|
Roman Phonetic Alphabet for English Lyubomir Ivanov ( (Paper published in Contrastive Linguistics, XXXII, 2007, 2, pp. 50-64) Abstract The present work describes the underlying principles of the 2002 Basic
Roman Spelling of English [4] aimed at providing an alternative English
orthography for international usage. This has been developed by Lyubomir Ivanov,
who introduces here a new construction for that system, and proposes a
closely related Roman phonetic alphabet to be used for the pronunciation
respelling of English without special characters or diacritics. A comparison is made with Interspel [11], a system developed by Valerie Yule that
attempts to maximize the advantages and remove the disadvantages of
traditional spelling of English, to benefit learners, users and international
communication. 1. Re-Romanization Re-Romanization is to replace an orthographic system that uses the Roman
(Latin) alphabet for writing the words of a certain language by another
writing system that is different yet based on the same alphabet. Traditional orthographies are often
modified by reform proposals when the spelling of languages is perceived as
reflecting past stages of development rather than actual present day spoken
language. Traditional English
orthography (Traditional Spelling of English, TS) is an obvious such case of outdated spelling, which according
to some research impedes the acquisition of literacy and efficient reading
and spelling [10]. Systems like the 2002 Basic Roman Spelling of English (BR) [4]
seek to re-Romanize the spelling for the purposes of academic research and education
in general, both for native and non-native English speakers. In one such application, BR is adapted in
the present work to serve as a phonetic alphabet for the pronunciation
respelling of English. This new
phonetic alphabet has no diacritics, and unlike the International Phonetic
Alphabet [2] has no additional special characters. The BR orthographic system could be arrived at from various starting
points. One of these is the original
construction [4], which makes use of an intermediary Cyrillic phonetic
transcription of English words (possibly leaving the misleading impression
that BR was somehow associated with or influenced by Bulgarian phonetics or
orthography). Here we shall present
another construction of BR starting from scratch, which provides a better
introduction and explanation of the system. 2. Spelling Principles Our first step is to formulate and substantiate the small number of basic
principles that are inherent to the BR system. 2.1. Strict
Romanization Principle:
Use the basic Latin alphabet, with no additional characters or diacritics. Traditional English orthography essentially adheres to this principle
too, with minor exceptions involving loanwords from other languages. In the case of languages with Latin-based
orthography employing diacritics, the mass practice of modern electronic
communication (e-mail, instant messaging, short message service etc.)
overwhelmingly uses not the true alphabets but their corrupted versions
stripped from all diacritical marks and special characters. Therefore, it is only natural to keep this
felicitous advantage of English spelling, that it is e-communication
friendly. 2.2. Consistency
Principle: Use
single-valued spelling, with no phoneme rendered by two or more graphemes. In the orthodox English orthography one and the same phoneme is often
spelled differently in different words.
For instance, the consistency principle is extensively violated by the
multiple-valued rendering of the English phonemes /æ/, /e/, /eɪ/, /ɪ/, /i:/, /ɒ/, /aɪ/, /aʊ/, /oʊ/, /ə/, /u:/, /ɜ:/ respectively in: have, salmon; red,
jeopardy, says, guess; paper, rain, way, eight, break; big, damage, pretty, women, busy,
myth, build, marriage; feel, beach,
shield, perceive, key, people; tall, walk; nine, try, high, tie, height, buy,
bye, eye, aisle, sign; out, now; no, know, boat, soul, toe; ago, anthem, awesome,
iridium, mountain; mood, soup, jewel, true, lose,
fruit, through; firm, fern,
turn, worst, earth, err. This list can easily be extended, as vowels,
diphthongs, and consonants have multiple-valued grapheme presentation in
numerous other instances. In particular, the above examples demonstrate
that the short vowels /ɪ/ and /ɒ/ are presented in TS by more than one single letter
each, with as much as six single letters representing the former as in big, damage, pretty, women, busy,
and myth. 2.3. Proportionality
Principle: Spell short
vowels by single letters; spell long vowels and diphthongs by digraphs. Orthodox English orthography defies the proportionality principle e.g. by
using a, eigh and aigh
for /eɪ/ as in paper, eight and straight; e for /i:/ as in delete; i for /aɪ/ as in fine; o for /oʊ/ as in no; u for /jʊ/ or /ju:/ as in duty and tune; y for /aɪ/ as in by; ea, ai,
ie, eo, ay, ue for /e/ as in leather, said, friend, jeopardy, says, guess; oo, ou for /ʊ/ as in book, should; ou and oe
for /ʌ/ as in touch, does; ai and ou for /ə/ as in mountain, famous; etc. It is worth mentioning the uniform
presentation by the same single letters of pairs of short/long values of
vowels, such as the pairs /æ/-/eɪ/; /e/-/i:/; /ɪ/-/aɪ/; /ɒ/-/oʊ/; and /ʌ/-/jʊ/ (or -/ju:/) in national/nation,
serenity/serene, finish/final, posture/pose,
and study/student.
These vowel alternations in stressed syllables have their origins in
the Great Vowel Shift of Early Modern English, the pivotal process of Modern
English phonology according to Chomsky and Obviously, there are three possible ways of
dealing with the Great Vowel Shift at orthography level. The first one would be to disregard it and
keep the pre-Shift (now dual) usage of the letters a, e, i, o, and u, thereby preserving the spelling
uniformity in relevant word families (such as the word pairs above) at the
expense of phonemicity. That is
precisely what the TS does. A technical advantage of this approach is
that the presentation of long value vowels by single letters contributes to
the brevity of TS, i.e. the lower grapheme to phoneme ratio in TS texts. A second tactic employed by Interspel [11] would be to maintain the dual
usage for the five single vowel letters, but indicate long vowels when
necessary, especially for learners, with a diacritic, as in Interspel national/nàtion,
repetition/repèt, finish/fìnal,
impotent/pòtent, study/stùdent. This
approach, proposed for investigation by Valerie Yule, is discussed in greater
detail below. The third approach, prescribed by the
proportionality principle and adopted by BR, would be to explicate the vowel
shift by a corresponding spelling shift, to the effect of preserving
phonemicity at the expense of the abovementioned uniformity of word families
spelling. 2.4. Context
Freeness Principle:
Spell diphthongs in accordance with the spelling of their components; spell
long vowels either as diphthongs or by doubling the letters spelling the
respective short vowels. In other words, xy renders /αβ/ if and only if x renders /α/, and y renders /β/.
That is, whenever some letters x, y represent respectively the
short vowels /α/ and /β/, then the digraph xy represents the
diphthong /αβ/; also, xx represents the long vowel /α:/.
And conversely, whenever the digraph xy
represents the diphthong /αβ/, then x and y should represent the
short vowels /α/ and /β/ respectively; and whenever xx represents the long vowel /α:/, then x should represent the short
vowel /α/. (We have used here α, β to denote both short vowels and
related diphthong components.
Sometimes the IPA notation uses slightly different shapes for that
purpose, e.g. /ɪ/, /ə/, /ɒ/ but /i:/, /ɜ:/, /oʊ/, not /ɪ:/, /ə:/, /ɒʊ/, indicating that the respective pairs
differ not only in length but also in quality.) The principle stipulates that if we
represent say /ɒ/ by o (as in not) and
/ʊ/ by u
(as in put), then we always do that including in the diphthong /oʊ/ which is always represented by ou as in BR nou, bout,
soul, tou (TS no, boat, soul, toe). Unlike certain other languages or dialects
such as Estonian, Finnish, Dutch, German, Frisian or Lombard, the traditional English orthography does
not normally use double letters for the long vowels. The digraphs ee
and oo are exceptions, and they represent /i:/, /u:/ and /ʊ/ (as in feel, mood and book), not /e:/
and /ɔ:/ as it would have been the case if the context freeness principle
explained above were applied. In the case of digraphs used for diphthongs, TS violates the context
freeness principle e.g. by the use of ai, ea
for /eɪ/ in main, break. Indeed, the context freeness principle, and
ai representing /eɪ/ would have implied that a represents
/e/ which is not the case. Besides, ea representing /eɪ/ would have implied by the context
freeness principle that a represents /ɪ/, which is not the case. Similarly, TS violates that principle by the use of ie, ei for /aɪ/ in tie, either,
because i and e do not represent /æ/ or /ʌ/; or by the use of oa, oe for /oʊ/ in boat, toe since a and e do not represent /ʊ/. 2.5. Universality
Principle: Spell short
vowels and consonants in a way that is common for the traditional orthography
of most Romanized languages including English. English spells consonants generally in a way
common for the traditional orthography of most Romanized languages, with few
exceptions like /ʃ/, /tʃ/ and /dʒ/.
Whereas most Romanized languages are likely to pronounce a, e, i, o, and u as in pasta, ballet,
police, depot and tabu,
the corresponding English short vowels are as in cat, pet,
big, fog and put.
The universality principle however refers to spelling not
pronunciation. We shall discuss the
spelling of English vowels in greater detail below, drawing a comparison
between the orthographic systems of BR [4] and Interspel [11]. We apply the above five principles to build from scratch the
re-Romanization system of Basic Roman Spelling of English (BR) [4]. 3. Building BR Orthography from Scratch The Basic Roman Spelling is aiming at a reasonably precise
approximation of Spoken English, for which purpose we use 48 phonemes comprising
the set of 45 English phonemes from [2], plus two rhotic
variant phonemes, plus the non-English consonant /ts/.
This system is not to serve some particular standard of English
pronunciation, but rather provide the means that could be used for the
spelling of different varieties of English. 3.1. Short
vowels In accordance with the strict Romanization, consistency, proportionality
and universality principles, we represent the short vowels /æ/, /e/,
/ɪ/, /ɒ/, /ʊ/ by a, e, i, o, u as in BR dam, net, big,
hot, put (TS dam, net, big, hot, put) respectively. The proportionality principle dictates that this same set of letters be used for the representation of the short vowels /ə/ and /ʌ/ as well. In order to facilitate disambiguation we choose a for /ə/, for /æ/ is more consistently reduced to /ə/ in unstressed syllables. The vowels /e/, /ɪ/ and /ɒ/ are not reduced to /ə/ e.g. in /endʒɔɪ/, /ɑ:tɪst/, and /ɒtɒrɪtɪ/, hence the representation of /ə/ by e, i or o would have increased ambiguity. As for the possible representation of /ə/ by u, in view of the context freeness principle that choice would have created the ambiguity of both the diphthong /ʊə/ and the long vowel /u:/ being represented by uu, as in tuu representing both /tʊə/ and /tu:/ (TS tour and too). No such ambiguity arises with the chosen representation of /ə/ by a as there is no /aə/ diphthong in English. Given the intermediate position of /ʌ/ between /æ/ and /ə/, and the
already fixed representation of both /æ/ and /ə/ by a, this dictates that /ʌ/ be represented by a as
well. Therefore, we represent the
three short vowels /æ/, /ʌ/ and /ə/ by a as in BR hav,
dast, ahed (TS have, dust, ahead). Let us stress that while the BR spelling
does not distinguish between these three short vowels, a is still
pronounced differently in these three words; namely, BR hav,
dast, ahed are pronounced /hæv/, /dʌst/ and /əhed/ respectively. (The more
precise RPA alphabet in Section 6 below has different presentations for /æ/, /ʌ/ and /ə/.) BR spelling reflects the spoken
language as the latter is, without hinting at, suggesting or advocating any
pronunciation distortions whatsoever.
In the case of homographs like e.g. BR spelling hat for both /hæt/ and /hʌt/ (TS hat and hut), the relevant
word and pronunciation would be differentiated from the context of the
sentence. 3.2. Consonants In accordance with the strict Romanization, consistency and universality
principles we represent the consonant sounds /b/, /tʃ/, /d/, /ð/, /dʒ/, /f/, /g/, /h/, /x/, /k/, /l/, /m/, /n/, /ŋ/, /p/, /r/, /s/, /ʃ/, /t/, /θ/, /ts/, /v/, /w/, /j/, /z/, /ʒ/ by b, ch, d, d, dzh, f, g, h, h, k,
l, m, n, ng, p, r, s, sh, t, t, ts, v, u, y, z, zh respectively. The English dental fricative consonants /θ/ and /ð/ (as in think and this)
are somewhat problematic for many non-native English speakers who tend to
pronounce /θ/ as /t/ or /s/, and /ð/ as /d/ or /z/. (A similar merger of /θ/ with /f/, and /ð/ with /v/ occurs in native English
varieties such as Cockney, Newfoundland English, African American English,
and Liberian English, making inroads into Estuary English too.) At the same time, while native English
speakers tend to distinguish /θ/ and /ð/ in their spoken language, they write both
with the same digraph, th and increasingly not
caring about the spoken distinction.
While we have opted to represent them in BR by t
and d respectively, in Section 4 below the system is extended to
differentiate between these consonants. 3.3. Long
vowels and diphthongs The representation of long vowels and
diphthongs is obtained by a straightforward application of the
proportionality and context freeness principles. In the case of long /ɪ/ we take y instead of i as a second
letter, following the pattern of the diphthongs /aɪ/, /eɪ/ and /ɔɪ/. Namely, we represent /ɑ:/, /i:/, /ɔ:/, /u:/, /ɜ:/ by aa, iy, oo, uu,
aa as in BR faam,
fiyl, soo, muud, baaning (TS farm, feel, saw, mood,
burning), and /aʊ/, /aɪ/, /eə/, /eɪ/, /ɪə/, /oʊ/, /ɔɪ/, /ʊə/ by au, ay, ea, ey, ia, ou,
oy, ua as in BR nau,
tray, hea, wey, dia,
lou, voys, pua
(TS now, try, hair, way, dear, low, voice, poor)
respectively. The initial impression of BR spelling may
well be one of inner city talk, unusual, and in any case un-TS which of
course it is. BR spellings like fiyl, soo, baaning, tray, hea,
wey, dia, and lou (TS feel, saw,
burning, try, hair, way, dear, and low) may well appear either
too dissimilar to present TS forms or confusing in their similarity to
contradictory TS conventions. This is
the place to remind and stress that BR derives from Spoken English, and from
traditional spelling patterns in the wider family of Romanized
languages. Furthermore, BR is self
contained; it neither derives from TS, nor is it designed with a view to a
step-by-step transition from TS to some reformed English spelling. Similarity to TS is sought at the basic level only, when choosing the
representation of consonants and vowels as in the case of sh, ch, y (and j, w in Section 4 below). Once that representation is fixed, then
because of the inconsistent nature of TS the two orthographies could be
expected to be confusingly contradictory in many cases. We are not concerned about that, for BR is
intended for independent usage rather than in combination with TS; texts in
BR are certainly not supposed to be read as if written in TS. For instance, once we fix o for /ɒ/ as in BR boks (TS box),
then we use oo for /ɔ:/ as in BR soo (TS saw), not
bothering that TS uses oo for /ʊ/ or /u:/ instead, as in book and mood. The obtained BR system uses 22 Roman letters
(the letters j, q, w and x are not used), with no special characters
or diacritical marks. The chosen
representation of short vowels, together with the derivative representation
of long vowels and diphthongs it entails, contributes most to shaping the
characteristic features that distinguish BR from TS, Interspel,
and other orthographic systems such as those discussed in Section 7 below. 4. Extensions and variants 4.1. Rhotic variety Rhotic dialects are accommodated by appending r to the relevant non-rhotic graphemes, so that /ɚ/, /ɝ/ are rendered by ar and aar respectively [4], as in BR tiycha, paasiyv becoming tiychar, paarsiyv (TS teacher, perceive). 4.2. Dental
fricatives Traditional English spelling represents both
dental fricative consonants /θ/ and /ð/ by th as in think and this,
and dialects may vary in which words are pronounced with what. BR differentiates between the two, but
spells /θ/ same like /t/, and /ð/ same like /d/. Seeking to expand BR as near
to one-to-one phoneme-grapheme correspondence as possible, one may
consider a version of the system using th for /θ/, and dh for /ð/ [4], as in think, dhis instead of BR tink,
dis (TS think, this). 4.3. w
vs. u The BR system could also be extended by
spelling the consonant /w/ (as in we, queen) by w instead of u [4]. This extended system provides for a
possible variant using uw instead of uu for /u:/ as in muwd instead of BR muud (TS mood), which however we do not take as standard. 4.4. j
vs. dzh The
BR orthographic system could be further extended by using the grapheme j
instead of dzh for /dʒ/, as in joy instead of
BR dzhoy (TS joy) a compact and
traditional English spelling pattern [4]. We call Extended Basic Roman Spelling of
English (EBR) the 24-letter system (the letters q and x are not used)
obtained from BR by incorporating all the extensions given in 4.1-4.4. Therefore, we consider two systems here: the
simpler BR, and the more elaborate EBR.
The former is the system originally introduced in [4], while the
options for an extended system were discussed in that work too. The better choice between these two systems
would derive from ones preference for the precision of the system, or for
its simplicity in using the available Roman letters and letter combinations
instead. 5. Phonemicity While the present approach is essentially
phonemic, the introduced orthography falls short of establishing a one-to-one
correspondence between phonemes and graphemes; hence it could be described as
semi-phonemic at word level. It
indicates the approximate rather than precise pronunciation of individual
words. In addition to the homophones
now receiving identical spelling, more homographs are created by the
presentation of /æ/, /ʌ/ and /ə/ by one and the same letter. At textual level however, readers could
retrieve the relevant word from among several homographs by taking into account
the context of the sentence.
Therefore, writing in BR is context-free, while reading is
context-dependent. This property may
possibly allow for the automated conversion of texts from BR into traditional
spelling. 6. Roman Phonetic Alphabet for English The BR orthography is close enough to one-to-one phoneme-grapheme correspondence, which makes it possible to engender one by means of a minor adaptation. We start from the full extended system EBR, then add stress marks as appropriate for a transcription system, and use them both to indicate stress and to disambiguate homographs as follows. As shown in the table below, we take the unstressed a to represent /ə/, and use two primary stress marks and , and two secondary stress marks , and ,,, with a in syllables stressed by or , representing /æ/, and a in syllables stressed by or ,, representing /ʌ/. This convention is extended to the two relevant long vowels /ɑ:/ and /ɜ:/ too, taking into account that the latter may occur in unstressed as well as stressed position. Namely, we take aa in syllables stressed by or , to represent /ɑ:/, and aa in syllables that are either unstressed or stressed by or ,, to represent /ɜ:/. In short, the set of stress marks , , is used in the case of /æ/ and /ɑ:/, while the stress marks , ,, are used in the case of /ʌ/ and /ɜ:/. Stress marks are placed before the syllable concerned (i.e. not necessarily next to the relevant vowel as in the table below). For instance, the BR homographs hat (IPA /hæt/, TS hat) and hat (IPA /hʌt/, TS hut) are differentiated now to become RPA / |