"Whit da heck is 'strut dee lug'?": Evidence for Accent-levelling in the Shetland Isles in an...
Transcript of "Whit da heck is 'strut dee lug'?": Evidence for Accent-levelling in the Shetland Isles in an...
“Whit da heck is ‘strut dee lug’?”:
evidence for accent-levelling in the
Shetland Isles in an intergenerational
analysis of variable pre-plosive vowel
mutation
William James Brown
ABSTRACT
In this paper I will be looking at whether the Shetland accent is
changing. The Shetland Isles have been documented as undergoing a
rapid shift from the very distinct regional dialect towards a hybrid
variety consisting of the dialect and Scottish Standard English. While
models put forward by Trudgill (1984) explain such hybridization in
terms of language contact and dialect-levelling I have found that
previous research hasn’t made a clear divide between dialect-levelling
at the lexical-syntactic level and accent-levelling at the
phonological, preferring to treat both as part of the dialect-
levelling model. To demonstrate that such accent-levelling occurs I
will be examining how Soft Mutation, a form of vowel allophony unique
to the Shetland Isles, behaved before recent immigration to the Isles
in the 1970s largely responsible for the dilution of the dialect and
after population mixing which accompanied immigration. I have employed
Page 1 of 48
an acoustic analysis study of Shetland and Scottish pseudowords to
examine how Soft Mutation behaves in both dialects and whether this
indicates bidialectalism or simply style-shifting in the Isles. My
prediction is that Soft Mutation is a feature of the dialect which
shouldn’t be applied by younger speakers in Scottish words, who have
the choice to use either Shetland or Scottish pronunciation. My
results demonstrate that Soft Mutation is becoming a less frequent
feature of the accent, especially of Well’s DRESS and THOUGHT vowel
sets, which indicates that the accent, like the dialect, is undergoing
accent-levelling.
Table of Contents
1. INTRODUCTION......................................................3
2. LITERATURE REVIEW.................................................62.1. Language Mixing Theories.......................................6
2.2. Phonological Subsystem Interaction.............................72.3. Contemporary Shetland Phonological Inventory...................9
2.4. Shetland Allophony............................................10
3. METHOD AND PROCEDURE.............................................143.1. Data and Materials............................................14
3.2. Vowel Formant Analysis........................................16
Page 2 of 48
3.3. Hypotheses....................................................16
4. RESULTS..........................................................184.1. Behaviour of Soft Mutation in ZE and SC environments..........18
4.2. + DRESS Variation.............................................204.3. + THOUGHT Variation...........................................21
5. LIMITATIONS AND CONSTRAINTS......................................22
6. DISCUSSION.......................................................25
6.1. Origin of Soft Mutation.......................................256.2. Future of Soft Mutation.......................................27
7. CONCLUSION.......................................................29
BIBLIOGRAPHY........................................................30
1. INTRODUCTION
Figure 1. The Shetland Isles and Fair Isle (source: Google Maps)
Page 3 of 48
The Shetland Isles (Fig. 1) are an archipelago of over 100
Islands, 15 of which are inhabited, which are situated in the North
Sea roughly equidistant between Aberdeen, Scotland and Bergen, Norway.
As a result of this geographical position the Isles have been exposed
to numerous language shifts in the past characterised by language
contact and mixing which resulted from social mobility to and from the
Isles. Thomason (2001) defines such language contact as “the use of
more than one language in the same place at the same time [when] at
least some people use more than one language” and his proposal that
“all aspects of language structure are subject to transfer from one
Page 4 of 48
language to another, given the right mix of social and linguistic
circumstances” explains the current optional usage of either Scottish
or Shetland terms presently at work in the Isles. However, this
optional usage is not recent. First evidence of habitation to the
Islands is in the prehistoric era and while discussion is ongoing as
to the specific origin and linguistic situation of the original
settlers the remnants of their culture have been the focus of much
archaeological work in the Isles. During the conquest of Britain
around the first century the Romans also wrote of Islands in the North
Sea which they called “Hermocidae” and “Thule”, but despite such
recorded sightings the Romans never settled. It wasn’t until the 9th
Century that the Islands were first exposed to language mixing when
Viking settlers arrived to settle in Orkney and Shetland and it is
clear, despite a lack of written records from the time, that the
Viking’s culture and Old Norse language eventually supplanted those of
the original population, most clearly seen in that all the placenames
are now Norse which have replaced the names of the original settlers.
Whether this was by force or simply by assimilation is unclear but
some of the Vikings traditions, such as the Up Helly Aa fire festival,
are still practised today and demonstrate the strong association
Islanders owe to their Scandinavian roots. Over the next six centuries
of Viking and subsequent Norwegian rule Old Norse developed into
Shetland Norn and Orkney Norn and the Isles remained monolingual in
their respective dialects until they were ceded to Scotland in the
15th Century. The influence of this Norn language can still be widely
seen in place names with common suffixes denoting geographical or
agricultural characteristics of the settlement. Some of the most
common are -wick (O.N. vik, ‘open bay with little shelter’ e.g.
Lerwick), -voe (O.N. vágr ‘a long narrow sheltered bay’, e.g. Ronas Page 5 of 48
Voe), -ness (O.N. nes ‘a headland’, e.g. Sandness), -bister (O.N.
bólstaðr ‘farm’, e.g. Westerbister) and Scandinavian genealogies are
also evidenced by the large number of patronymic surnames of the
current population, such as Jamieson, Laurenson and Anderson. This
led to increased trade with Scotland, as well as the Netherlands and
Germany, and a second language mixing when Scottish trade brought with
it the Scots language. As trade increased and Scottish traders
settled, Scots eventually prevailed and replaced Norn as the main
language, which was extinct by the 17th Century (Trudgill 1984). Prior
to this, however, mixing of Scots and Norn meant Norn elements had
already been incorporated into the Shetland Scots dialect which
resulted from from the mixing of these varieties.
The present linguistic context in the Isles owes to a third
mixing of varieties. Population levels have risen 30% from around
17,000 in the 1960s to 23,000 today due to the success of Sullom Voe
oil terminal which brought profitable industry to the islands when oil
was struck during the 1970s1. Sullom Voe brought with it a large number
of Scottish workers from the English-speaking mainland, some of whom
permanently settled in Shetland. As a result, the Shetland Scots
variety of the inhabitants is competing with Standard Scottish English
(StScE), which was brought by immigrants. The dialect appears to be
losing and as a result is in noticeable decline. Smith’s (2009) study
comparing a set of questionnaire results of 13-16 year olds in 1983
and 2010 was the first to demonstrate a real-time shift in use of
unique dialect features, which was also accompanied by a change in
attitude toward both varieties. The result is a 30-year gap in which
time StScE has become much more favourable and Shetland much less
1 http://www.shetlandtimes.co.uk/2014/04/17/boom-time-again-at-sullom-voePage 6 of 48
used. “Whereas in 1983 nearly 80 per cent of children could be
expected to have spoken the Shetland dialect at home, now only 65 per
cent might.” The dialect was the only variety spoken in the home in
1983 while at the same time not actively encouraged, and sometimes
openly discouraged, in the education system where marks could be
deducted for spoken dialect use in the classroom. Presently, StScE has
been encouraged, which may be another cause for 'children [who] are
moving away from a diglossic use of the Shetland dialect and English
towards a more mixed use'. Smith proposes that this must be the result
of immigration to the Shetland Isles during the 1970s oil boom and its
subsequent effects on language interaction, accompanied by a sharp
decrease in speakers self-identifying as dialect-only speakers and an
increase of a mixed Shetland-StScE variety.
In the case of language contact in both Original Settlers/Old
Norse and Norn/Scots periods dialect-formation required ‘dialect-
levelling’ (Trudgill 1997) where the loss of L1 features was
concurrent to increased variability in the use of L1 and L2 variants
among speakers. It is clear there has been a marked loss of
distinctive features in younger speakers as opposed to previous
generations (Tait 2001; Van Leyden 2004), thus the dialect may be said
to be levelling in the case of the present Shetland Scots/StScE
language mixing as well. As for attitudes towards this mixing, the
Shetland verb ‘knap’, ‘to speak with affectation, especially
Shetlander trying to speak ‘proper English’’, demonstrates both the
eminence of the present Shetland/StScE mixing and the negative
connotations ascribed thereto2.
2 http://www.shetlanddialect.org.uk/john-j-grahams-shetland-dictionary.php?word=1315Page 7 of 48
2. LITERATURE REVIEW2.1. Language Mixing Theories
An analysis of how Norn and Older Scots phonological inventories
historically mixed to result in Shetland Scots is discussed by
Knooihuizen (2009). He agrees with Miller (2008), who explains the mix
in terms of Trudgill’s model of diffusion and intergenerational
transmission in new dialect-formation. Miller categorizes Shetland
Scots as an L2 Scots variety which evolved from Angus, Fife and
Lothian Scots dialects, while owing strong influences to the L1 Norn
variety of the time. The influence of L1 Norn on L2 Shetland Scots
were principally lexical-phonological and include:
● initial and medial-position stopping of /θ, ð/ to /t, d/ (th-
stopping)
● merger of onset /ʍ/ and /kw/ (hw-confusion)
● T/V distinction in second person pronouns du/you
● use of to be as auxiliary verb to construct the perfect tense (be-
perfective)
● COT-GOAT merger
Page 8 of 48
● prosodic features, which includes Scandinavian Syllable Structure
● gendered anaphoric pronominal reference.
Features attributable to the phonological systems of Older Scots
include:
● the Scottish Vowel Length Rule
● simplification of alveolar clusters /nd/ and /ld/
● BIT variation /ɜ~ɪ/
● Millar’s BUIT vowel /y~ø/, which is represented orthographically
as ö and has traditionally been viewed as a Norn relic
● fronted TRAP vowel /æ/.
However, the date of such change is in doubt. Knooihuizen (2008)
argues that Scots wholly replaced Norn as the functional spoken
language of the Isles before 1774 and that the dialect-formation
process must have begun no later than 1600, since marriage records
from the early 17th century show considerable interethnic marriage
between Scots-speaking immigrants and Norn-speaking Shetlanders
(Knooihuizen 2005). Miller, however, proposes a much later and more
gradual dialect-formation period and places Trudgill’s Stage II at
around 1800.
Disagreement also lies over which parent variety contributed
which features to the daughter variety. The smaller consonant
inventory of Shetland is largely agreed to be Norn, owing to a smaller
consonant inventory in Old Norse, while vowels are largely Scots-
derived since the Older Scots’ richer system of monophthongs (10+) can
more readily explain the number of present Shetland monophthongs
(Section 2.3) than can Old Norse (6-8). However, this has not always
Page 9 of 48
been certain. Th-stopping, commonly regarded as a Norn feature due to
the absence of voiced fricatives in Old Norse, may in reality be a
feature disconnect from both Scots and Norn influence, owing to
Aitken’s Scottish Vowel Length Rule (1981) from the Scots substratum
which states that in Scottish dialects short vowels, with the
exception of KIT /ɪ/ and STRUT /ʌ/ vowels which are invariably short,
are lengthened when followed by: /v, z, ð, r/; a morpheme boundary or
an inflectional suffix (Millar, 2004). The argument depends on the
fact that meid /mi:d/ has a long /i:/ in an unaffected environment (-
SVLR), and that since meid derives from Old Norse meið, which is an
affected environment (+SVLR), th-stopping must have postdated the
effect of the Scottish Vowel Length Rule. This is also evidence by a
second feature affecting vowel quality: Scandinavian Syllable
Structure from the Norn substratum (Melchers 1991; Bandle et. al:
2004) which is itself a result of the Great Scandinavian Quantity
Shift which caused stressed monosyllables in West Scandinavian
language between 1250-1550 to conform to either CVVC or CVCC syllable
structure. This predicts that meid, if it were unaffected by the
Scottish Vowel Length Rule, should have the realization /mid:/. Since
we do not see this, we have good reason to suppose that th-stopping
postdates Norn influence.
2.2. Phonological Subsystem Interaction
Byers-Heinlein et al (2013) ascertain that “exposure to language
mixing might obscure cues that facilitate young bilingual children’s
separation of their languages and could hinder the functioning of
learning mechanisms that support the early growth of their
vocabularies”, while also allowing the possibility that “some parents
modulate the frequency of their language mixing in response to their Page 10 of 48
children’s developing vocabularies”. In this way children using a
variety characterised by mixing of Shetland/StScE are exposed to a
kind of linguistic confirmation bias. They are exposed to StScE in
school and thus adopt these forms, then return home and have those
forms reinforced by parents who employ these same forms to accommodate
the heavily-influenced speech of their children. In keeping with this,
Ferguson’s model allows us to understand the influence of StScE in
Shetland by differentiating the H variety of prestige and ruling-class
style, and the L variety of the working classes which is ‘closer to
the real thinking and feeling of the people’ (Ferguson 1972: 247-248).
As Dorian (1994) notes, wherever a hybrid variety exhibits
considerable variation for L-variant x this precludes the loss of x
through supplanting by x’s equivalent H-variant.
Cruickshank argues that this occurred previously in the move
towards StScE in Shetland in the past, which was encouraged by three
primary factors: the spread of English-language education; the
promotion of StScE as H-variety above Scots; and generally favourable
attitudes towards the English language following the union of Scotland
and England. In the case of present-day Shetlanders the effect of
English-language education is seen in high literacy in English (H-
variety), due to promotion of English in education, and general
illiteracy in Shetland3 (L-variety). The question regarding rate of
change is to what extent such changes are taking place within the
speech community itself. Previously, there have been three ways of
examining how Shetland and StScE variants inhabit and interact in the
mental linguistic space:
3 “I mebbe spik Shetlan’ but I dinna read it” 0:36 - https://www.youtube.com/watch?v=v37bgydws0EPage 11 of 48
● bilingualism: Shetland and StScE occupy two sets of
generally non-overlapping grammars which do not interact;
● bidialectalism: Shetland and StScE occupy the same grammar,
which is characterised by a “mixing of variants” (Smith &
Durham, 2012: 80).
● style-shifting, where there exist multiple context-dependent
forms which may exhibit variation and overlap in a speaker’s
potentially-bidialectal space (Hazen, 2001; Labov, 1994).
This same shift noted in dialect-levelling should, in my mind,
also be applied to phonological categories of dialects which are also
subject to some form of change in a process of ‘accent-levelling’: a
process of phonological change which accompanies lexical, semantic and
syntactic change. This is generally grouped together with dialect-
levelling and is hinted at during Smith & Durham’s analysis of
HOUSE/HOOSE frequency among younger speakers (2012). This research
demonstrates that 50% of younger speakers exhibit style-shifting
between H- and L-variants depending on the linguistic preference of
the interlocutor present, accompanied by a general decrease in the use
of traditional variants in favour of StScE variants by younger
speakers. Three choices are thus available to younger Shetland
speakers, who may: continue to use only Shetland dialect; use StScE
and Shetland dialect in a style-shifting or bidialectal manner;
abandon Shetland dialect and in favour of StScE.
Page 12 of 48
2.3. Contemporary Shetland Phonological Inventory
While most consonants are similar to their Scottish realizations,
with the exception of dental in place of alveolar placement of
plosives, disagreement on the realization of /l/ sheds light on
confusion of both historical strata influence and present language
mixing. Melchers (2010) argues that /l/ is clear in all positions and
illustrates a phoneme totally distinct from StScE phonology, which has
velarized /ɫ/. Sundkvist (2010) argues the opposition for a
velarized /ɫ/ in all positions, similar to StScE. Catford (1957) has a
more systemic solution and posits that the realization of /l/ may be
conditioned by environment: palatalised following front vowels and
velarised following non-front vowels, giving rise to a clear/dark /l/
allophony illustrated in the minimal pair bell ~ ball [bɛɪɫ ~ bælʲ].
Palatalisation of coda-/l/ matches previous observations of similar
effects on palatalisation of coda-/dʲ, nʲ/. Sundkvist, while arguing
for a general velarised realization, expands on Catford’s coda-
palatalisation and suggests this may also specifically affect
consonants which trigger application of the Scottish Vowel Length Rule
while failing to show /l/-allophony. Catford’s /l/-opposition is
certainly systemic and Sundkvist’s illustration of a preference for
velarized realizations is certainly indicative of accent variation
within the Island speech community.
Disagreement also exists over the exact number of vowels which
should be included in the Shetland vowel system. Catford (1957) adopts
Mather’s system of 12 monophthongs, 9 short and 3 long, and at least 2
diphthongs to propose that the similarity of North Angus and Shetland
Page 13 of 48
vowel systems is not coincidental but rather serves as evidence of the
Middle Scots vowel system, which has not been preserved in other Scots
dialects. Melchers adds three monophthongs to Mather and Catford’s
original system. The most common canon for examining vowel sets within
Scottish phonological systems has been Wells’ which was adopted by
Aitken (1981) in describing the Scottish Vowel Length Rule. However,
Miller prefer Johnson’s (1997) sets on the basis that Well’s sets are
unhelpful for examining Scots or StScE, since both varieties owe to
separate linguistic histories of English and Scots phonological
inventories. Scottish varieties contain an already greatly reduced set
of vowels from English as the result of multiple mergers, which leads
to far too considerable overlapping of tokens among Well’s sets for
this to be used as a reliable guideline for use in Scottish dialects.
This is most clearly illustrated by Johnson’s BEAT vowel set. The BEAT
set constituted a separate set in Middle Scots preceding a historic
merger with FLEECE which is still retained in Shetland. The digraph
<ea>, present StScE /ɛ/, is preserved by the equivalent digraph in
Shetland <ae> which recognises the original BEAT set. Therefore in
Shetland there exists a threeway distinction which is illustrated by
regional orthographic variations on head to represent regional spoken
variants [hed ~ hid ~ hɛɪd] as the result of three separate mergers
with BEAT:
haed - Mainland DRESS merger, in which [ɛɪ] is the regular hard
mutation of /ɛ/;
heed - North Isles FLEECE merger, as in historic Scots, which
doesn’t undergo soft mutation;
hed - Geographically unspecified BAIT merger (Miller 2004).
Page 14 of 48
For the present study Wells’ sets have been adopted as they are
suitable for the vowel sets examined and have been done so in previous
explanations of the accent (Sundkvist 2010). The Linguistics Atlas of
Scotland further records:
● -SVLR DRESS lengthening
● +SVLR THOUGHT lengthening preceding nasals and /l/
● +SVLR TRAP lengthening
● compensatory lengthening of vowels resulting from
monophthongization of diphthongs and l-vocalisation.
Regional variation also exists in the exact distribution of vowel
sets within the Shetland isles. Miller’s BUIT vowel may appear on
Burra Isle as /ø/ where in Lerwick it is absent, so school may have two
regional spoken variants [skul ~ skøl]. Further examples of regional
variation of Shetland vowel sets are described by Graham (1993):
● North Isles and Fair Isles AA is rounded to a short O,
demonstrated orthographically by haund in place of haand.
The DRESS vowels merges with FLEECE, e.g. steen for Mainland
stane.
● Whalsay diphthongization of DRESS /steən/,
● Whalsay and Fair Isle long E in MAIN diphthongizes to OI,
eg. doy for day
● Burra Isle, Cunningsburgh and Fair Isles AA becomes short E,
e.g. lesh for lash
Page 15 of 48
2.4. Shetland Allophony
In Section 2.1. I discussed two features of Shetland Scots
phonology which affect vowel quantity: the Great Scandinavian Quantity
Shift and the Scottish Vowel Length Rule. Here I will describe a
feature which also affects vowel quality. Soft Mutation is a form of
vowel allophony which is unique to Shetland among all Scottish
dialects whereby Wells’ KIT /ɜ/, DRESS /ɛ/, TRAP /a/, THOUGHT /ɔ/ and
STRUT /ʌ/ sets are raised and/or fronted (Figure 1) when preceding
particular voiced consonants (Tait 2000). This is recognised by the
Linguistics Atlas for Scotland (2010) for TRAP mutation which occurs
most frequently preceding voiced alveolar plosives, followed by velar
and labial plosives. Millar (2007) recognises this as it applies to
the DRESS vowel but doesn’t record variation for the others sets,
while Johnson’s Insular Scots Clockwise Vowel Shift (1997) notes the
general direction of the shift to rising and fronting. Tait’s Soft
Mutation (Fig. 2) builds upon this more extensively in documenting the
precise regional realizations of these pre-voiced ‘soft’ allophones
and their pre-voiceless ‘hard’ realizations, which can also be shown
in contrasting minimal pairs in Table 1.
Figure 2. Tait’s Soft Mutation
Page 16 of 48
Table 1. Minimal pairs for sets in bVt/bVd environments
Two long vowels are also affected: /a: ~ ɑ:/ which mutates to /æ/
and /eə ~ ɛ:/ which mutates to /e:/ while those which do not undergo
Soft Mutation are long /o/ and both short and long /i, u, ø/.
Sundkvist’s recognition of vowel mutation is generally correct but
confusion may arise for Wells’ KIT set, which reflects an earlier
Angus Scots KIT /ɜ/, while the /ɪ/ phoneme is reserved for the hard
KIT mutation. Tait notes that mutation also applies in pre-stop
cluster environments, thus where affected vowels are followed by
/r/, /l/, /m/, /n/ + /stop/ it is the voicing of the second consonant
in the cluster which will determine whether vowel mutation applies.
Page 17 of 48
While the regular trigger environment is a voiced plosive regional
variation for these triggers exist which produce shibboleths. For
example, in Hamnavoe on Burra Isle bag is affected by Soft Mutation
trigger /g/ and is realized [bɛg] while in Westside /g/ is not a
trigger and so realized as [bag]. Similarly in Hamnavoe voiceless
fricatives act as triggers so that match is realized [mɛt͡ʃ] while in
Westside voiceless fricatives are not regular triggers. This may also
lie behind Graham’s observation that AA becomes E in Burra Isle, which
would mean this isn’t a phonological change affecting AA but rather
allophonic for AA preceding certain fricatives.
While both a grammar and dictionary of the dialect exists (Graham
1993) the lack of regulated written standard, both past and present,
may explain why neither quantitative nor qualitative distinctions are
recognised orthographically, while in West Scandinavian languages
which have both a written standard some form of allophony we see this
represented orthographically, e.g. Norwegian tak~takk /ta:k~tak:/,
demonstrating Scandinavian Syllable Structure. The lack of written
standard, and the failure to show this in the orthography, would also
explain why most Shetland speakers are not consciously aware of either
quantitative or qualitative changes which apply to the dialect.
However, Tait argues this is not necessary due to the unconscious
recognition of Soft Mutation vowel quality in pre-voiced environments,
which is confirmed by the success of the methodology used in this
study.
All discussion of dialect-levelling thus far has been employed
only for lexical, semantic or syntactic variants. Auer et al. (1998)
assert that those features of a phonological system which are least
Page 18 of 48
salient are those most likely to be lost and since it is clear that
Soft Mutation (Section 2.4) is not a salient feature for Shetland
speakers the prediction is that Soft Mutation will be employed with a
reduced frequency or lost entirely by younger speakers. Thus,
examining how non-salient vowel allophony behaves in StScE and
Shetland contexts offers the best look at how the mental language
space treats both varieties in terms of both phonological subsystem
content and interaction which affects accent-levelling. This is based
in Smith & Durham’s (2012) assertion that th-stopping’s allophonic
usage is considered “below the level of the speaker’s social
awareness”. If younger speakers do exhibit bidialectalism this may
suggest separate phonological inventories, however if Soft Mutation is
not applied this still requires explanation. Are vowels unaffected due
to the speaker’s subsystem:
1. inhibiting Soft Mutation in only a number of Shetland
environments, which may simply be either accent-levelling or
regional variation;
2. not extending Soft Mutation in any Shetland environments
where expected, which supports non-salience of Soft Mutation
and;
3. inhibiting Soft Mutation to facilitate comprehension on
behalf of non-Shetlander interlocutors, i.e. ‘knappin’’?
To remove option 3 I remained absent from the recordings and had
personal Shetland locals gather data on my behalf. To limit regional
variation of individual lexical tokens a list of pseudowords
containing KIT, DRESS, TRAP, LOT and STRUT vowels in hard and soft
mutated environments were used. Participants were told that these Page 19 of 48
lists contained a mixture of old Shetland words and old Scottish
words. While this was true insofar as the Shetland set contained no
obviously Scottish words and vice versa in reality both sets contained
a mixture of English words and pseudo tokens. By disguising these
false words as local or non-local I expected to see authentic and
natural application of Soft Mutation to both Shetland and non-Shetland
environments without the risk of speakers moulding pronunciation on
previous exposure to tokens. Two Scottish participants with no ties to
Shetland who did not use any form of pre-voiced vowel allophony were
used as controls to give mean values against which Shetland speaker
values were compared. However, my focus is not to categorize speakers
as speaking either specifically Shetland or StScE but rather whether
expected features are absent, which will serve both as evidence that
the accent is changing concurrent to the dialect and that these less
salient features are more likely to erode. Since Soft Mutation is not
a generally salient feature of the accent, or at least one which like
th-stopping is above the level of conscious awareness, I have chosen it
as the principal variable to analyse accent change in Shetland. This
will also serve as an insight into the relationship between accent
change compared with previously documented dialect-levelling.
Page 20 of 48
3. METHOD AND PROCEDURE3.1. Data and Materials
This study employs standard acoustic analysis methodology
performed in PRAAT speech analysis software (Boersma & Weenink 2005)
and all diagrams in this section were made either in Microsoft Excel,
R statistical software’s phonR package (2008; package update March
2014) or NORM (2014) and inserted from Window’s Snipping tool. The
data consists of one exercise containing two words lists of numerous
stressed monosyllable tokens primarily concerning isolated speech.
This was intended to show whether participants contained two separate
phonological systems for Shetland (ZE) and Scottish (SC) words or one
into which all tokens fell. All tokens were uttered in carrier phrases
“Shö kens ___ laekly (ZE) / She knows ___ likely (SC)” to frame tokens
in appropriate Shetland/Scottish environments (Torre & Barlow: 2006).
While this didn’t ensure whether Soft Mutation would be extended, it
is clear from the results (Section 4) that this difference had a
successful and noticeable effect. Participants were also told to
discuss whether they found token words familiar, firstly to elicit
connected speech, which wasn’t used in recordings, and to distract
participants from the true purpose of the study into thinking the
study examined lexical knowledge or recognition. The majority of
tokens were pseudowords to remove any speaker bias towards
pronunciation, with the occasional authentic word to give participants
the impression that real, albeit unfamiliar, Shetland had been chosen,
and also to examine how regularly Soft Mutation was extended in tokens
which the participants believed were of Shetland or Scottish origin.
In discussing the effect of Soft Mutation in ZE/SC environments my own
Page 21 of 48
notation for Soft Mutation environments will show hard unaffected
environments (-DRESS) and soft affected environments (+DRESS). This
does not indicate actual realized vowel quality but rather expected
vowel quality. Recordings were collected by three trusted contacts in
Shetland who sampled participants according to age group and oversaw
both the recording and completion of ethical forms. Contacts informed
participants of the purpose of the study retrospectively and did not
pass on personal information of participants to myself. Participants,
after sampling, completed a consent form with tick boxes to provide
information on age (<18; >50), gender (Male/Female) and parental
heritage (Both Shetland, One Shetland, Both Scottish). Participants
were divided into two age groups based on likelihood of exposure to
Scottish tokens: Old speakers (O) aged over 50, with the assumption
that pre-1970s speakers who were born, raised and still settled in the
Isles were far less likely to have been affected by immigrant
varieties and more likely to preserve historical variants; and Young
(Y) speakers, who were aged under 18 to represent the post-
immigration, mixed-parentage, and thus mixed-variety generation. The
total number of participants after completion of data analysis was 16
whose gender, age and parental origin are shown in Fig. 3, as well as
the distribution of O/Y speakers in Fig. 4.
Figure 3. Graph illustrating participants’ age, gender and parentage
Page 22 of 48
Figure 4. Participants’ age as a percentage of total (N=16)
Contacts presented participants with consent forms and exercise
sheets which were completed before being recorded as audio files onto
a smartphone microphone. This were then sent to me and imported into
Page 23 of 48
PRAAT on default settings where F1 and F2 values were taken and noted
into an excel spreadsheet document, before being analysed in phonR and
NORM. F3 was to be measured for THOUGHT and STRUT vowels, which
undergo rounding shown by a simultaneous drop of 400Hz in F2 and F3
(Raphael et al. 1978), but this was not uniformly applied to all
participants due to time constraints. Therefore, all values in Section
4 only demonstrate vowel fronting and height rather than rounding.
This was not expected to be problematic, since both THOUGHT and STRUT
undergo both rounding and fronting. All contacts were present at the
time of recording and were fellow Shetlanders with noticeable accents,
which was intended to remove style-shifting as a variable. The
presence of native islanders and accents was expected to place
participants, especially younger participants, at greater ease during
the recording (Souza et. al: 2013).
3.2. Vowel Formant Analysis
Monophthong values were taken according to Ladefoged (2003) in
what was perceived to be the midpoint of the vowel’s duration, rather
than in either the middle of a period where there is little change, or
“central tendency”, or as the mean value of the vowel over a specified
duration. However, the presence of voiced coda consonants incurred
formant transitions which proved especially difficult in the case of
KIT tokens, which appeared to be more greatly affected by offset
consonant transitions while still not being perceived as diphthongs.
KIT was thus measured at the central point of what was perceived to be
its central tendency. Diphthong values were only expected for +DRESS
but also showed notably for +THOUGHT. Here, the first vowel in the
Page 24 of 48
diphthong was measured either as close to the initial point of the
vowel unaffected by onset transitions as possible or in what was
perceived to be the point which showed the least effects from
transitions of the onset consonant. The second element was measured in
the same way at a point which was perceived to be unaffected by offset
formant transitions.
3.3. Hypotheses
H Y will show a general reduced frequency of Soft Mutation in
soft environments compared with O, which will support the premise
that dialect-levelling at the lexical level in Shetland is also
accompanied by phonological accent-levelling.
H0 Y will show no significant change from O, in which case
future research is required to analyse how and whether the pace
of accent change accompanies dialect change.
Page 25 of 48
4. RESULTS4.1. Behaviour of Soft Mutation in ZE and SC
environments
Here I will be analysing how Soft Mutation behaves in ZE and SC
environments. Fig. 5 illustrates a vowel plot for mean vowels in hard
environments according to age group with control values added, which
will be taken as the default unaffected values against which Y and O
values will be examined. These control values predictably did not
change depending on hard or soft environments and were thus suitable
controls. What can be seen is that -TRAP, -DRESS and -THOUGHT largely
overlap across age groups with accompanying control values. The
Page 26 of 48
position of /ɪ/ for -KIT vowel seems to imply a mean loss of /ɜ/ for Y, who instead adopt a similar phoneme to Scottish speakers, while O
clearly employ a separate /ɜ/ phoneme. -STRUT is also generally raisedfor Shetland speakers compared to Scottish speakers, which illustrates
a phonological change in point of articulation rather than mutated
variation. What this illustrates is that even in hard environments
Shetland speakers employ the same Shetland phonology regardless of age
with the exception of -KIT which has been replaced by Scottish values
in Y. Thus, already in the hard phonology we can see the effect of
accent influence.
Mean +SC values illustrated in Fig. 6 demonstrate that +TRAP,
+DRESS and +STRUT don’t have greatly differing values compared to
their -SC equivalents. +DRESS and +KIT have mean values which are
raised in the vowel space by around 100 Hz compared to hard
environments, but overall only +THOUGHT appears to indicate any
notable movement, with far more pronounced fronting in hard Scottish
environments for Shetland speakers which overlaps with +STRUT.
Mean +ZE values illustrate in Fig. 7 were expected to show the
greatest change, with Soft Mutation most greatly applied across
speakers. This does not appear to apply to +KIT, which demonstrates
little change from its +SC values. O shows a pronounced fronted +TRAP
vowel /æ/ and raises +DRESS vowel to overlap with +KIT, which
indicates diphthongization. Both Y and O fronted +THOUGHT and raised
+STRUT with little variation.
What may be concluded for +SC/+ZE environments is that all vowel
sets appear to undergo change, most visibly here are +TRAP, +DRESS and
+THOUGHT, with less pronounced differences for +KIT and +STRUT.
Page 27 of 48
However, this change is not uniform and only +THOUGHT illustrates
shift for both +SC and +ZE environments. This seems to support the
claim that speakers possess multiple phonological subsystems which
affect Soft Mutation realization.
Figure 5. Mean positions for - environments Figure 6. Means positions for +SC
environments
Figure 7. Mean positions for +ZE environments
Page 28 of 48
What is clear from the previous results is that three vowels
illustrate the most notable change dependent on environments: +TRAP,
+DRESS and +THOUGHT. +TRAP fronting in +ZE environments but not in +SC
environments has been clearly demonstrated in Fig. 7, which at least
suggests some level of phonological subsystem dichotomy. I will here
analyse the behaviour of the two further sets which showed most
promising variation between +SC and +ZE environments: +DRESS and
+THOUGHT. All individual values in diagrams for Sections 4.2 and 4.3
are Lobanov normalized and portray vowel spaces situated between
Catford’s (1988) mean vowel positions for /i, a, u/, which are
highlighted, to give an indication of relative position of vowels
within the vowel space.
4.2. + DRESS Variation
Figure 8. Individual O speakers’ +DRESS values Figure 9. Individual Y speakers’ +DRESS values
Page 29 of 48
Table 2. Percentage of monophthongs/diphthongs for +DRESS environments
+DRESS O (N=32) Y (N=40)
Monophthong[ɛ]
17 (53%) 32 (80%)
Diphthong[ɛɪ]
15 (47%) 6 (20%)
All hard environments for both speaker groups contained 100%
monophthongs as expected and Fig’s 8 and 9 demonstrate +DRESS values
for individual O and Y speakers in both +ZE and +SC environments. The
distinction is most clearly seen in the application and frequency of
monophthongs to diphthongs. While O employed diphthongs in 47% of soft
environments, Y only did so in 20% of soft environments. Thus, Y are
using Soft Mutation with 27% less frequency than O. However, this
figure is more dramatic when considering that the 47% value appears to
be based on triggers, rather than simply random variation. This is
seen where O speakers have diphthong [ɛɪ] in 100% of pre-/d/
Page 30 of 48
environments and monophthong [ɛ] in 100% of pre-/b/ environments. It
could then be concluded that /b/ is not a Soft Mutation trigger at all
while /d/ is a trigger which affects 100% of values. The effect of
this pre-/b~d/ environment trigger also applies to 66% of Y values
while the remaining 33% were used by one speaker for 100% of ZE
tokens. While the data set is admittedly small we may conclude that
for 100% of O participants a distinction between +/-DRESS was
preserved which applies to both +ZE and +SC values, while for Y
speakers the same distinction only applies for 20% speakers and 10%
apply Soft Mutation in only ZE environments, indicating loss of
expected +DRESS along and age gradient.
4.3. + THOUGHT Variation
Figure 10. Individual O speakers’ +THOUGHT values Figure 11. Individual Y speakers’ +THOUGHT values
Table 3. Percentage of monophthongs/diphthongs for +THOUGHT environments
+THOUGHT O (N=31) Y (N=38)
Page 31 of 48
Monophthong[ɔ]
5 (19.2%) 27 (71.1%)
Diphthong[œ͡ɛ/œ͡ə]
26 (83.8%) 11 (28.9%)
All hard environments for both speaker groups contained 100%
monophthongs as expected and Fig’s 10 and 11 demonstrate +THOUGHT
values for individual O and Y speakers in both +ZE and +SC
environments. The position of the second diphthong element seems to
have two main variants for both speakers: either fronting towards /ɔ͡ɛ/
or centralizing towards /ɔ͡ə/. Due to the absence of F3 it is difficult
to tell whether these values are rounded or unrounded, however in
vowel analysis there was perceived to be a general tendency for
diphthong values to be rounded, thus producing variants [œ͡ɛ] and [œ͡ə]
All tokens occurred preceding /d/ triggers, therefore we may conclude
that for both age groups two variants for +THOUGHT a common rising
diphthong and less common centralizing diphthong, may occur
irrespective of environment with frequency of diphthongizations
reducing along an age gradient.
Page 32 of 48
5. LIMITATIONS AND CONSTRAINTS
The largest constraint on this study was time management
surrounding compilation and analysis of data. Exercises were finalized
one month into research after several trials on participants whose
results were not included in the study in order to do background
reading to select proper variables for research, while the following
month was spent sending exercises to colleagues and giving colleagues
time to gather and return the data from participants. The following
month was spent performing formant analysis, where transcription
averaged one hour per participant. The intended final number was 30
participants: 10 each in Old, Middle and Young age groups, however a
smaller number was received than expected and after removing the one
Middle group speaker, which was to analyse speakers aged 25 to 50 to
give greater indication of the speed of accent loss, the final number
Page 33 of 48
of speakers reached only 16. This speaker’s results largely matched
those of the Old group, which would indicate that + values have been
lost most rapidly within the past generation, however this doesn’t say
anything for the remainder of the age group and is therefore
inconclusive.
Originally a duration analysis test was considered which was
expected to demonstrate that if Soft Mutation weren’t extended by
Young speakers, which would serve as evidence of accent-levelling,
that this would be accompanied by a drop in the frequency of
Scandinavian Syllable Structure application. Young speakers were
expected to both omit Soft Mutation and adopt non-allophonic Scottish
syllable structures instead. In order to test for such change I
adopted Van Leyden’s (2004) method and guidelines for measuring
boundaries between the nucleus-vowel and the coda-consonant. My method
differed in that minimal pairs were primarily based on vowel quality
affection rather than quantity affection and thus consisted primarily
of VCC environments. The only VVC environments expected were in
diphthong +DRESS and +THOUGHT environments. This study also differed
by examining age range. While vowel duration is a contrasting feature
of Shetland speech (Tait 2000) the reliance on pseudowords should have
excluded this as a variable, however pseudowords may have been
mistaken for actual Shetland words with the result that participants
mistook final consonants for a +SVLR past tense morpheme boundary,
such that trid may have been understood as tri#d. This may also have been
mistaken for lexical exceptions where vowel contrast applies, e.g.
/lumi ~ lu:mi/, and thus have unexpected syllable structures. All
tokens qualified for examination and were plosive-final, in –SVLR
environments and thus expected to be regularly VCC or VVC. However, Page 34 of 48
analysis was dropped due to both time constraints and a lack of
significant variation to display. After mapping syllable length for
several participants in both age categories it was concluded that
VVC/VCC structure exhibits considerably less, if any, variation
compared to Soft Mutation for younger speakers. This may mean syllable
length is a more salient feature of which speakers which according to
Auer determines more resilience to change, however when asking several
Shetland-speaking contacts about the feature they seemed relatively
unaware of Scandinavian Syllable Structure as a general principle and
far less so in their own speech until it was pointed out to them.
Future research is required on the behaviour of Scandinavian Syllable
Structure and whether this varies across generations which would be
beneficial to understanding the interaction of dialect- and accent-
levelling in Shetland. In the end this was deemed beyond the scope of
this study.
Another potential limitation for this study was my choice to
employ apparent-time methodology to examine generational accent change
where real-time studies have also been employed to study generational
dialect change in Shetland. Thus far apparent-time studies have been
the most common method in sociolinguistic research as they demand less
time, however the results are less revealing. Apparent-time
observations may demonstrate either stable variation within a speech
community which is unaffected by external influence or external
influence and unnatural speech variation. (Bailey Guy et. al, 1991).
Eckert (in Coulmas 1997) comments that “increasing age correlates with
increasing conservatism in speech, however in that case it is not
clear whether or not age-stratified patterns of variation actually
reflect any change in progress or can be seen as a stable variable.” Page 35 of 48
In this light Coulmas (1997) advises that “only real-time studies can
unequivocally deliver clear-cut results for changes in progress”.
Durham and Smith’s real-time analysis over a 30-year period (2009)
comes close to delivering this. Ideally, this present study would have
combined both apparent and real-time data to show both that accent
change is occurring across an age gradient and the lesser likelihood
that this change is simply stable variation in a speech community.
Suppose that a real time study concludes the older generation ‘knap’
to the same frequency as younger speakers, presumably to encourage
communication between speakers of different backgrounds. It would
still be logical to presume the younger generation have built upon
changes in the older generation’s speech concurrent to blending
external imports in speech. An example of variation which is not
attributable to purely internal variation in Shetland would be the
HERD/BIRD/CURD merger, which was present in the connected speech of
some younger Shetlander’s speech in this study. Since such a merger is
regarded as a specifically Scottish feature with no parallel in other
language varieties affective in Shetland this is likely to be the
result of external influence.
In his study on age-related real-time variation Harrington (2006)
concludes the Queen’s use of the KIT vowel shows raising over time
which may be the result of Cockney influence. He argues that vowels in
multiple sets may overlap in positions where they are not obliged to
contrast phonemically. It is to be expected from this that such
overlapping leads to increased variation between vowels in such
positions. The pseudo words used in this study were crafted under the
assumption that Soft Mutation was a uniform change which occurred
generally in pre-voiced plosive environments. However, both data from Page 36 of 48
the Linguistics Atlas of Scotland and the present results for +DRESS
suggest that plosive place of articulation may affect Soft Mutation
values in a regular way. In the absence of information on speaker
origin it is difficult to ascertain whether this is simply regional
variation or a general trend.
Page 37 of 48
6. DISCUSSION6.1. Origin of Soft Mutation
Of most interest regarding Soft Mutation is that no published
literature examines how it emerged in the first place. Here I believe
I have a satisfactory explanation which treats Soft Mutation as an
instance of I-mutation triggered by palatalisation of coda-position
/lʲ, nʲ, dʲ/. Evidence of I-mutation, a process whereby the vowel in
Syllable x may be fronted and raised where Syllable x+1 contains /i, ɪ,
j/ or a palatalised consonant /Cʲ/, may be most strongly seen by the
presence of /ɪ/ in +DRESS mutation /ɛɪ/, where palatalisation has
introduced an /ɪ/ preceding the coda consonant, and for +KIT by the
supplanting of /ɜ/ by /ɪ/. Why /ɪ/ isn’t retained for the remaining
sets may have an explanation in variants for +THOUGHT which has
standard hard mutation [œ͡ɛ] and common variant [œ͡ə]. Retention of
/ɛ~ə/ schwa in the both variants may indicate former /i/ which has
gradually been lowered. While identifying palatalised final consonants
was not the principle exercise of this study, Catford’s claim that
coda-position /l, n, d/ were palatalised to /lʲ, nʲ, dʲ/ may be
accepted as an example of historical realization. Sundkvist also
notes that “/d, n, ŋ/ are palatalised in contexts that overlap with
the application of Shetland Vowel Mutation: bad [bædʲ], ban [bænʲ],
bang [bæŋʲ]” (2008). Thus, where /d, l, n, ŋ/ are all [+voiced] and
thus predictable triggers for Soft Mutation and where these
environments were historically /lʲ, nʲ, dʲ/, of which this study
observes /d/ to be the most reliable trigger, a plausible history for
all the mutation of all sets is shown in Table 4 where original
Page 38 of 48
palatalised coda-consonants insert /ɪ/ preceding the coda which
triggers mutations before /ɪ/ is either lowered in +THOUGHT, replaces
original vowel in +KIT, retained in +DRESS or lost in +TRAP and
+STRUT. It should be noted that in results one O speaker contained
+TRAP diphthong /æə/, which was disregarded as an exception, however
this may indicate retention of /ə/ preceding loss in O speakers.
Palatalisation of coda-consonants was not observed for all speakers,
therefore an added stage of loss of coda palatalization may be
suggested.
Table 4. Proposed historical development of Soft Mutation
Page 39 of 48
Further evidence is required to validate this hypothesis and my
solution is not conclusive. For example, if /l, n/ have always been
palatalised why has previous literature concluded there exists more
considerable regional variation for these in soft environments than
for /d/? Further questions demand research regarding the exclusivity
of historical palatalisation of /l, n, d/, in which case the extension
of Soft Mutation to other pre-voiced environments, e.g. /g/, requires
explanation. Regional variation in pre-/g/ and pre-fricative
environments in 2.4 may shed light on this.
6.2. Future of Soft Mutation
This study has sought to demonstrate that the Shetland
accent is being lost at a similar rate to the dialect and that Sullom
Voe oil terminal has played a major role in such change. These results
support this hypothesis. As a result of the data compiled here across
two age groups we may plot a future projection for the frequency of
Soft Mutation as it will be applied in successive generations over the
Page 40 of 48
next 30 years that Sullom Voe is projected to be profitable, which
places the present study at roughly half-way through the present
period of expected immigration. By the end of the profitability in
2045 Soft Mutation may cease to exist as a standard feature in the
accent of younger speakeres, as indicated in Fig. 12. While this
forecast is obviously not conclusive and in the absence of a middle
age group to compare values it may be that the rate has slowed down in
recent years it does illustrate that the accent is losing distinctive
forms of the phonology which may disappear within the next few years
and be entirely lost by the next generation.
Figure 12. Predictions for future application of Soft Mutation
1975 1985 1995 2005 2015 2025 2035 2045020406080100
Forecast for +DRESS and +THOUGHT diphthongization
DRESS Linear (DRESS)THOUGHT Linear (THOUGHT)
Page 41 of 48
7. CONCLUSION
In conclusion, these results support the claim that the Shetland
accent is changing and the results of this study specifically give
evidence which supports hypothesis H: that Soft Mutation varies
according to age and environment and that these may be due to accent-
levelling.
For hard environments there are no visible affects of age for -
TRAP, -DRESS and -THOUGHT vowels, which also broadly overlapped with
Scottish control values and indicated no separate of phonological
subsystems. -KIT values give reasons to suppose that while older
speakers retain /ɜ/ younger speakers are replacing this with Scottish
KIT value /ɪ/. This preference for the H-variant supports accent-
levelling. Also evident was that -STRUT has a raised /ʌ/ irrespective
of age, which does not appear in the literature. For soft Shetland
Page 42 of 48
environments there was no great change to the +KIT vowel compared to
hard environments, which would indicate that +KIT does not undergo
mutation although perception to the contrary suggests that this is not
conclusive. Shetland speakers of both ages treated +THOUGHT and +STRUT
similarly, generally fronting +THOUGHT with a raised +STRUT, although
without F3 values it is unclear whether this undergoes mutation for
both sets. However, age appeared to affect +TRAP, which older speakers
realized as predicted /æ/, and +DRESS which showed diphthong values.
For soft Scottish environments there was no change across age groups
to +TRAP, +DRESS and +STRUT when compared with hard environments,
indicating that these do not undergo change. +DRESS and +KIT exhibited
a lowered F1 of around 100 Hz for both ages, however +THOUGHT appeared
to be fronted, and possibly rounded, in hard environments for Shetland
speakers.
Also presented was a possible account for the development of Soft
Mutation which resulted from palatalisation of voiced coda-position
consonants. This study also makes it possible to dismiss certain
claims as either false or at the very least no longer valid. For
example LAS, which notes lengthening of -DRESS and of +TRAP, neither
of which were evident in results.
Page 43 of 48
BIBLIOGRAPHY
Aitken, A. J. ‘The Scottish vowel length rule’. In Benskin, Michael
and M. L. Samuels (eds.) 1981. So Meny Peoples Longages and Tonges: Philological
Essays in Scots and Mediaeval English Presented to Angus McIntosh. Edinburgh: Middle
English Dialect Project, 131–57.
Auer, P. Birgit. B, Grosskopf. B. (1998). “Subjective and Objective
Parameters Determining ‘Salience’ in Long-Term Dialect Accommodation.”
Journal of Sociolinguistics 2.2: 163-87. doi:10.1111/1467-9481.00039
Bailey, G., Wikle, T., Tillery, J., & Sand, L. (1991), The apparent time
construct, Language Variation and Change, 3, pp. 241–264
Bandle, O., Braunmuller, K., Jahr, E. H., Karker, A., Naumann, H.,
Telemann, U., (2004) The Nordic Languages 2: An International Handbook of the
History, Mouton de Gruyter: London
Boersma, P., & Weenink, D. (2005). Praat: Doing phonetics by computer
(Version 4.3.14). (computer program): Retrieved May 26, 2005, from
http://www.praat.org
Catford, J.C. (1988). A Practical Introduction to Phonetics, Oxford University
Press
Catford, J. C. (1957). Shetland Dialect, Shetland Folk Book vol. 3.
Coulmas, F. (1997). The Handbook of Sociolinguistics, Blackwell
Publishing: United KingdomPage 44 of 48
Graham, J. J. (1993). The Shetland Dictionary. The Shetland Times: Lerwick
Dorian, N. C. (1994). Varieties of variation in a very small place:
Social homogeneity, prestige norms, and linguistic variation. Language,
70, 631-696
Harrington, J. (2006). An acoustic analysis of ‘happy-tensing’ in the
Queen’s Christmas broadcasts (2006), Journal of Phonetics 34, 439-457
Hazen, K. (2001). “An Introductory Investigation into Bidialectalism.”
In “Selected Papers from NWAV 29,” edited by Tara Sanchez and Daniel
Ezra Johnson, 85-99. University of Pennsylvania Working Papers in linguistics 7.3
Johnston, P. (1997). 'Regional Variation' in Jones, C., ed., Edinburgh
History of the Scots Language. Edinburgh.
Knooihuizen, R. (2005). The Norn-to-Scots language shift: another look
at socio-historical evidence. Northern Studies 39. 105-17
Knooihuizen, R (2008). Fishing for Words: The Taboo Language of
Shetland Fisherman and the Dating of Norn Language Death. Transactions of
the Philological Society 106. 100-13
Knooihuizen, R. (2009). Shetland Scots as a new dialect: Phonetic and
Phonological Considerations, English Languages and Linguistics 13, 3, 483-501
Labov, W. (1994). Principles of Linguistic Change, Vol. 1, Internal Factors,
Oxford: BlackwellPage 45 of 48
Ladefoged, . (2003). Phonetic Data Analysis. An Introduction to Fieldwork and
Instrumental Techniques. Oxford: Blackwell.
Mather, J. Y., & Speitel, H. H. (2010). The Linguistics Atlas of Scotland,
Routledge: UK
Melchers, G. English spoken in Orkney and Shetland: Phonology, in A Handbook of
Varieties of English (2004), edited by Kortmann, B. and Schneider, E.,
Mouton de Gruyter
Miller, R. M., (2007). Dialects of English: Northern and Insular Scots, Edinburgh
University Press: UK
Miller, R. M., (2008). The origins and development of Shetland dialect
in light of dialect contact theories. English World-Wide 29, 237-67
R Development Core Team (2008). R: A language and environment for
statistical computing. R Foundation for Statistical Computing, Vienna,
Austria. ISBN 3-900051-07-0, URL http://www.R-project.org.
Raphael, L. J., Bell-Berti, F., Collier, R., & Baer, T. (1978). Tongue
Position in Rounded and Unrounded Vowel Pairs. Status Report on Speech
Research SR-54
Smith, Jennifer (2009). Obsolescence vs. stability in a Shetland
dialect: evidence from three generations of speakers: Full Research
Report ESRC End of Award Report, RES-000-22-2052. Swindon: ESRC
Page 46 of 48
Smith, J. & Durham, M. (2012). Bidialectalism or dialect death?
Explaining generational change in the Shetland Islands, Scotland,
American Speech 87, DOI 10.1215/00031283-1599959
Souza, A. L., Byers-Heinlein, K., Poulin-Dubois, P., Bilingual and
monolingual children prefer native-accented speakers. Frontiers in
Psychology, Published 23 December 2013
Sundkvist, P. (2010). Scottish Standard English as spoken in Lerwick:
an overview of pronunciation features. In Millar, Robert McColl (ed.)
2010. Northern Lights, Northern Words. Selected Papers from the FRLSU Conference, Kirkwall
2009. Aberdeen: Forum for Research on the Languages of Scotland and
Ireland, 98-106. ISBN: 978-0-9566549-1-5
Tait, J. (2000). Some Characteristics of the Shetland vowel system,
Scottish language, 19, 83-99
Tait, J. (2001). “Whit is Shetlandic?” Lallans 58: 7-26
Thomas, E. R. & Kendall, T. (2007) NORM: The vowel normalization and
plotting suite. [ Online
Resource: http://ncslaap.lib.ncsu.edu/tools/norm/ ]
Thomason, S. G. (2001). Language Contact. Washington, DC: Georgetown
University Press
Trudgill, P. (1997). Third-person singular zero: African-American
English, East Anglian dialects and Spanish persecution in the Low
Countries. Folia Linguistica historica 18, 139-48
Page 47 of 48
Trudgill, P. (1984). Language in the British Isles, Cambridge University Press:
UK
Van Leyden, K. (2004). Prosodic Characteristics of Orkney and Shetland
Dialects: An Experimental Approach, Ph.D. Diss., Leiden University
Torre, P., & Barlow, J. A. (2006), Age-related changes in acoustic
characteristics of adult speech, Journal of Communication Disorders, 42, pp.
324-333
Page 48 of 48