Arabic phonology

Vowel chart representing the pronunciation of long vowels by a Palestinian speaker educated in Beirut. From Thelwall (1990:38) (Notice that these values vary between regions across North Africa and West Asia)

Vowel chart representing the pronunciation of diphthongs by a Palestinian speaker educated in Beirut. From Thelwall (1990:38)

Modern Standard Arabic has six vowel phonemes forming three pairs of corresponding short and long vowels (/a, aː, i, iː, u, uː/). Many spoken varieties also include /oː/ and /eː/. Modern Standard Arabic has two diphthongs (formed by a combination of short /a/ with the semivowels /j/ and /w/). Allophony in different dialects of Arabic can occur, and is partially conditioned by neighboring consonants within the same word. As a general rule, for example, /a/ and /aː/ are:

• /a, aː/
  • retracted to [ɑ] in the environment of a neighboring /r/, /q/ or an emphatic consonant (one that is uvularized, though customarily transcribed as if pharyngealized): /sˤ/, /dˤ/, /tˤ/, /ðˤ/, /ɫ/ and in a few regional standard pronunciations also /x/ and /ɣ/;[4]
  • only in Iraq and the Persian Gulf: [ɐ] before a word boundary;[4]
  • advanced to [æ] in the environment of most consonants:
    • labial consonants (/m/, /b/ and /f/),
    • plain (non-emphatic) coronal consonants with the exception of /r/ (/θ/, /ð/, /n/, /t/, /d/, /s/, /z/, /l/, /ʃ/ and /d͡ʒ~ɡ~ʒ/)
    • pharyngeal consonants (/ħ/ and /ʕ/)
    • glottal consonants (/h/ and /ʔ/)
    • /j/, /k/ and /w/;[5]
  • Across North Africa and West Asia, the allophones [æ] and [ɑ] may be realized differently, either as [a ~ ɑ ~ ɛ], or both as [a ~ ä]
  • In northwestern Africa, the open front vowel /æ/ is raised to [ɛ] or [e].
• /i, iː, u, uː/
  • Across North Africa and West Asia, /i/ may be realized as [ɪ ~ e ~ ɨ] before or adjacent to emphatic consonants and [q], [r], [ħ], [ʕ]. /u/ can also have different realizations, i.e. [ʊ ~ o ~ ʉ]. Sometimes with one value for each vowel in both short and long lengths or two different values for each short and long lengths. They can be distinct phonemes in loanwords for a number of speakers.
  • In Egypt, close vowels have different values; short initial or medial: [e], [o] ← instead of /i, u/. /i~ɪ/ and /u~ʊ/ completely become [e] and [o] respectively in some other particular dialects. Unstressed final long /aː, iː, uː/ are most often shortened or reduced: /aː/ → [æ ~ ɑ], /iː/ → [i], /uː/ → [o~u].

However, the actual rules governing vowel-retraction are a good deal more complex, and have relatively little in the way of an agreed-upon standard, as there are often competing notions of what constitutes a "prestige" form.[7] Often, even highly proficient speakers will import the vowel-retraction rules from their native dialects.[8] Thus, for example, in the Arabic of someone from Cairo emphatic consonants will affect every vowel between word boundaries, whereas certain Saudi speakers exhibit emphasis only on the vowels adjacent to an emphatic consonant.[9] Certain speakers (most notably Levantine speakers) exhibit a degree of asymmetry in leftward vs. rightward spread of vowel-retraction.[9][10]

The final heavy syllable of a root is stressed.[6]

The short vowels [u, ʊ, o, o̞, ɔ] are all possible allophones of /u/ across different dialects, e.g. قُلْت /ˈqult/ ('I said') is pronounced [ˈqʊlt] or [ˈqolt] or [ˈqo̞lt] since the difference between the short mid vowels [o, o̞, ɔ] and [u, ʊ] is never phonemic and they're mostly found in complementary distribution, except for a number of speakers where they can be phonemic but only in foreign words.

The short vowels [i, ɪ, e, e̞, ɛ] are all possible allophones of /i/ across different dialects, e.g. مِن /ˈmin/ ('from') is pronounced [ˈmɪn] or [ˈmen] or [ˈmɛn] since the difference between the short mid vowels [e, e̞, ɛ] and [i, ɪ] is never phonemic and they're mostly found in complementary distribution, except for a number of speakers where they can be phonemic but only in foreign words.

The long mid vowels /oː/ and /eː/ appear to be phonemic in most varieties of Arabic except in general Maghrebi Arabic where they merge with /uː/ and /iː/. For example لون ('color') is generally pronounced /loːn/ in Mashriqi dialects but /luːn/ in most Maghrebi Arabic. The long mid vowels can be used in Modern Standard Arabic in dialectal words or in some stable loanwords or foreign names.[11] as in روما /ˈroːma/ ('Rome') and شيك /ˈʃeːk/ ('cheque').

Foreign words often have a liberal sprinkling of long vowels, as their word shapes do not conform to standardized prescriptive pronunciations written by letters for short vowels.[12] The long mid vowels /eː/ and /oː/ are always rendered with the letters ي and و, respectively. In general, the pronunciation of loanwords is highly dependent on the speaker's native variety.

Even in the most formal of conventions, pronunciation depends upon a speaker's background.[13] Nevertheless, the number and phonetic character of most of the 28 consonants has a broad degree of regularity among Arabic-speaking regions. Note that Arabic is particularly rich in uvular, pharyngeal, and pharyngealized ("emphatic") sounds. The emphatic coronals (/sˤ/, /dˤ/, /tˤ/, and /ðˤ/) cause assimilation of emphasis to adjacent non-emphatic coronal consonants. The phonemes /p/ ⟨پ⟩ and /v/ ⟨ڤ⟩ (not used by all speakers) are not considered to be part of the phonemic inventory, as they exist only in foreign words and they can be pronounced as /b/ ⟨ب⟩ and /f/ ⟨ف⟩ respectively depending on the speaker.[12][14] The standard pronunciation of ⟨ج⟩ /d͡ʒ/ varies regionally, most prominently [d͡ʒ] in the Arabian Peninsula, parts of the Levant, Iraq, northern Algeria and Sudan, it is also considered as the predominant pronunciation of Literary Arabic outside the Arab world, [ʒ] in most of Northwest Africa and the Levant, and [g] in most of Egypt and a number of Yemeni and Omani dialects.

Note: the table and notes below discusses the phonology of Modern Standard Arabic among Arabic speakers not individual dialects; Algerian, Egyptian, Syrian, etc.

Long (geminate or double) consonants are pronounced exactly like short consonants, but last longer. In Arabic, they are called mushaddadah ("strengthened", marked with a shaddah), but they are not actually pronounced any "stronger". Between a long consonant and a pause, an epenthetic [ə] occurs,[6] but this is only common across regions in West Asia.

Arabic syllable structure can be summarized as follows, in which parentheses enclose optional components:

• (C₁) (S₁) V (S₂) (C₂ (C₃))

Arabic syllable structure consists of an optional syllable onset, consisting of one or two consonants; an obligatory syllable nucleus, consisting of a vowel optionally preceded by and/or followed by a semivowel; and an optional syllable coda, consisting of one or two consonants. The following restrictions apply:

• Onset
  • First consonant (C₁): Can be any consonant, including a liquid (/l, r/). (Onset is composed only of one consonant; consonant clusters are only found in loanwords, sometimes an epenthetic /a/ is inserted between consonants.)
• Nucleus
  • Semivowel (S₁)
  • Vowel (V)
  • Semivowel (S₂)
• Coda
  • First consonant (C₂): Can be any consonant.
  • Second consonant (C₃): Can also be any consonant.

The placement of word stress in Arabic varies considerably from one dialect to another, and has been the focus of extensive research and debate.

In determining stress, Arabic distinguishes three types of syllables:[25]

• Light:
  • An open syllable containing a short vowel (i.e. CV), such as wa 'and'
• Heavy:
  • An open syllable containing a long vowel (i.e. CVV), such as sā.fara 'he travelled'
  • A closed syllable containing a short vowel followed by one consonant (i.e. CVC), such as min 'from' or ka.tab.tu 'I wrote'
• Super-heavy:
  • A closed syllable containing a long vowel followed by one consonant (i.e. CVVC), such as bāb# 'door' or mād.dun 'stretching (NOM)'
  • A closed syllable containing a vowel of any length followed by two consonants (i.e. CVCC, CVVCC), such as bint# 'girl' or mādd# 'stretching'

The word stress of Classical Arabic has been the subject of debate. However, there is consensus as to the general rule, even though there are some exceptions. A simple rule of thumb is that word-stress falls on the penultimate syllable of a word if that syllable is closed, and otherwise on the antepenultimate.[26]

A more precise description is J. C. E. Watson's. Here the stressed syllable follows the marker ' and variant rules are in brackets:[27]

1. Stress a pre-pausal superheavy (CVVC, CVVGG, or CVCC) syllable: [kiˈtaːb] ‘book’, [ˈmaːdd] ‘stretching (MASC SG)’, [ʃaːˈribt] ‘I/you (MASC SG) drank’.
2. Otherwise, stress the rightmost (ending) non-final heavy (CVV, CVC, or CVVG) syllable (up to the antepenult): [daˈrasnaː] ‘we learnt’, [ṣaːˈbuːnun] ‘soap (NOM)’, [ˈmaktabah] ‘library’, [ˈmaːddun] ‘stretching (NOM)’, [ˈmaktabatun] ‘library’ (non-pause) (or [makˈtabatun]).
3. Otherwise, stress the leftmost (beginning) CV syllable (or antepenult): [ˈkataba] ‘he wrote’, [ˈkatabatuhu] ‘she wrote it’ (or [kataˈbatuhu]).

Modern Arabic dialects all maintain rules (1) and (2). But if there is neither a final superheavy syllable nor a heavy penultimate syllable, their behaviour varies. Thus in Palestinian, rule (3) is instead 'otherwise stress the first syllable (up to the antepenult): [ˈkatab] ‘he wrote’, [ˈzalama] ‘man’', whereas the basic rules of Cairene (to which there are exceptions) are:[28]

1. Stress a superheavy ultima.
2. Otherwise, stress a heavy penult.
3. Otherwise, stress the penult or antepenult, whichever is separated by an even number of syllables from the rightmost non-final heavy syllable, or, if there is no non-final heavy syllable, from the left boundary of the word.

Spoken varieties differ from Classical Arabic and Modern Standard Arabic not only in grammar but also in pronunciation. Outside of the Arabian peninsula, a major linguistic division is between sedentary varieties, largely urban varieties. Inside the Arabian peninsula and in Iraq, the two types are less distinct; but the language of the urbanized Hejaz, at least, strongly looks like a conservative sedentary variety.

Some examples of variation:

Consonants

In Modern Standard Arabic (not in Egypt's use), /ɡ/ is used as a marginal phoneme to pronounce some dialectal and loan words. On the other hand, it is considered a native phoneme or allophone in most modern Arabic dialects, mostly as a variant of ق /q/ (as in Arabian Peninsula and Northwest African dialects) or as a variant of /d͡ʒ/ ج (as in Egyptian and a number of Yemeni and Omani dialects). It is also considered a separate foreign phoneme that appears only in loanwords, as in most urban Levantine dialects where ق is /ʔ/ and ج is /d͡ʒ~ʒ/.

The phoneme represented by the Arabic letter ǧīm (ج) has many standard pronunciations: [d͡ʒ] in most of the Arabian Peninsula and as the predominant pronunciation of Literary Arabic outside the Arab world, [ɡ] in most of Egypt and some regions in southern Yemen and southwestern Oman. This is also a characteristic of colloquial Egyptian and southern Yemeni dialects.[17] In Morocco and western Algeria, it is pronounced as [ɡ] in some words, especially colloquially. In most north Africa and most of the Levant, the standard is pronounced [ʒ], and in certain regions of the Persian Gulf colloquially with [j]. In some Sudanese and Yemeni dialects, it may be either [ɡʲ] or [ɟ] as it used to be in Classical Arabic.

The foreign phonemes /p/ and /v/ are not necessarily pronounced by all Arabic speakers, but are often pronounced in names and loanwords. /p/ and /v/ are usually transcribed with their own letters ﭖ /p/ and ﭪ /v/ but as these letters are not present on standard keyboards, they are simply written with ب /b/ and ف /f/, e.g. both نوفمبر and نوڤمبر /nu(ː)fambar/, /novambar, -ber/ or /nofember/ "November", both كاپريس and كابريس /ka(ː)pri(ː)s, ka(ː)bri(ː)s/ "caprice" can be used.[12][14] The use of both sounds may be considered marginal and Arabs may pronounce the words interchangeably; besides, many loanwords have become Arabized, e.g. باكستان or پاکستان /pa(ː)kistaːn, ba(ː)kistaːn/ "Pakistan", فيروس or ڤيروس /vi(ː)ru(ː)s, vajru(ː)s/ "virus".

/t͡ʃ/ is another possible loanword phoneme, as in the word سندوتش‎ or ساندوتش‎ (sandawitš or sāndwitš 'sandwich'), though a number of varieties instead break up the [t] and [ʃ] sounds with an epenthetic vowel.[29] Egyptian Arabic treats /t͡ʃ/ as two consonants ([tʃ]) and inserts [e], as [teʃC] or [Cetʃ], when it occurs before or after another consonant. /t͡ʃ/ is found as normal in Iraqi Arabic and Gulf Arabic.[30] Normally the combination تش (tā’-shīn) is used to transliterate the [tʃ]. Otherwise Arabic usually substitutes other letters in the transliteration of names and loanwords like the Persian character چ which is used for writing [tʃ]

Other Variations include:

• Split of original /r/ into two phonemes, distinguished primarily by how they affect neighboring vowels. This has progressed the farthest in North Africa. See Moroccan Arabic, Algerian Arabic, Tunisian Arabic and Libyan Arabic
• Loss of the glottal stop in places where it is historically attested, as in /samaːʔ/ → /sama/.

Vowels

• Development of highly distinctive allophones of /a/ and /aː/, with highly fronted [a(ː)], [æ(ː)] or [ɛ(ː)] in non-emphatic contexts, and retracted [ɑ(ː)] in emphatic contexts. The more extreme distinctions are characteristic of sedentary varieties, while Bedouin and conservative Arabian-peninsula varieties have much closer allophones. In some of the sedentary varieties, the allophones are gradually splitting into new phonemes under the influence of loanwords, where the allophone closest in sound to the source-language vowel often appears regardless of the presence or absence of nearby emphatic consonants.
• Spread of "emphasis", visible in the backing of phonemic /a(ː)/. In conservative varieties of the Arabic peninsula, only /a/ adjacent to emphatic consonants is affected, while in Cairo, an emphatic consonant anywhere in a word tends to trigger emphatic allophones throughout the entire word. Dialects of the Levant are somewhere in between. Moroccan Arabic is unusual in that /i/ and /u/ have clear emphatic allophones as well (typically lowered, e.g. to [e] and [o]).
• Monophthongization of diphthongs such as /aj/ and /aw/ to /eː/ and /oː/, respectively (/iː/ and /uː/ in parts of the Maghrib, such as in Moroccan Arabic). Mid vowels may also be present in loanwords such as ملبورن (/milboːrn/ Melbourne), سكرتير (/sikriteːr/ '(male) secretary') and دكتور (/duktoːr/ 'doctor').[11]
• Raising of word final /a/ to [e]. In some parts of Levant, also word-medial /aː/ to [eː]. See Lebanese Arabic.
• Loss of final short vowels (with /i/ sometimes remaining), and shortening of final long vowels. This triggered the loss of most Classical Arabic case and mood distinctions.
• Collapse and deletion of short vowels. In many varieties, such as North Mesopotamian, many Levantine dialects, many Bedouin dialects of the Maghrib, and Mauritanian, short /i/ and /u/ have collapsed to schwa and exhibit very little distinction so that such dialects have two short vowels, /a/ and /ə/. Many Levantine dialects show partial collapse of /i/ and /u/, which appear as such only in the next-to-last phoneme of a word (i.e. followed by a single word-final consonant), and merge to /ə/ elsewhere. A number of dialects that still allow three short vowels /a/ /i/ /u/ in all positions, such as Egyptian Arabic, nevertheless show little functional contrast between /i/ and /u/ as a result of past sound changes converting one sound into the other.[31] Arabic varieties everywhere have a tendency to delete short vowels (especially other than /a/) in many phonological contexts. When combined with the operation of inflectional morphology, disallowed consonant clusters often result, which are broken up by epenthetic short vowels, automatically inserted by phonological rules. In these respects (as in many others), Moroccan Arabic has the most extreme changes, with all three short vowels /a/, /i/, /u/ collapsing to a schwa /ə/, which is then deleted in nearly all contexts. This variety, in fact, has essentially lost the quantitative distinction between short and long vowels in favor of a new qualitative distinction between unstable "reduced" vowels (especially /ə/) and stable, half-long "full" vowels /a/, /i/, /u/ (the reflexes of original long vowels). Classical Arabic words borrowed into Moroccan Arabic are pronounced entirely with "full" vowels regardless of the length of the original vowel.

The main dialectal variations in Arabic consonants revolve around the six consonants; ⟨ج⟩, ⟨ق⟩, ⟨ث⟩, ⟨ذ⟩, ⟨ض⟩ and ⟨ظ⟩:

The most frequent consonant phoneme is /r/, the rarest is /ðˤ/. The frequency distribution of the 28 consonant phonemes, based on the 2,967 triliteral roots listed by Wehr[14] is (with the percentage of roots in which each phoneme occurs):

This distribution does not necessarily reflect the actual frequency of occurrence of the phonemes in speech, since pronouns, prepositions and suffixes are not taken into account, and the roots themselves will occur with varying frequency. In particular, /t/ occurs in several extremely common affixes (occurring in the marker for second-person or feminine third-person as a prefix, the marker for first-person or feminine third-person as a suffix, and as the second element of Forms VIII and X as an infix) despite being fifth from last on Wehr's list. The list does give, however, an idea of which phonemes are more marginal than others. Note that the five least frequent letters are among the six letters added to those inherited from the Phoenician alphabet, namely, ḍād, ṯāʾ, ḫāʾ, ẓāʾ, ḏāl and ġayn.

The Literary Arabic sample text is a reading of The North Wind and the Sun by a speaker who was born in Safed, lived and was educated in Beirut from age 8 to 15, subsequently studied and taught in Damascus, studied phonetics in Scotland and since then has resided in Scotland and Kuwait.[36]