knows I do my best and
often fail, But it'ss no skin off my nose if our enemies sneer at some of the good work we do.
Truth will out.'
iv