Any
romanization system has to make a number of decisions which are dependent on its intended field of application.
Vowels One basic problem is that written Arabic is normally
unvocalized; i.e., many of the
vowels are not written out, and must be supplied by a reader familiar with the language. Hence unvocalized
Arabic writing does not give a reader unfamiliar with the language sufficient information for accurate pronunciation. As a result, a pure
transliteration, e.g., rendering as '
, is meaningless to an untrained reader. For this reason, transcriptions are generally used that add vowels, e.g. '. However, unvocalized systems match exactly to written Arabic, unlike vocalized systems such as Arabic chat, which some claim detracts from one's ability to spell.
Transliteration vs. transcription Most uses of romanization call for
transcription rather than
transliteration: Instead of transliterating each written letter, they try to reproduce the sound of the words according to the orthography rules of the target language:
Qaṭar. This applies equally to scientific and popular applications. A pure transliteration would need to omit vowels (e.g.
qṭr), making the result difficult to interpret except for a subset of trained readers fluent in Arabic. Even if vowels are added, a transliteration system would still need to distinguish between multiple ways of spelling the same sound in the Arabic script, e.g. '
vs. ' for the sound '
, and the six different ways () of writing the glottal stop (hamza, usually transcribed '). This sort of detail is needlessly confusing, except in a very few situations (e.g., typesetting text in the Arabic script). Most issues related to the romanization of Arabic are about transliterating vs. transcribing; others, about what should be romanized: • Some transliterations ignore
assimilation of the
definite article al- before the "
sun letters", and may be easily misread by non-Arabic speakers. For instance, "the light"
an-nūr would be more literally transliterated along the lines of
alnūr. In the transcription
an-nūr, a hyphen is added and the unpronounced removed for the convenience of the uninformed non-Arabic speaker, who would otherwise pronounce an , perhaps not understanding that in
nūr is
geminated. Alternatively, if the
shaddah is not transliterated (since it is strictly not a letter), a strictly literal transliteration would be
alnūr, which presents similar problems for the uninformed non-Arabic speaker. • A transliteration should render the "closed tāʼ" (
tāʼ marbūṭah, ) faithfully. Many transcriptions render the sound as
a or
ah and
t when it denotes . •
ISO 233 has a unique symbol, ẗ. • "Restricted alif" (''
, ) should be transliterated with an acute accent, á
, differentiating it from regular alif , but it is transcribed in many schemes like alif, ā'', because it stands for . •
Nunation: what is true elsewhere is also true for nunation: transliteration renders what is seen, transcription what is heard, when in the Arabic script, it is written with diacritics, not by letters, or omitted. A transcription may reflect the language as spoken, typically rendering names, for example, by the people of Baghdad (
Baghdad Arabic), or the official standard (
Literary Arabic) as spoken by a
preacher in the
Mosque or a TV newsreader. A transcription is free to add phonological (such as vowels) or morphological (such as word boundaries) information. Transcriptions will also vary depending on the writing conventions of the target language; compare English
Omar Khayyam with German
Omar Chajjam, both for , (unvocalized '
, vocalized '). A transliteration is ideally fully reversible: a machine should be able to transliterate it back into Arabic. A transliteration can be considered as flawed for any one of the following reasons: • A "loose" transliteration is ambiguous, rendering several Arabic phonemes with an identical transliteration, or such that digraphs for a single phoneme (such as
dh gh kh sh th rather than
ḏ ġ ḫ š ṯ) may be confused with two adjacent consonants—but this problem is resolved in the
ALA-LC romanization system, where the
prime symbol ʹ is used to separate two consonants when they do not form a digraph; for example: '''' ('she honored her'), in which the
t and
h are two distinct consonantal sounds, or where the
middle dot is used in the same way in the BGN/PCGN romanization. • Symbols representing phonemes may be considered too similar (e.g., ʻ and ' or ʿ and ʾ for
ʻayn and
hamzah); • ASCII transliterations using capital letters to disambiguate phonemes are easy to type, but may be considered unaesthetic. A fully accurate transcription may not be necessary for native Arabic speakers, as they would be able to pronounce names and sentences correctly anyway, but it can be very useful for those not fully familiar with spoken Arabic and who are familiar with the Roman alphabet. An accurate transliteration serves as a valuable stepping stone for learning, pronouncing correctly, and distinguishing phonemes. It is a useful tool for anyone who is familiar with the sounds of Arabic but not fully conversant in the language. One criticism is that a fully accurate system would require special learning that most do not have to actually pronounce names correctly, and that with a lack of a universal romanization system they will not be pronounced correctly by non-native speakers anyway. The precision will be lost if special characters are not replicated and if a reader is not familiar with Arabic pronunciation. == Examples ==