Road sign romanization in Oman
The linguistic landscape close-up
Rafik Jamoussi and Thomas Roche
Sohar University / Southern Cross University
Throughout the Arab Gulf States, bilingual road signs are the norm, employing
both Arabic and a romanized counterpart for the large expatriate population.
The existing romanization is inconsistent, with potentially misleading variant
spellings of place names signposting the region. This study provides a linguistic
analysis of signs on the arterial road running from Muscat, Oman, toward Dubai
in the United Arab Emirates in an effort to identify linguistic factors leading to
discrepant renderings. The authors identify the wavering between transcription and transliteration and the hesitancy between the local and the standard
language varieties as the main sources of discrepancies. These findings constitute
the groundwork for initiatives intent on addressing the situation. Guidelines
are provided for policy makers indicating the various aspects remedial work on
signage in Oman should take into account.
Keywords: romanization, Arabic script, road signs, language policy,
transcription, transliteration
1.
Introduction
In toponymy, geographical names are usually used to refer to particular places
or features (see for instance United Nations Group of Experts on Geographical
Names, UNGEGN, 2006). Their management is of major importance to state planning organizations and has attracted considerable attention from national and
international organizations such as the United Nations, the United States Board
on Geographic Names (BGN), and the Permanent Committee on Geographical
Names (PCGN) in the UK. In instances where the resulting documentation is to
be made available to a non-native audience, the name management process involves translation. When the Latin alphabet is used, this is often referred to as
Latinisation, or more frequently Romanization (Aboelezz, 2009; Beesley, 1998).
Australian Review of Applied Linguistics 40:1 (2017), 40–70. doi 10.1075/aral.40.1.04jam
issn 0155–0640 / e-issn 1833–7139 © John Benjamins Publishing Company
Road sign romanization in Oman
Standardising name management is of major importance, and this is even
more the case when romanization is involved; in fact, standardization is the very
raison d’être of the organizational focus on romanization. The list of advantages
of standardization can be quite long (UNGEGN, 2006), but special emphasis is
usually put on the negative impact on the credibility of documents with issues
of consistency and accuracy (Albin, as cited in Al-Nabhani, 2007). For example,
in preparation for the 2008 Olympic games, the Chinese authorities launched in
2005 the “Use Accurate English to Welcome the Olympics – Public Bilingual Sign
Standardization Drive” campaign which aimed at “improving the country’s international image and enhancing public awareness about using correct English by
identifying and correcting bad translations on bilingual public signs and printed
materials’ (China Education and Research Network, 2005). Recognizing the importance of geographical names, and the necessity to have accurate documentation, the first UN conference on the Standardization of Geographical Names, held
in 1967, recommended “that efforts should be made to arrive at an agreement on a
single romanization system, based on scientific principles, from each non-Roman
alphabet or script, for international application” (UNGEGN, 1967, “Resolution
No. 9”; see Spolsky, 2009, p. 242.). In compliance with this resolution, Arabicspeaking countries at the Beirut conference in 1971 established the Arabic romanization system, the amended version of which was adopted by the Arab League
and the UNGEGN (United Nations Group of Experts on Geographical Names) in
1972 (see Atoui, 2012 for a review of the main developments leading to the adoption of the Beirut system and the issues that still affect its implementation; see
also Al-Nabhani, 2007, p. 71; Zagórski, 2010). The UNGEGN effort hence represents much of the framework within which research into romanization in the Arab
world is carried out (see inter alia Aboelezz, 2009; Atoui, 2012; Khraish, 2008; see
also work undertaken by the United States Board on Geographic Names).
Follow up on these UN recommendations in the Sultanate of Oman is ensured by the National Survey Authority (NSA). Place name management research
focusing on Oman has already been undertaken (see Al-Nabhani, 2007; Kharusi
& Salman, 2011), with most issues addressed therein applicable to many of the
neighboring Arab Gulf States.
Being but a subcomponent of the field of place names, road signs have not
attracted much specific attention in research. Road signs, however, do require special attention, due to the importance of their infrastructural function. In the literature on road signs, there is emphasis on two aspects; simplicity, that is ridding
the system of uncommon Roman-based graphemes such as diacritics (e.g., the
circumflex accent "û" used to disambiguate phonemes in some Latin based alphabets as in Turkish and French) and features that are only important in technical
descriptions; and, recognizability whereby the sign is identifiable by a local when
41
42
Rafik Jamoussi and Thomas Roche
pronounced by a foreigner, also defined as reversibility, that is, when a transcription can be traced back to its original word (Kharusi & Salman, 2011; UNGEGN,
2003).Within the context of the Sultanate of Oman, road sign bilingualism is a
reflection of the multilingual situation of the country1 as well as of the will of the
authorities to cater for both Arabic and non-Arabic speaking communities (see
Sloboda, Szabó-Gilinger, Vigers, & Šimičić, 2010, for an investigation of the significance of bilingual signage and the interaction dynamics of actors involved in
the bilingualization process).
The romanized component on these road signs, however, is often affected by
inaccuracies and inconsistencies that undermine the initial intention to provide
assistance to those foreign communities. Present signage frequently includes two
or more variations of the same place name (Table 1), resulting in significant confusion for non-Arabic speaking travellers and workers in the country (see Jamson,
Tate, & Jamson, 2005, for an investigation of the driving and safety implications
of bilingual road signs). The existing linguistic mélange suggests a standardization
issue in the management of those road signs.
The present project is thus motivated by a need to address these inaccuracies
through an examination of the existing signage and a categorization of the different
structural issues affecting those inaccuracies. This categorization may be used, at a
subsequent stage, to generate a romanization system that bridges existing lacunas.
2. Methodology
The project’s initial phase consisted of constituting a corpus of road signs along
the Muscat-Dubai road with an incursion of approximately 1 kilometre on both
sides of each roundabout along that road. Over 660 photos representing 862 entries were thus collected between December 2012 and January 2013. The photos
were indexed, watermarked and uploaded onto a web album. For ease of referencing, all photos were taken with geographical coordinate tags allowing for an automatic locating of each sign on a web map, such as Google Earth© or Picassa©.2
Named CAHRS (Corpus of Al-Batinah Highway Road Signs), the corpus, was
analyzed with a combination of spreadsheet and database tools. Each entry in the
corpus consists of the Arabic place name, the romanized counterpart, a phonetic
1. According to the 2013 census, the expat population in Oman has reached 1,684,625, i.e. almost 44% of a total population of 3,831,553.
2. The web album is accessible at the following address: https://picasaweb.google.
com/104731972347591208298/ProjectORGEHR12003?authuser=0&authkey=Gv1sRgCMDP
wv-8-q_hzgE&feat=directlink
Road sign romanization in Oman
transcription, a hyperlink to the multimedia file that merges both the photo and
an audio file, and a series of analyses that track the variations in the romanization
of different phonetic and morphological features.
The use of the spreadsheet allows for different layers of filtering that help identify inconsistencies and other phenomena that affect the corpus. The database tool
was used to generate basic statistics that detail the extent and variety of the transcription features.
3. Data analysis
The approach adopted for data analysis is primarily descriptive in nature, with the
implication that the analysis does not sort findings into “correct” and “incorrect”
categories. The research does not presuppose an already established evaluative system against which findings are appraised. From a strictly linguistic perspective,
any instance of romanization can be considered an acceptable representation of
the original sign so long as it succeeds in assuming two interrelated functions:
–
–
A signifier for the sign it represents (the signified)
An acceptably close representation of the original (for referenceing and communication with the local population, see inter alia Kharusi & Salman, 2011)
Most signs in the CAHRS database satisfy the second criterion. Except for rare
cases (e.g., Mehat Bani Kueem (P506) and Mehat Bani Queem (P507) for the
Arabic “‫[ ”ميحة بني كيوم‬mi:ħət bani: kəjju:m]) all romanizations in the CAHRS corpus are phonologically close enough to enable recognition by the local population
as representations of specific names. For instance, all of the following alternatives
from the corpus, Ruwilat (P339), Rwilat (P340), Rewilat (P341), Ruwailat (P353),
Ruwaylat (P605), and Rwailat (P607), are recognisable representations of the runon instance “‫”رويلة‬, [rəwəjlat].3 It is only in relation to the first criterion that these
alternatives become problematic. From the perspective of a non-Arabic language
user, these alternatives can be understood as signifiers of different signifieds (i.e.
different place names) instead of variants of the same signified. Consequently,
mistakes aside, the different representations of the same word or morpheme or
sound will all be referred to here as variants.
3. Numbers with the format (P + three digits) refer to the reference numbers given to the entries in the database. Italicized names refer to the romanized entries in the database.
43
44
Rafik Jamoussi and Thomas Roche
3.1 Word level variations
At this level, the analysis considers inconsistencies appearing in different occurrences of the same names. The corpus consists of single-word and multiple-word
entries, each of these can be a single occurrence or an instance of multiple occurrences (duplicate entries). A total of 377 words have been identified. Out of these
entries, 72 represent single occurrences, that is, entries that occur only once in the
corpus. These have been eliminated from the word-level variation analysis. Of the
remaining names, the number of variants ranges from 2 to 6, with “‫[ ”رويلة‬rəwəjlat]
representing the entry with most variants in the corpus.
3.2 Sound level variation
Word-level inconsistencies are due to morphosyntactically and sociolinguistically motivated variations in the representations of sounds or other features within
these words. The analysis will now turn to these.
Table 1. Sample multiple entry words with their variant distribution
No. of
No. of entries Example
variants in this category
Transcription
Variants
2
‫األبيض‬
əl ʔəbjəd
Al Abyad, Al Abiyad
‫املنطيفة‬
əl mint̪ ˤi:fa(h)*
Al Manteafah, Al Mintayfah
‫الصويحرة‬
əsˤwi:ħra(h)
A ‘Swaihrah, A’Suwaihrah,
A’Suwaihra
‫الغريفة‬
lə ɣrəjfa
Al – Gharifah, al Ghurayfah,
Al-Gharifah
‫الخابورة‬
əl xa:bu:ra(h)
al Khaburah, Al Khabourah,
Khaborah, Al Khaboorah
‫الهاملية‬
əl ha:milija
AL Hamliyah, Al-Hamiliyah,
Al Hamaliya, Al–Hamliyah
3
4
89
15
8
5
2
‫املصنعة‬
lə msˤnʕa(h)
al Musanaah, Al Musna’ah,
al Musanna, Al Masanah, Al
Musnaah
6
1
‫**رويلة‬
rwe:lat
Ruwailat, Ruwaylat, Rwailat,
Ruwilat, Rwilat, Rewilat
* Bracketed sound refers to the light version of the feminine case marker (see Section 3.5 below). The
authors resolved to adopt the brackets following indecision and disagreement between informants over its
enunciation.
** This entry occurs in a run-on context
Road sign romanization in Oman
3.2.1 Consonant sounds
Arabic has a larger repertoire of consonant sounds than Latin/English (Table 2),
which means that tradeoffs are necessary for those sounds that are not available in
the Latin system. In the absence of clear guidelines, these are expected to yield a
larger number of variants.
3.2.1.1 Common consonant sound variation. The analysis of the romanization
of common consonant sounds (Table II)4 shows results that consistently adhere to
the above hypothesis as most sounds are represented with one variant only. Three
sounds derogate to this rule:
–
–
The voiceless velar stop “‫ ”ك‬/k/ has two variants; “k” (96.15%), as in Hillat Al
Kahahil (P293) and less frequently “q” (3.84%) as in Mehat Bani Queem (P507).
The voiced alveopalatal affricate “‫ ”ج‬/j/ has 3 variants; “j” (88%), “g” (9.33%),
and “q” (2.66%).
Glottal
Pharyngeal
Uvular
Velar
Palatal
Alveolar
Interdental
Labiod ental
Labial
Table 2. Phonemic chart of MSA consonants (from Ryding, 2005: 13) with phonemes
lacking an equivalent in Roman script shaded
Stops
Voiceless
Voiced
K‫ﻙ‬
t‫ﺖ‬
b‫ﺐ‬
ɔ
‫ﺀ‬
d‫ﺪ‬
Affricates
Voiceless
Voiced
j‫ﺝ‬
Fricatives
Voiceless
f‫ﻑ‬
Voiced
Nasals
m‫ﻡ‬
th ‫ﺙ‬
s‫ﺲ‬
dh ‫ﺫ‬
z‫ﺰ‬
n‫ﻥ‬
Laterals
l‫ﻝ‬
Flaps
r‫ﺭ‬
Semivowels
(approximants)
w‫ﻭ‬
sh ‫ﺵ‬
Y‫ﻱ‬
4. Tables with Roman numerals are situated in the appendix.
h‫ﻩ‬
45
46
Rafik Jamoussi and Thomas Roche
–
The palatal semi vowel “‫ ”ي‬/y/ has three variants; “y” (72.15%), as in Al Abyad
for “‫”األبيض‬, [əl ʔəbjəd] (P015), “i” (13.92%), as in Dubai for “‫”ديب‬, [dbaj], otherwise the sound is dropped leaving just the vowel that accompanies it (11.39%),
as in Seih Al Taibat for “‫”سيح الطيبات‬, [si:ħ al t̪ ˤaiba:t].5
The glottal stop Hamza (or Hamzah) “‫ ”ء‬is thought to deserve specific attention.
This consonant sound is available in European languages and/or dialects, such as
Cockney English. While it is usually phonemically represented through the sign
/ʔ/, as in ‘little bottle’ [liʔl boʔl], it has no corresponding sign in the English alphabet (see inter alia, Abu-Chacra, 2007; Ryding, 2005; UNGEGN, 2003).
The analysis of the database shows a consistent dropping of the Hamza when it
is in word-final position, as in Wadi Rajma (P649) for “‫”وادي رجامء‬, [wa:di rəʒma] (34
cases out of a total of 35).6 This tallies with the fact that the letter in this position is
usually elided, except in the case of running speech standard Arabic.
When appearing in other positions, the Hamza is pronounced. However, analysis of the database shows that it is equally dropped in the romanization leaving
just the vowel of the syllable (Table III). The database contains only one instance
(out of 32) where the medial Hamza is represented through an apostrophe: (Al
Kaza’in (P058) for “‫”الخزائن‬, [ləxza:jən]).
3.2.1.2 Specific consonant sound variation. As expected, the analysis of this
group of sounds shows less consistency, with up to 4 variants per consonant sound
(Table IV). It equally reveals that despite the avoidance of diacritics, there is a clear
attempt to intimate that a sound is specific to Arabic through the combination of
different consonant letters, thus forming digraphs as in “kh” for “‫”خ‬, the voiceless
velar fricative /x/, or “gh” for “‫”غ‬, the voiced velar fricative /ɣ/.
The voiced pharangeal fricative “‫”ع‬, /ʕ/, is only represented in about 20% of the
cases through the use of an apostrophe.
3.2.2 Vowel sounds
Arabic has three basic short vowels represented through diacritics appearing
above or below the consonant letter they accompany (Abu-Chacra, 2007; Eades,
5. It is interesting to note that a number of cases show a dropping of the vowel preceding the
semi vowel when it is a [i] (P103) Al Saifya for the Arabic “‫”السيفية‬, [əssi:fijja(h)] (as opposed to
cases where this does not happen: (P130) AL Uqdah A’Sharqiyah for the Arabic “‫”العقدة الرشقية‬,
[l ʕəqdə əʃʃərqijja(h)]).
6. The one case where a word-final hamza is not dropped is Humairah (P297) for “‫”حمرياء‬,
[ħmi:ra], where the hamza is rendered through an “h”.
Road sign romanization in Oman
2009; Ryding, 2005; Watson, 2002). A fourth diacritic sign, “◌ْ ”, indicates a sukun
or absence of the vowel.
Table 3. Phonemic chart of MSA vowels (Ryding, 2005, p. 25)
Front
High
Central
i/ii ‫ﻱ‬/‫ﹺ‬
Back
u/uu ‫ﻭ‬/‫ﹸ‬
Mid
Low
a/aa ‫ﺍ‬/‫ﹶ‬
These diacritics are not part of the Arabic alphabet and their script representation, called vowelization or vocalization, is only required for language teaching
and assessment purposes or in Qur’anic texts (Abu-Chacra, 2007; Ryding, 2005).
Still, the reading process requires restoring at least some, if not all, of these vowels through an inference process that relies on a combination of contextual syntactic and semantic clues (Abu-Rabia, 1997). The vowel-restoring process can be
straightforward when a single alternative is morpho-syntactically evident, as in “‫”آ‬
or the case of the Hamza where the vowel affects the transcription: “‫”أ‬, “‫”إ‬, and “‫”ؤ‬.
When the context accommodates a larger number of alternatives7 the vowelization process becomes more arduous, especially for learners of Arabic as a foreign
language (Zagórski, 2010).
Arabic transcriptions within the CAHRS corpus of road signs adhere to the
principle of non-vowelization; rare exceptions are usually used for disambiguation purposes, i.e. when alternative vowelizations produce different words as in
“‫( ”شارع ال ِعلم‬P653), where “‫ ”علم‬can be interpreted as [əlʕalam] “flag” or [əlʕilm]
“knowledge”. Since road signs are made up of single words or, at most, phrases,
the morpho-syntactic context is reduced to a bare minimum and when different
vowelization alternatives exist, they usually relate to the domain of sociolinguistics
as in the adoption of standard Arabic or a spoken variety. For instance, the sign
“‫ ”الغيزين‬can be interpreted as [əlɣi:zi:n], as in Al Gizen (P165), or [əlɣi:zain], as in
Al Ghizayn (P405) (see discussion).
The absence of vowelization in the Arabic corpus implies that the romanization is bound to be more explicit, and equally more restrictive, as it involves opting for only one of the alternatives available in the non-vowelised Arabic transcription. Because of this restrictive nature, romanization reflects a number of
language management decisions that need not appear in the Arabic transcription
(see discussion). Another implication is that the absence of vowelization deprives
the corpus of a script-based reference point for mapping and analysis along the
7. Beesley (1996) speaks of an average of almost five morphophonologically valid analyses per
word. See also Zagórski (2010).
47
48
Rafik Jamoussi and Thomas Roche
lines of the investigation undertaken for consonants. For this reason, the analysis
of vowels in this study has taken as reference point the pronunciation provided by
informants adopting the spoken language varieties.8
3.2.2.1 Short vowels. The romanization of the three vowels adheres in the main
to the MSA rendering with “a” for the “fatha” َ (99.38%), “u” for the “thamma” َ
(90.62%), and “i” for the “kasra” ِ (77.58%) (Table V). Other renderings are considered dialectal or even idiosyncratic interpretations (see Watson, 2002: 21–2), as
in Al Khushdah (P060) for “‫”الخشدة‬, [əl xəʃda(h)].
3.2.2.2 Long vowels. Long vowels are represented in Arabic script through the
adding of the “‫”ا‬, “‫”و‬, and “‫ ”ي‬letters, respectively after the letter carrying the fatha,
thamma, and kasra (Abu-Chacra, 2007: 21).
The analysis of the corpus shows a tendency in romanization to reduce these
long vowels into short ones (over 88%). Thus, the long fatha is predominantly represented through an “a” (99.52%), the long thamma is mainly represented though
a “u” (52%), but equally through the “o” (15.2%), and the long kasra is represented
through an “i” (76.56%), with some instances featuring an “e” (2.6%).
Attempts to reproduce the length of the vowel feature the doubling of the same
vowel sign (“oo”, “ee”), Al Khaboorah for ‫[ الخابورة‬əl xa:bu:ra(h)] (P169), Hillat AL
Jood for ‫[ حلة الجود‬ħalat əl dʒu:d] (P279), Ajeeb, for ‫[ عجيب‬ʕadʒi:b] (P434), or reinforcing it through adding another vowel or consonant (“ou”, “ow”), Al Khabourah, for
‫[ الخابورة‬əl xa:bu:ra(h)] (P057), al Huwqayn, for ‫[ الحوقني‬əl ħu:qi:n] (P056), (Table VI).
3.2.3 Diphthongs
The two available diphthongs in Arabic, /ai/ and /au/ (Kharusi & Salman, 2011;
Watson, 2002), may represent phonological aspects or may assume a grammatical
function (e.g., the dual causative case as in “‫”الحيلني‬, [lħi:li:n], (P054)). They equally
represent a salient feature of the standard variety and typically tend to be monophthongised in spoken varieties, such as the dialects spoken in Oman (Eades, 2009;
Holes, 2009; Watson, 2002) and turned into long vowels (see Section 3.6 on dialectal variation).
The corpus shows that diphthongs are maintained in 79.77% of the cases, for
example, A ‘Swaihrah (P395) for “‫”الصويحرة‬, [əsˤwi:ħra(h)]. The romanization here
makes use of the following digraphs in descending order of frequency: “ay”, “ai”, “ei”,
and “ey”. The majority of the remaining cases (17.97%) tend to use the long vowel,
“ee”, or short ones, “i”, “e”, and “y”, all taken here as representing the spoken variety,
8. The Batinah coast region through which the Muscat-Dubai Road passes is recognized as a
mixed dialectical area (see Holes, 2009).
Road sign romanization in Oman
e.g. Khadrawin (P307) for “‫[ ”خرضاوين‬xad̪ ˤra:wi:n]. These results show a clear inclination for the adoption of the standard Arabic variety in romanization (Table VII).9
3.2.4 The sukun
Sukun, or absence of the vowel, represents a major feature, especially for the input it
brings in the contrast between the MSA and spoken varieties whereby vowels appearing in a pronunciation along the MSA lines tend to be deleted (replaced by sukuns)
in dialectal renderings (Holes, 2001). The analysis of the corpus shows that where
there is a choice between a sukun (spoken variety) and a vowel (MSA), the vowelized
alternative is chosen in almost 85% of the cases.10 Included here are segments that
follow the article “al” (as with [x] in “‫”الخويرات‬, [ləxwi:ra:t], romanized as in Al Kuwerat
(P062)) or medial segments (as with [s] in “‫”وادي الحواسنة‬, [wa:di əlħawa:snə(h)], romanized as Wadi Al Hawasinah (P0526). This is a clear indication of the preference
for MSA rather than the regional dialect in the romanization process.
3.3 The definite article
The definite article in Arabic is represented through the combination of the alif
lam sounds or “al”. Depending on the initial letter of the word that follows, this “al”
is either pronounced fully or assimilated with the segment that follows it, resulting
in idgham; a dropping of the “l” combined with a doubling of the initial consonant
sound of the following word (Abu-Chacra, 2007).
3.3.1 The non-assimilated definite article
With 344 records, 4 variants have been recorded, the most statistically significant
being “Al” (77.9%), i.e. where use is made of the initial upper case and the article
is kept separate from the name, as in Al Fahgrah (P015) for “‫”الفاغرة‬, [əl fa:ɣra(h)]
(Table VIII).
3.3.2 The assimilated definite article
The most recurrent variant here is “A’ ” (uppercase “a” followed by an apostrophe,
either separated from the following word or attached to it) with 62.75% of a total
of 196 cases, as in A’ Dammos (P066) for “‫”الدموس‬, [əddəmmu:s]. Next in terms of
9. Monophthongising diphthongs is an attested feature of some dialectal varieties of the Middle
East. See Eades (2009), Holes (2001).
10. The investigation of this aspect discarded established sukun cases, which belong with the
MSA morphology of the word, as with the [s] in “‫”الرستاق‬, [ʔərrusta:q] (P076), or in “‫[ ”أسود‬aswad]
(P014). Equally discarded are cases associated with the voiced pharyngeal fricative “‫”ع‬, /ʕ/, as
this sound is often dropped together with its vowel, which may affect the counts.
49
50
Rafik Jamoussi and Thomas Roche
frequency is the use of assimilation, “A” followed by the doubling of the initial
letter of the next word (16.83%) as in “As’Sawaba” in Hellat As’Sawaba (P286) for
“‫”حلة السوابع‬, [ħəllət əssawa:bəʕ]. Instances where assimilation is not applied represent
16.83%11 (Table VIII).
3.4 Gemination
Gemination, or consonant elongation, is the doubling of a consonant sound (AbuChacra, 2007). It is represented through the diacritic “◌ّ” placed over the affected
consonant letter, this feature is treated here as distinct from the gemination that
occurs with the assimilated definite article “al”. The romanization of gemination is
characterized as follows in the database (Table IX):
–
–
For all sounds except the semivowel “‫( ”ي‬67 records), 94% of the gemination
cases have been represented through the doubling of the consonant letter, as
in Sallan (P397) for “‫”نلاص‬, [sˤəlla:n].
For cases of gemination occurring with the semivowel “‫( ”ي‬39 records) only
3 instances (7.69%) feature a doubling of the letter “y”. This happens in a
word final context as in Hayy As Salam, ‫[ مالسلا يح‬ħaj ɪssala:m] (P298). For
the rest of the cases, either a non-geminated sound is represented through
“y” (74.35%), or the sound is not represented at all (17.94%); Al Saifya for
“‫”ةيفيسلا‬, [ʔəsi:fi:a(h)] (P103).
3.5 Feminine case
The feminine case is represented in Arabic through a word final “‫”ة‬, /t/, followed by
an inflectional vowel. However, this feature is rarely pronounced outside Modern
Standard Arabic (MSA) and is almost solely confined in that register to instances
of running speech. If not dropped altogether, a light /h/ sound is used in some
spoken varieties (Abu-Chacra, 2007; Ryding, 2005).
In our corpus (Table X), 86.72% of 212 records of single-word hits have an “h”
to represent this silent letterfeature, as a reference to the light variant mentioned
above, as in Al Fahgrah for “‫”الفاغرة‬, [əl fa:ɣra(h)] (P015), compared to only 13.27%
of cases where the letter is dropped; al Ghashba (P002) for “‫”الغشبة‬, [əlɣaʃba(h)].
Most cases occurring in run-on contexts carry a representation of the feminine marker. This is a “t” or an “h” depending on whether the word bears a
11. This citation form alternative (Wells, 2006) is advocated by Kharusi and Salman (2011),
though they further restrict it to the lower case mainly for database compilation constraints.
Road sign romanization in Oman
definite article or not. Only 2 cases (6.66%) in this category do not represent the
feminine marker.
The plural feminine in Arabic is equally indicated through a /t/ – though this
is graphically different from the one representing the singular feminine case: “‫”ت‬.
This feature is consistently represented by a “t” in our corpus.
3.6 Dialectal variation
Two dialectal sound shifts are present in the database, both minimally. The [g] replacing the voiced alveopalatal affricate /j/ represents 9.33% of the cases (Table XI),
as in Al Higra for “‫”الحجرة‬, [əl ħigra(h)], (P593). The [g] replacing the voiceless uvular stop “‫”ق‬, /q/, is featured in a single instance out of a total of 153 (0.65%). This
is Sur Al Bogryn (P347) for “‫”سور آل بوقرين‬, [su:r ʔa:l buqri:n], (compare with Sur Al
Boqrayn (P348) (Table IV)).
3.7 Romanization or translation
The discussion of the romanization of road signs is based on the assumption that
these are proper names (UNGEGN, 2006) that have no function other than that of
referring to a place (hence, place names). This principle should apply even when the
name in question serves as a content word outside its function as a proper name.
Hence, (P393) “‫( ”حديقة النسيم‬Breeze Garden), and (P379) “‫( ”شارع البهجة‬Elation Street)
are place names containing words referring originally to attributes or objects. Still,
they are handled as proper names, A’ Naseem Garden and al Bahjah Street respectively. On the other hand, words that serve primarily as content words should be
translated, as in “street” and “garden” in the two examples above, or in (P367) “‫سوق‬
‫( ”الجمعة‬Friday Market). The database, however, shows instances that deviate from
this rule: As Sumhan South (P116) for “‫”الصومحان الجنوبية‬, [əssu:mħa:n əlganu:bijja(h)].
4. Discussion
The analysis of the corpus confirms that road signs are adversely affected by orthographic inconsistencies. Most sounds, including some of those which are common
to both Arabic and Latin/English phonological repertoires, present two or more
variants in the CAHRS corpus. This, however, needs to be further qualified as the
analysis reveals that there is often a statistically prevalent variant for each sound
representation and that these variants tends to coincide with those proposed in the
literature (Kharusi & Salman, 2011; Ryding, 2005). These findings show how the
proportions of individual sound transcription inconsistencies, though relatively
51
52
Rafik Jamoussi and Thomas Roche
small, are dramatically compounded at the word level, with up to 6 variants for a
single sign-posted name (see word statistics in Section 3.1). This further highlights
the obvious need for a clearly delineated romanization model to warrant consistency, which, in itself, could prove difficult to achieve (see, for instance, the history
of the romanization of Egyptian geographic names within the United States Board
on Geographic Names (USBGN, 2012).
Over and above the segmental issues brought to the fore in the transcription
chart sensu stricto, the analysis equally brings forth a number of other issues that
cannot be captured by a romanization chart, together with aspects of general implementation policies. These features require specific attention and in some instances clear language management policy. These come under three broad categories addressed in the following sections.
4.1 Transcription or transliteration
The romanization process can be achieved through two close but distinct techniques: transliteration and transcription. While transliteration is the process
whereby the source language letters of the alphabet are mapped against the Latin
ones, transcription is achieved by representing the source language phonemes (i.e.,
the actual sounds produced) in Latin script (Beesley, 1998; Halpern, 2007; Spolsky,
2009). For both transliteration and transcription, the resulting representation still
relies on a letter-to-sound correspondence which, for obvious globalization reasons, is often that of English (Aboelezz, 2009).
Though the end result of both techniques can at times be similar, a clear distinction between both is necessary due to the variances occurring between the
written and spoken target forms. Hence, for the place name “‫[ ”بركاء‬bərka] a transliteration would yield Barka’, with the apostrophe representing the final hamza “‫”ء‬.
However, because this hamza is usually not pronounced, it is not represented in a
transcription, thus Barka.
The literature is characterized by a rather loose use of the terminology whereby the terms romanization, transliteration, and transcription seem to be used interchangeably (Aboelezz, 2009; for the sake of illustration, see inter alia Kharusi &
Salman, 20115; Khraish, 2008; Al- Nabhani, 2007).
This can be accounted for in terms of the different purposes for which romanization is used. With cartography, for instance, there is an obvious need for extreme accuracy in order to secure recognizability. This is addressed by Kharusi &
Salman (2011) and equally expressed in terms of reversibility (UNGEGN, 2003),
that is, the ability for a local speaker to trace the foreign pronunciation back to
its original (in this case Arabic) name. The output in this case is more akin to
transliteration, and can be cluttered with diacritics and consequently opaque to
Road sign romanization in Oman
the non-specialist, especially when source language sounds are not available in the
Latin/English repertoire (see UNGEGN, 2003). As an illustration, applying the
UNGEGN romanization system to the entry “‫( ”الحميضة‬P048), gives al h̹myd̹ahc.
Recognizing that this level of sophistication is superfluous in public domains such
as road signs,12 a tendency is emerging that opts for a more economic notation
system (see Alghamdi, 2009; Kharusi & Salman, 2011). Though representing a
significant improvement, this simplification effort is often undermined by the fact
that it is still principally based on transliteration (Beesley, 1998; Halpern, 2007),13
with a consequential heightened importance often given to the graphological level,
the letters of the alphabet, at the expense of sounds.
This ambivalence towards the adoption of transliteration or transcription for
romanization is visible in the corpus, as in entries providing a representation of
the feminine marker (Section 3.5, Table X) despite the fact that this suffix is usually elided in pause form. Thus, “‫[ ”املصنعة‬lə msˤnʕa(h)], which appears 9 times in
the database, is represented 8 times with a final [h], as in Al Musnaah (P189).
Vowel length is another illustration of this ambivalence. As discussed above
(Section 3.2.2.2, Table VI), long vowels as in “‫”الخابورة‬, [əl xa:bu:ra(h)], or “‫”مجيس‬,
[mdʒi:s], tend to be shortened in romanization without this representing a hindrance to recognisability/reversibility (UNGEGN, 2003: 2). Still, the corpus shows
attempts to represent the length of vowels. The case of “‫ ”آل‬constitutes another
illustration of the issue. Contrary to the article “al” discussed in Section 3.3, the
segment “‫[ ”آل‬ʔa:l], as in “‫”برج آل خميس‬, [burg ʔa:l xami:s] (P240), featuring a maddah (long [a]), is a content word conveying a meaning akin to the plural “s” in
English when added to a family name to refer to the members of that family, e.g.
the Daltons, the Kennedys, so that “‫[ ”آل سعد‬ʔa:l səʕd] Al Sa’ad (P394) means “the
Sa’ads”. Out of a total of 28, one entry, Qasbiyat All Brayk (P456) reproduces the
length of the vowel in an attempt to preserve the distinction this word has with the
definite article “al”. This attempt is considered counter intuitive, a consequence of
the adoption of transliteration.
The issue is equally illustrated through the propensity to distinguish the
voiced velarized alveolar stop “‫ ”ض‬and the voiced velarized interdental fricative
“‫”ظ‬, which, though originally representing distinct phonemes in MSA, tend more
now to be regarded as a dialectal feature, i.e. allophones of the same phoneme
(Abu-Chacra, 2007; Eades, 2009).
12. As T. E. Lawrence writes, such systems are “helpful to people who know enough Arabic not
to need helping, but a wash-out for the world” (Lawrence, as cited in Gorgis, 2009, p. 19).
13. This is reinforced by the fact that romanization charts often adopt a one-to-one mapping
between the letters of the Arabic alphabet and their Latin equivalents (see for instance Kharusi
& Salman, 2011; UNGEGN, 2003), thus giving every letter a unique equivalent.
53
54
Rafik Jamoussi and Thomas Roche
Two isolated cases stick to transliteration to the extent of handling letters of
prolongation, which are used to signal long vowels (Abu-Chacra, 2007; Ryding,
2005), as full-fledged sounds: Owtab (P383), for “‫”عوتب‬, [ʕu:tab], and al Huwqayn
(P056) for “‫”الحوقني‬, [əl ħu:qi:n] (Table VI; see Isaksson, 2013 for a comprehensive
account of the features that have no significance in a simplified transcription
system).
It is interesting to note that at times the tension between transcription and
transliteration is not within the corpus but between the corpus and official recommendations. Hence, the voiced interdental fricative “‫”ذ‬, /ð/, though rare in the
corpus, is consistently transcribed through the digraph “th”. Kharusi and Salman
(2011), however, adopt the “dh”, aligning in so doing with the UNGEGN (2003)
recommendations, and reserving the otherwise more intuitive “th” in the English
system for the romanization of “‫”ث‬: /θ/ (Table II).14
More generally, the use of digraphs in romanization to represent sounds that
are specific to Arabic (Table IV) is considered superfluous as it still cannot help the
end user, typically a non-Arabic speaker, retrace the original sound represented.
Thus, these digraphs only serve the purpose of signaling unfamiliar letters, a feature of transliteration.
A more systematic use of the apostrophe may serve disambiguate cases of fake
diphthongs that can appear as a result of the eliding of the voiced pharyngeal fricative /ʕ/ and the Hamza /ʔ/ in romanization, or cases of consonant digraphs appearing with a sukun. In the first instance, the elision of the voiced pharyngeal
fricative results in the formation of a vv cluster as in Al Badiah (P627) for “‫”البديعة‬,
[ləbdi:ʕa], or A’Zaab (P339) in “‫”رويلة الزعاب‬, [rəwəjlat ʔəzʕa:b]. Under certain
conditions,15 this cluster may be mistaken by the end user for a diphthong as with
“ai” in “Masaid” in “‫”غليل املساعيد‬, [ɣali:l əl masa:ʕi:d], or “Busaid”, in “‫”ضيان البوسعيد‬
(P421). The introduction of an apostrophe in these cases, as in Ghalil Al Masa’id
(P551), can help avoid the confusion. In Falaj Al Qabail, (P391), “‫[ ”فلج القبائل‬falag
əlqaba:jəl], the vv cluster is the result of the elision of the Hamza. Again, the introduction of an apostrophe would help the end user keep the two vowels “a” and “i”
distinct and therefore have a pronunciation of the place name that is closer to the
14. At stake here is the issue of the foreign language on which romanization is based. “dh”, advocated as an equivalent sign in the Arab Division of Experts on Geographical Names (ADEGN)
system, shows an influence of French in the letter-to-sound correspondence – this affects primarily countries from the Maghreb area plus Lebanon – where the sound to be represented, /ð/,
is not available in the repertoire and a tradeoff is necessary, despite the fact that English is increasingly used as the basis of the letter-to-sound correspondence (Aboelezz, 2009; Atoui, 2012).
15. The voiced pharyngeal fricative should be medial, carry a vowel that is different from that of
the previous segment, and is not followed by the semivowel /y/ or the feminine marker.
Road sign romanization in Oman
one of the local speakers. This option is available in the corpus with the entry Al
Kaza’in (P058) as a romanizations of “‫”الخزائن‬, [ləxza:jən] as opposed to Al Khazain
(P059). The usefulness of the apostrophe is equally evident with cases of sukun
where the resulting CC cluster in romanization may be interpreted by the end user
as a digraph, such as /θ/ or /ð/ for “th”, as is the case for “Batha” in Batha Al Ashkhar,
“‫”بطحاء األشخر‬, [bat̪ ˤħa:] (P252). The use of an apostrophe, “Bat’ha” would again prevent the erroneous interpretation.
Working towards the elimination of romanization inconsistencies implies intervention at the level of road sign production and taking into account the level
of linguistic expertise of the personnel involved. Steps toward the simplification of
the system of the kind proposed in Kharusi and Salman (2011) can only be considered to address this question from the perspective of the end user (those negotiating the roads). It is believed here that inconsistencies of the type addressed in this
section can be avoided through the adoption of transcription instead of transliteration. This option would represent a further step towards simplification, though
this time it is a simplification that targets primarily the production personnel as
it eases the system of non-significant distinctions through the exclusive focus on
phonemes (see Wells, 2006)
As a further step, the mapping of Arabic sounds can be reorganised in a fashion that serves the transcription approach better. The categories would thus bring
out pertinent differences between the Arabic and the Latin systems while eliminating insignificant ones from the discussion. This is in line with the proposal made at
the 2007 UNGEGN conference (originally quoted in Al-Nabhani, 2007: 73).
4.2 Language management: Which Arabic?
From its classical origins Arabic has evolved into a range of spoken dialects (see the
discussion in Eades, (2009) about the main Bedouin/sedentary dialectal dichotomy in Arabic) and Modern Standard Arabic (MSA), a high prestige register variety
made use of in formal settings, and more specifically in written communication
(Eades, 2009). The context of Oman adds to this with a number of endogenous
languages/dialects that do not originate from Arabic including the South Semitic,
Mehri and the Indo-Iranian, Kumzari languages (Holes, 1984, 2004). With this
socio-linguistic background, the need emerges for consensus on the question of
the Arabic language variety on which any road signage romanization system is to
be based (Aboelezz, 2009).
What emerges from the analysis of the CAHRS corpus is a clear tendency
to opt for MSA. This is predominately evident through structural features, such
as: the quasi absence of dialectal consonant sound shifts (Section 3.6); the retention of diphthongs which otherwise tend to shift to long vowels in the spoken
55
56
Rafik Jamoussi and Thomas Roche
variety (as in A 'Swaihrah, P395, P556, P559, for “‫”الصويحرة‬, [əsˤwi:ħra(h)] [see also
Section 3.2.3 and Table VII]); the preference for the vowelized variant over the
non-vowelised one (Section 3.2.4); and, the clear delineation of the non-assimilated definite article, albeit with a number of variants (Section 3.3), when the spoken
variety would reduce the feature to a [lə] merged with the next word, as in Al
Hadirah (P595) for “‫”الحظرية‬, [ləħði:ra(h)].
Let us briefly add here that the adoption of MSA with these features is by
no means consistent, though the practice occurs in high-frequency. For instance,
parallel to the two entries al Ghurayfah (P155,) for “‫[ ”الغريفة‬lə ɣrəjfa], and Al
Mintayfah (P204) for “‫[ ”املنطيفة‬əl mint̪ ˤi:fa(h)], both of which are clearly based on
an MSA reading through the use of the diphthong and the vowelization, we have
romanizations which are based on spoken forms, respectively Al–Gharifah (P154)
and Al Manteafah (P203).
This language variety fluctuation sometimes occurs even within a single entry.
Hence, “‫[ ”سور بني خزمية‬su:r bani: xzi:ma(h)] is romanized as Sur Bani Khazaymah
(P363, P364) where the diphthong “ay” in “Khazaymah” clearly refers to the standard variety, while the first vowel “a” refers to the spoken one as it neutralizes the
/u/ sound into a schwa. Likewise, “‫[ ”سيح املحامد‬si:ħ ləmħa:məd], is rendered Sayh
al Mahmed (P377), where the diphthong “ay” in “Sayh” contrasts with the “e” in
Mahmed. For the sake of comparison, Sayh Al Mahamid (P378) provides a more
consistent rendering.
Explicit official positions of the kind Kharusi & Salman (2011) supply16 provide directions to an otherwise seemingly random practice, and help eliminate
some of the indecision that characterises the issue. However, it is thought here that
the official stance (as seen in Kharusi & Salman, 2011) in favour of the adoption
of MSA as a basis for romanization still leaves room for inconsistencies as it does
not specify the full range of the linguistic features it entails. This is evident in the
corpus in features that seem to be out-of- sync by opting for the regional spoken
variety or being at least a step away from MSA. These features cluster around the
“full form” and “pause form” pronunciation variants, as is shown below.
Case vowels: full form vowel pronunciation includes word-final case vowels,
or inflectional vowels that usually indicate the nominative, accusative and genitive
cases (Abu-Chacra, 2007, Isaksson, 2013; Ryding, 2005). This type of pronunciation is rare and “pause form” pronunciation (waqf), implicating the dropping of
the final short vowel, is used instead. When adopting formal speech, the use of the
16. Kharusi & Salman (2011) show an acute awareness of this aspect and make room for different treatments depending on whether a word is of Arabic or non-Arabic origin. However, when
an Arabic origin is established for the word in question, they advocate the use of MSA, showing
an understandable urge for standardization.
Road sign romanization in Oman
pause form is usually restricted to the end of the sentence or, when the sentence
is long, to the end of clauses or phrases (Ryding, 2005). At a less formal level,
the pause form is generalised over most if not all words, thus avoiding to specify
inflections, which require a sound knowledge of the language rules. The database
shows that multiple-word entries make use of the pause form characteristic of
the less formal variety (99.57% of 233 applicable records). Thus, “‫ ”سور‬in “‫”سور قطيط‬
(P365) is pronounce [su:r] instead of [su:ru] which would signal the genitive case.
The romanization follows the same pattern, thus Sur Qteyt (P365).
The feminine case: except in liaison cases, the corpus shows a tendency to
make use of the pause form for the feminine case, achieved through the use of the
weaker version “h” (Table X). Alternatively, the feminine case is simply dropped.
This treatment is thought to be one step removed from the standard variety. A
transcription fully based on the standard variety would have featured a “t” together with a full inflection vowelization.
Nunation: “tanween” or nunation for indefinite nouns and adjectives is usually signalled through “doubling the final vowel sign and pronouncing it with a
final /n/” (Abu-Chacra, 2007). For an entry such as “‫”صالن‬, the usual pronunciation
is devoid of nunation; [sˤəlla:n] instead of [sˤəlla:nun]. This is reflected in the romanization; Sallan (P397). The abandoning of this MSA feature in the database is
a salient feature of the adoption of simplified spoken varieties.
The word-final Hamza, as in " ‫"جامء‬, is normally elided outside MSA, hence
[gəmma]. In this respect, the database shows a clear preference for the spoken
variety.
Over and above the issue of dialectal variation featuring alternative pronunciations of some sounds (Section 3.6), the corpus analysed here clearly shows that
it would be erroneous to deal with the standard/spoken variation as a clear-cut binary. Rather, the variation encompasses a number of discrete features that cluster
along a continuum between the two extremes, which represent the relatively elusive concepts of standard and spoken varieties. The lack of consistency in adopting
either the standard or the spoken variety in the features above clearly illustrates
the need to further specify the adoption of MSA pronunciation as a basis for romanization, if such is the official position.
4.3 Institutionalized names
Some entries do not seem to be affected by romanization inconsistencies as they
have a single variant each throughout their occurrences in the corpus (Table I). The
argument that these entries involve none of the multi-variant features that cause
inconsistencies is quickly dismissed, with counterexamples such as Khatmat, [xət̪
ˤmət]; Muscat, [məsqət̺ ˤ]; Sur, [su:r]; Al Buraimi, [ləbri:mi], or Salalah, [sala:la(h)],
57
58
Rafik Jamoussi and Thomas Roche
all including consistent renderings of otherwise multi-variant features, respectively the consonants [x] and [q], the long vowel [u:], the diphthong [ayj], and the
representation of the feminine marker (Tables IV, VI, VII, X).
A more plausible explanation is that we are dealing here with names that
have acquired a collective renown warranting their uniform transcription.
Institutionalised romanizations of this sort reveal the challenges and limitations of
a strict adoption of the prescriptive approach. A revision of road signs that strictly
implements a specific romanization chart would signify a change in the way these
particular terms are written, such as to adopt “Musqat” as the Romanization of
the name of the capital city instead of the established, though deviant variant,
“Muscat”. Practical aspects and even cultural significance cannot be underestimated here. This issue is exacerbated in the case of trans-boundary names. For
instance, Ar Rub Al Khali, a place name Oman shares with Saudi Arabia, makes
use of the assimilated variant of the definite article “Al” (Al-Nabhani, 2007: 68). A
decision to amend this feature according to the recommendation of the transcription system cannot be made unilaterally. The case of institutionalized names thus
suggests that the system should be flexible enough to accommodate exceptions;
special cases where usage overrides the strict application of the romanization system (see Kharusi & Salman, 2011).
5. Conclusion and recommendations
The present work brings a confirmation of the fact that road sign romanization
in the CAHRS corpus is plagued by inconsistencies. However, what the analysis
equally shows is that these inconsistencies are not the sole outcome of random
practice, but often result from competing and contradictory romanization principles. Remedies in this case cannot only consist in providing a system to harmonize the romanization practice across the country, as the system itself should be
based on carefully thought out decisions with broader linguistic policy overtones.
One such decision relates to the clear identification of the romanization
method to be adopted, as fluctuations between transcription and transliteration
account for a large proportion of inconsistencies. As seen in the above, the present
authors favour the adoption of transcription as it seems to provide more userfriendly results from an end-user perspective and is equally less demanding on the
personnel in charge of the implementation of romanization. The importance of
such language policy decisions, as discussed here, has implications not just for the
Omani context, but equally for road sign projects and other romanization endeavours throughout the Arab world and beyond more broadly.
Road sign romanization in Oman
Another issue of major influence is the language variety to be adopted for
romanization. This issue is more complex than choosing between the MSA or spoken variety binary. The investigation provides ample evidence that, rather than a
binary, the situation could be more helpfully portrayed as a continuum with MSA
and spoken varieties as the two extremes, between which a whole range of considerations come into force.
From an implementation perspective, the appearance of mistakes, renderings
which deviate for no apparent reason from the Arabic pronunciation to the extent of compromising recognisability, though rare, points toward the necessity of
introducing a quality control system as a safeguard against gross typos and misinterpretations. Equally important is perhaps to opt for centralising any road sign
production process as a safeguard against the vagaries of decentralisation.
Funding
This work was supported by The Research Council of Oman (TRC) under Grant
ORG/HER/12/003.
References
Aboelezz, M. (2009). A Latinized Arabic for all? Issues of Representation, Purpose and Audience.
Paper presented at The International Symposium on Arabic Transliteration Standard:
Challenges and Solutions, Abu Dhabi.
Abu-Chacra, F. (2007). Arabic: An essential grammar. London: Routledge.
Abu-Rabia, S. (1997). Reading in Arabic orthography: The effect of vowels and context on reading accuracy of poor and skilled native Arabic readers in reading paragraphs, sentences,
and isolated Words. Journal of Psycholinguistic Research, 26(4), 465–482.​
doi: 10.1023/A:1025034220924
Al-Nabhani, Y. H. N. (2007). The Role and Standardisation of Geographical Names on Maps:
Oman as a Case Study. (Master of Science), University of Glasgow, Faculty of Physical
Sciences. Retrieved from http://theses.gla.ac.uk/460/1/2007al-nabhaniMSc.pdf
Alghamdi, M. (2009). Romanizing Arabic Proper Names: Saudi Arabia Experience. Paper presented at The International Symposium on Arabic Transliteration Standard: Challenges and
Solutions, Abu Dhabi.
Atoui, B. (2012, July-August). The Issue of the Romanisation System for the Arab Countries:
Between Legitimacy and Practices. Which Solutions? Paper presented at the Tenth United
Nations Conference on the Standardization of Geographical Names, New York. Retrieved
from
http://unstats.un.org/unsd/geoinfo/UNGEGN/docs/10th-uncsgn-docs/econf/E_
CONF.101_96_The%20issue%20of%20Romanization.pdf
59
60
Rafik Jamoussi and Thomas Roche
Beesley, K. R. (1996). Arabic Finite-State Morphological Analysis and Generation. Paper presented at The 16th Conference on Computational Linguistics. http://acl.ldc.upenn.edu/C/C96/
C96-1017.pdf ​doi: 10.3115/992628.992647
Beesley, K. R. (1998). Romanization, transcription and transliteration. Retrieved from https://
open.xerox.com/Services/arabic-morphology/Pages/romanization.
China Education and Research Network. (2005). Welcome the 2008 Games with Correct
Bilingual Signs. China Education and Research Network. Retrieved from http://www.edu.
cn/Newswin_1547/20060323/t20060323_127180.shtml.
Eades, D. (2009). The Arabic Dialect of a Šawāwī Community of Northern Oman. In E. Al-Wer
& R. d. Jong (Eds.), Arabic Dialectology. In Honour of Clive Holes on the Occasion of his
Sixtieth Birthday (Vol. 53, pp. 77–98). Boston: Leiden.​
doi: 10.1163/ej.9789004172128.i-298.36
Gorgis, D. T. (2009). Transliterating Arabic: The Nuisance of Conversion Between Romanisation
and Transcription Schemes. Paper presented at The International Symposium on Arabic
Transliteration Standards: Challenges and Solutions, Abu Dhabi.
Halpern, J. (2007, July). The Challenges and Pitfalls of Arabic Romanization and Arabization.
Paper presented at The Second Workshop on Computational Approaches to Arabic ScriptBased Languages, Stanford University. Retrieved from http://www.cjk.org/cjk/arabic/arannana.pdf
Holes, C. (1984). Colloquial Arabic of the Gulf and Saudi Arabia. London: Routledge & Kegan
Paul.
Holes, C. (2001). Dialect, culture, and society in eastern Arabia (Vol. 3). Leiden: Brill.
Holes, C. (2004). Modern Arabic: Structures, functions, and varieties (Rev. ed.). Washington, D.C:
Georgetown University Press.
Holes, C. (2009). Omani Arabic. In L. Edzard & R. de Jong (Eds.), Encyclopedia of Arabic
Language and Linguistics, (pp. 478–491). Leiden: Brill.
Isaksson, B. (2013) Transcription of written Arabic. Uppsala Universitet. Retrieved from http://
www.lingfil.uu.se/digitalAssets/94/94977_transcription-of-arabicEN.pdf.
Jamson, S., Tate, L. F. N. & Jamson, A. H. (2005). Evaluating the effects of bilingual traffic signs
on driver performance and safety. Ergonomics, 48(15), 1734–1748.​
doi: 10.1080/00140130500142191
Kharusi, N. S., & Salman, S. (2011). The English transliteration of place names in Oman. Journal
of Academic and Applied Studies, 1(3), 1–27.
Khraish, M. (2008). Fourth Arab conference on geographical names. Lebanon report concerning
the activities related to geographical names. Beirut.
Ryding, K. C. (2005). A reference grammar of modern standard Arabic. Cambridge: Cambridge
University Press. ​doi: 10.1017/CBO9780511486975
Sloboda, M., Szabó-Gilinger, E., Vigers, D., & Šimičić, L. (2010). Carrying out a language policy
change: Advocacy coalitions and the management of linguistic landscape. Current Issues in
Language Planning, 11(2), 95–113. ​doi: 10.1080/14664208.2010.505067
Spolsky, B. (2009). Language management. Cambridge: Cambridge University Press.​
doi: 10.1017/CBO9780511626470
UNGEGN. (1967). Resolution 1/9. Retrieved from http://www.eki.ee/wgrs/res/res_1_9.htm.
UNGEGN. (2003). Report on the Current Status of United Nations Romanization Systems for
Geographical Names. Arabic. Retrieved from http://www.eki.ee/wgrs/rom1_ar.pdf
UNGEGN. (2006). Manual for the national standardization of geographical names. New York:
United Nations.
Road sign romanization in Oman
USBGN. (2012). Geographic Names Standardisation Policy for Egypt. Retrieved from http://
earth-info.nga.mil/gns/html/Policies/Egypt_Country_Policy_webversion_Jan2012.pdf.
Watson, J. C. E. (2002). The phonology and morphology of Arabic. Oxford: Oxford University
Press.
Wells, J. C. (2006). Phonetic Transcription and Analysis. In K. Brown (Ed.), Encyclopedia of
Language and Linguistics (pp. 386–396). Amsterdam: Elsevier.​
doi: 10.1016/B0-08-044854-2/00014-6
Zagórski, B. R. (2010). Endonyms versus exonyms: A case study in standardization: With a list
of names of Arab countries and their major cities, In P. Jordan (Ed.), Trends in Exonym
Use. Proceedings of the 10th UNGEGN Working Group on Exonyms Meeting Tainach, 28–30
April, 2010 (pp. 95–129).
Appendix 1
Table I. Top words with a single variant (out of 72 records)
Word
Frequency
Word
Frequency
Sohar
48
Khatmat
12
Muscat
42
Malahah
12
Wadi
36
Salalah
9
Sur
24
Barka
9
Bani
22
Wudam
9
Dubai
12
Abu
9
ADEGN
Kharusi &
Salman
No of records
in the database
CAHRS
No of variants
Phonetic description
Arabic transcription
Sound
Table II. Common consonant sounds variation*
b
b
263
b
1
‫ب‬
/b/
voiced bilabial stop
‫ت‬
/t/
voiceless alveolar stop
t
‫ث‬
/θ/
voiceless interdental fricative
th
‫ج‬
/j/
voiced alveopalatal affricate
j
j
t
76
t(75), th (1) (P441)
2
th
6
th
1
75
j (66), g (7), q (2)
3
‫د‬
/d/
voiced alveolar stop
d
d
173
d
1
‫ذ‬
/ð/
voiced interdental fricative
dh
dh
2
th
1
‫ر‬
/r/
voiced alveolar flap or trill
r
r
360
r
1
‫ز‬
/z /
voiced alveolar fricative
z
z
41
z
1
‫س‬
/s/
voiceless alveolar fricative
s
s
220
s
1
61
Rafik Jamoussi and Thomas Roche
‫ش‬
/sh/
voiceless palatal fricative
sh
sh
77
sh
1
‫ف‬
/f /
voiceless labiodental fricative
f
f
99
f
1
k
k
26
k (25), q (1)
2
‫ك‬
/k/
voiceless velar stop
‫ل‬
/l/
voiced lateral
l
l
197
l
1
‫م‬
/m/
voiced bilabial continuant
m
m
285
m
1
‫ن‬
/n/
voiced nasal continuant
n
n
165
n
1
‫ه‬
/h/
voiceless glottal fricative
h
h
51
h
1
‫و‬
/w/
bilabial semivowel
w, ū
w
145
w
1
‫ي‬
/y/
palatal semivowel
y, ī
y
79
y (57), i (11), nil (9),
3
dropped
67
hamza /ʔ/
glottal stop
* Phonetic transcription is based on Ryding (2005, pp. 13–15). For all the tables displaying sound mappings the ADEGN and the Kharusi & Salman systems are provided as reference points.
CAHRS
No of variants
Position
Dropped
No of records
in the database
Glottal stop /ʔ/
Kharusi &
Salman
‫ء‬
ADEGN
Phonetic
description
Table III. Hamza sound variation (67)*
Arabic transcription
62
35
Word final
– Dropped (34)
– h (1) (P297) Humairah,
“‫[ ”حمرياء‬ħmi:ra(h)]
2
32
Other
– Dropped (31)
– Apostrophe (1) (P058) Al
Kaza’in, “‫[ ”الخزائن‬ləxza:jən]
2
* Hamza cases occurring with the definite article are treated in Table VIII.
Road sign romanization in Oman
ADEGN
Kharusi & Salman
voiced velarized alveolar
stop /d/
ḏ
d
36
– d (30) (83.33%)
– dh (6) (16.66%)
2
‫ظ‬
voiced velarized interdental
fricative /ðˤ/
dh
dh
1
– d
1
‫ص‬
voiceless velarized alveolar
fricative /sˤ/
s
s
135
– s
1
‫خ‬
voiceless velar fricative /x/
kh
kh
106
– kh (100) (94.33%)
– k (6) (5.66%)
2
‫ح‬
voiceless pharyngeal fricative /ħ/
h
h
248
– h
1
‫ع‬
voiced pharyngeal fricative
/ʕ/
‘
Dropped
except in
few cases
(y”, “w”)
153
– Dropped (119) (77.77%)
–Vowel is kept (100 cases) (Al
Musana/‫)املصنعة‬
–Vowel is dropped with the
consonant (3 cases) (P291) (Al
Qanain/‫)القعانني‬
–Consonant is without a vowel
(16 cases) (P001) (Al Zafaran/
‫الزعفران‬
–Apostrophe (34) (22.22%), (Al
Masa’id/ ‫)املساعيد‬
2
‫غ‬
voiced velar fricative /ɣ/
gh
gh
47
– gh (40) (85.1%)
– g (7) (14.89%)
3
‫ق‬
voiceless uvular stop /q/
q
q
148
–q (104), (al Rustaq / ‫( )الرستاق‬P076)
(70.27%)
–c (39), (Muscat/ 26.35%( )‫)مسقط‬
–k (1), (Souk/ ‫( )سوق‬P095/353)
(0.45%)
–g (1), (Al Bogryn/ ‫( )آل بوقرين‬P347)
(0.45%)
4
‫ط‬
voiceless velarized alveolar
stop /t̪ ˤ/
t
t
103
– t
No of variants
Phonetic description
& representation
‫ض‬
No of records in the
database
CAHRS
Arabic transcription
Table IV. Specific consonant sounds variation (sounds not belonging to the English
repertoire)
63
֜
֜
֜
Vowel
‫ الكرسة‬/i/
‫ الضمة‬/u/
‫ الفتحة‬/a/
i
u
a
i
u
a
58
32
488
ADEGN Kharusi No of
&
records
Salman
Table V. Short vowel variation
45
2
8
3
i
u
a
e
3
o
2
nil
29
1
e
u
485
a
5.17%
13.79%
3.44%
77.58%
9.37%
90.62%
0.4%
0.2%
99.38%
[ʔabu d̪ d̪ ru:s]
[ləʕði:bə(h)]
[əssəlla:ha]
[ʔabu nnxi:l]
Phonetic
Transcription
‫أبو الرضوس‬
‫العذيبة‬
‫السالحة‬
‫أبو النخيل‬
Arabic transcription
[əl xəʃda(h)]
[əl bɪllə(h)]
Hellat As’sawaba (P286) [ħəllət əssawa:bəʕ]
Rassat Al Malah (P335) [rəssat əl məlħ]
Al Khushdah (P060)
al Billah (P031)
‫حلة السوابع‬
‫رسة امللح‬
‫الخشدة‬
‫البلة‬
Wadi Bani Omar (P093) [wa:di bani: ʕumər] ‫وادي بني عمر‬
Abu A’Dhurus (P003)
Aluthaiybh (P128)
A’SELLAHAH (P091)
Abu A’Nakheel (P004)
CAHRS Variant hits Frequency Example
Variant
64
Rafik Jamoussi and Thomas Roche
Kharusi &
Salman
Reduced (p 16)
Kharusi &
Salman
dismissed
ADEGN
ū
َ‫ – و‬Total frequency 125
ā
ADEGN
َ‫ – ا‬Total frequency 418
1
Other
65
19
22
13
1
1
2
2
u
o
ou
oo
ow
uw
aw
Other
Variant
hits
1
a‘a
CAHRS
Variant
Variant
hits
416
CAHRS
Variant
a
Table VI. Long vowel variation
1.6%
1.6%
0.8%
0.8%
10.4%
17.6%
15.2%
52%
Frequency
0.23%
99.52%
Frequency
AL Bardah (P021)
Sur Bani Hamma’ad (P361)
Al Khaborah (P165)
Abu A’Dhurus (P003)
mistake
Misinterpreted:
3.2%
Mehat Bani Kueem (P506)
Khawr Al Malh (P316)
al Huwqayn (P056)
Owtab street (P383)
Vowel length
A’Souq (P087)
reproduced: 28% Hillat Al Jood (P279)
Reduced: 67.2%
Freq. by category Example
mistake
Reduced
Freq. by category Example
[mi:ħət bani:
kəjju:m]
[xu:r əl malh]
[əl ħu:qi:n]
[ʃa:ri ʕ ʕu:tab]
[ħəllət əl dʒu:d]
[əs su:q]
[l xa:bu:ra(h)]
[ʔabu d̪ d̪ ru:s]
Phon. transcription
[su:r bani:
ħamma:d]
‫ميحة بني كيوم‬
‫خور امللح‬
‫الحوقني‬
‫شارع عوتب‬
‫حلة الجود‬
‫السوق‬
‫الخابورة‬
‫أبو الرضوس‬
‫سور بني حامد‬
Phon. transcription
‫الباردة‬
[əlba:rda]
Road sign romanization in Oman
65
Kharusi &
Salman
dismissed
ADEGN
ī
َ‫ – ي‬Total frequency 192
3.64%
7
24
2
1
6
y
ee
ie
ea
ai/ay
2.08%
0.52%
1.04%
12.49%
2.6%
5
e
76.56%
Frequency
147
Variant
hits
i
CAHRS
Variant
A’Shiresah (P281)
Ghalil Al Masa’id (P551)
Diphthongised
3.12%
Hillat al Qanain (P291)
Al Manteafah (P057)
Al Ghalieleh (P160)
Vowel length re- Timyt (P430)
produced: 17.7% Abu A’Nakheel (P004)
Reduced:79.16%
Freq by category Example
‫املنطيفة‬
[əl mint̪ ˤi:fa(h)]
[ħəllət ləqʕa:ni:n] ‫حلة القعانني‬
‫الغليلة‬
[ləɣli:la(h)]
‫أبو النخيل‬
[t̪ ˤəmi:t]
[ʔabu nnxi:l]
‫الرشيسة‬
‫طميت‬
‫غليل املساعيد‬
[ʔə ʃri:sa(h)]
[ɣali:l lə
msa:ʕi:d]
Phon. transcription
66
Rafik Jamoussi and Thomas Roche
49 (27.52%)
11 (6.17%)
3 (1.68%)
2 (1.12%)
1 (0.56%)
1 (0.56%)
3 (1.68%)
6 (3.37%)
17 (9.55%)
5 (2.8%)
4 (2.24%)
2 (1.12%)
1 (0.56%)
ai
ei
ey
aiy
ayy
uy
ea*
ee
i
e
y
ay
a
mistake
misinterpreted
reduced (32):
17.97%
reproduced (142):
79.77%
Frequency by
category
* This instance represents a spoken interpretation of the diphthong.
72 (40.44%)
ai
Not addressed
Frequency
ay
Kharusi
CAHRS
& Salman Variant
ADEGN
Table VII. Dipthongs (178 records)
Al Felaj (P169)
Khadra Al Bu Rishayd (P304)
A’Rumylah (P079)
Al Kuwerat (P062)
Ruwilat A’Zaab (P339)
A’Tareef (P108)
Al Manteafah (P203)
Ruwuyhah (P336)
As Suwayyq (P616)
Aluthaiybh (P128)
Sur Qteyt (P365)
Murtafa’at Al Breik (P028)
Al Godairah (P153)
Al Musaylah (P546)
Example
‫الفليج‬
‫خرضاء آل بورشيد‬
[xəd̪ ˤra a:l burʃi:d]
[ləfli:ʒ]
‫الرميلة‬
‫الخويرات‬
[ərmi:la(h)]
[ləxwi:ra:t]
‫رويلة الزعاب‬
‫املنطيفة‬
‫الطريف‬
[əl mint̪ ˤi:fa(h)]
[ʔət̪ ˤri:f]
[rəwi:lat ʔəzəʕʕa:b]
‫رويحة‬
‫السويق‬
[ərwi:ħə(h)]
[sswi:q]
‫العذيبة‬
‫سور قطيط‬
[su:r qt̪ ˤi:t̪ ˤ]
[ləʕði:bə(h)]
‫مرتفعات الربيك‬
‫الغديرة‬
‫املسيلة‬
[murtafaʕa:t lə bri:k]
[ləɣdi:ra(h)]
[ləmsi:la(h)]
Phon. transcription
Road sign romanization in Oman
67
al
With assimilation
(196 occurrences)
‫شمس ّية‬
Without assimilation (344 occurrences) ‫قمريّة‬
Kharusi & Type
Salman
268 (77.9%)
49 (14.24%)
19 (5.52%)
4 (1.16%)
1 (0.29%)
3 (0.87%)
77 (39.28%)
46 (23.46%)
31 (15.81%)
2 (1.02%)
39 (19.89%)
1 (0.51%)
al
Al-
Al (linked to the next word)
A’ (linked to the next word)
Not represented
A’ (with apostrophe separate
from the next word)
A’ (apostrophe merged with
next word)
Al (both with upper and lower
case)
Al- (with hyphen)
Assimilation (dropping of the
“l” + doubling of the initial letter of the following word)
Not represented
Frequency
Al
CAHRS Variants
Table VIII. Definite article variations
[ʔəssaħil]
Al-Shizaw (P557)
[ʔəʃʃi:za:w]
[ʔəssəqsu:q]
Subarah R/A (P561) [duwa:r ʔəsˤba:ra(h)]
as Saqsuq (P090)
‫دوار الصبارة‬
‫السقسوق‬
‫الشيزاو‬
‫الروضة الصبارة‬
‫الساحل‬
‫الدموس‬
‫مزرعة الباطنة‬
[əl ba:t̪ ˤina(h)]
[əddəmmu:s]
‫الحلة الجنوبية‬
‫العذيبة‬
‫الغوايب‬
‫املغرس‬
‫الفاغرة‬
Example
[lħəlla lganu:bijja(h)]
[ləʕðeibə(h)]
[əl ɣawa:bi]
[əl məɣsər]
[əl fa:ɣra(h)]
Phon. transcription
Al Rawdhah (P082) / [ʔəru:d̪ ˤa(h)] /
al Subarah (P288)
[ʔəssˤba:ra(h)]
A’Sahil (P610)
A’ Dammos (P066)
Batinah live stock
(P488)
Al – Hillah
A’Janubiya (P611)
Aluthaiybh (P128)
Al-Ghawabi (P609)
al Maghsar (P008)
Al Fahgrah (P015)
Example
68
Rafik Jamoussi and Thomas Roche
7
nil
nn
rr
ss
tt
‫( ن‬2 records)
‫( ر‬2 records)
‫( س‬4 records)
‫( ت‬5 records)
‫( م‬5 records)
‫( ل‬37 records)
Doubling
ll
the letter
except with l
digraph
mm
5
4
2
2
Hatta (P263)
Rassat Al Misbakh (P332)
Wadi A’ Sarrami (P529)
Hajr Al Sinnanat (P265)
Jamma (P260)
Hilat al Sheikh (P112)
4
5
al Billah (P031)
Al Muladdah (P200)
Khabbat an Nawafil (P303)
Al Hindiah (P220)
Hayy As Salam (P298)
Al Hamliya (P216)
Example
33
5
7
3
yy
dd
29
Frequency
y
CAHRS
Variant
bb
Kharusi &
Salman
‫( د‬5 records)
Doubling
the letter
ADEGN
‫( ب‬7 records)
‫( ي‬39 records)
Geminated
sound
Table IX. Gemination variations (106 records)
[ħattæ]
[rəssat əl mʊsˤba:x]
[wa:di sˤərrami]
[ħagər əsna:na:t]
[gəmma]
[ħəllət əʃ ʃi:x]
[əl bɪllə(h)]
[ʔəlmlæddæ(h)]
[xəbbət ən nawa:fil]
[əlhindija(h)]
[ħaj ɪssala:m]
[al ha:mli:ja(h)]
Phon. transcription
‫حتا‬
‫رسة املصباخ‬
‫وادي الرصمي‬
‫حجر السنانات‬
‫جام‬
‫حلة الشيخ‬
‫البلة‬
‫امللدة‬
‫خبة النوافل‬
‫الهندية‬
‫حي السالم‬
‫الهاملية‬
Road sign romanization in Oman
69
70
Rafik Jamoussi and Thomas Roche
Table X. Feminine case variation
Kharusi
& Salman
CAHRS
Variants
Frequency
SINGULAR Not repre- Not repre(211 resented
sented
cords)
h
SINGULAR t
RUN ON
(30 records)
Phon. transcription
28 (13.27%) Al
[ləxwi:ri:a(h)]
Khuwayriya
(P064)
183 (86.72%) Al
Ashikhrah
(P018)
[la ʃi:xra(h)]
‫الخويرية‬
‫األشيخرة‬
t (with words
bearing
no definite
article)
21        Qasbiyat Al [qəsˤbiət
Hawasneh əlħawa:snə(h)]
(P007)
‫قصبية‬
‫الحواسنة‬
h (with
words bearing a definite
article)
7
[l ʕəqdə l
ɣarbi:jə(h)]
‫العقدة‬
‫الغربية‬
[lħa:ra
li:di:da(h)]
‫الحارة‬
‫الجديدة‬
[ləħsana:t]
‫الحسنات‬
Not represented
PLURAL
t
(30 records)
Example
Al Uqdah
Al
Garbiyah
(P132)
2 (6.66%) Al Hara
Al Jadidah
(P042)
t
30 (100%)
Al Hasnat
(P044)
Table XI. j/g sound shift/variation (75 records)
ADEGN
Kharusi CAHRS
& Salman Variants
j
j
j
Frequency
Example
Phon. transcription
66 (87.99%) Burj al Khamis (P242) [burg ʔa:l xami:s] ‫برج آل خميس‬
g
7 (9.33%) Hilat Al Burg (P276)
[ħəllət əl burg]
‫حلة الربج‬
q
2 (2.66%) Fareq A’Zaab (P446)
[fri:g əzəʕʕa:b]
‫فريج الزعاب‬
Corresponding author’s address
Thomas Roche
SCU College, Southern Cross University
Military Rd
Lismore, 2480, NSW
Australia
thomas.roche@scu.edu.au