Harry van der Hulst and Sam Hellmuth 1. Introduction 1 This chapter deals with accentual systems in the languages spoken in the Middle East and includes a discussion of a number of isolate ancient lan- guages and some Afro-Asiatic languages, notably Egyptian and Semitic languages. The notion Middle East is not well-dened linguistically, and some languages that could have been included here are dealt with in other chapters. All Altaic languages, Indo-Iranian languages (such as Kurdish and Avestan) are discussed in Schiering and van der Hulst (this volume). All Caucasian languages as well as Indo-European languages such as Hittite (and related Anatolian languages) and Armenian are treated in van der Hulst (this volume). Afro-asiatic languages spoken in North Africa can be found in Downing (this volume). This survey, incomplete as it is, was included because the relevant area was not included in the areas covered in the other chapters of this volume. The organization of each section (or subsection) with accentual data is as follows: a. Genetic structure of the (sub)family; these are based on sources such as Ruhlen (1991), Comrie et al (2003), the Ethnologue (15th edition) and several others of the many sources that oer classications. We have tried to strike a compromise in cases of conicting groupings and no claim is made here that the resulting groupings are superior to those oered in other sources. In each case (sub)family names are presented in capitals. We have not included information of the numbers of lan- guages per (sub)family and in most cases list only a (sometimes rather arbitrary) subset of the languages in each (sub)family. Languages that are included in StressTyp (see section 3) are indicated in bold. Lan- guages that are not in StressTyp about which this chapter provides information are underlined. 1. We wish to acknowledge the help of various people who gave feedback on parts of this chapter: Bob Hoberman, Janet Watson and Alan Kaye. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 615) b. Extracts from StressTyp entries (language name followed by StressTyp Code, and examples). More complete extracts (including references) are oered in Part II of this volume and, of course, in StressTyp itself. We included the Stress information unchanged, i.e. as it is in the data- base and Part II, despite the fact that in some cases this information has been questioned by specialists. (Future work on StressTyp must, of course, aim at removing or changing such information.) c. Additional accent information: This might involve additional informa- tion on languages already in StressTyp, or information on languages that are not in StressTyp. (In some larger families that are treated in one section we have presented the StressTyp extracts and additional information per subfamily.) d. Generalizations: We have tried to make general statements about the accentual patterns in the relevant (sub)family, in some cases accom- panied by remarks about diachronic developments. If (b) is absent, (c) will be labeled accent information. Sometimes (c) and (d) are conated into a single section. 2. A note on the information on which this chapter is based The rst source for the accentual data presented here is StressTyp (see chapter 1, 2, 11 and Part II of this book for information on this database). Secondly, we consulted books that oer surveys of language families or languages in a specic geographical area. Thirdly, we have consulted grammars of individual languages and, fourthly, we have sent email queries to colleagues; where we rely on information that they directly have given to us (via email or in personal communication) we note this in the text. In line with previous work on word accentual systems (van der Hulst 1999, 2002, 2005), we use the term word accent where many others would use the term word stress (as in done in StressTyp). We refer to chapter 1 of this volume for a justication and clarication of this termi- nological choice. 3. Isolate ancient languages There are many now extinct languages from this area. Here we provide some information about four languages. Others that are not dealt with include: Median, Ancient Macedonian, Had(r)ani, Minaean, Old Nubian, 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 616) 616 Harry van der Hulst and Sam Hellmuth Sabaean, many of which are known too fragmentedly to provide researchers with information on word accent. USumerian. This language isolate (once suspected to be an articial, secret language), spoken in Southern Mesopotamia, is known rst from clay tablets found in Uruk from around 3200 BCE. It was superseded by the unrelated Semitic language Akkadian (cf. below) in the beginning of the 2nd millenium, but remained in limited use for 3000 years. Michalowski (2004) dismisses the suggestion that Sumerian, an agglutinative language, might have been a tonal language which has been suggested because the language otherwise has an unusually high number of homophones. He does not provide further information about word prosodic properties. Hayes (1997: 1013) agrees that very little is known about word accent in Sumerian. Vowel deletions suggest a strong stress accent: amar-utu-(k) bull of the Sun > Akkadian: marduk suggests a second syllable accent: amarutuk. Not enough such examples are known to suggest whether stress was lexi- cally marked or predictable by rule. A recent grammar is Edzard (2003). UElamite (also known as Scythian, Median, Susian, Anzanite), the ocial language of the Persian empire from the 6th to the 4th century BCE. A genetic link to the Dravidan family has been suggested. Stolper (2004: 73) thinks that accent could have been non-nal, perhaps initial. UHurrian and UUrartian (Vannian, Chaldean). These languages are related to each other, with no known genetic connection to any other language family, although links to Northeast Caucasian (notably Georgian) have been proposed. Wilhelm (2004a: 100) says that Hurrian had a penulti- mate accent on words (including suxed words), but excluding enclitics. Wilhelm (2004b: 123) suggest the same pattern for Urartian. 4. Afro-Asiatic: Egyptian and Semitic Genetic information The Afro-Asiatic family contains: EGYPTIAN SEMITIC BERBER CHADIC CUSHITIC OMOTIC 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 617) Word accent systems in the Middle East 617 Omotic is sometimes regarded as a sister to the rest forming a group, while Semitic and Berber are usually seen as a subgroup. This family extends over North Africa, the Middle East and Asia Minor. In this section we will deal with the Afro-Asiatic languages that are spoken outside Africa, roughly in the Middle East. This means that we will deal with the Egyp- tian branch and the Semitic branch (excluding Ethiopian Semitic and Western Arabic languages). Egyptian, Semitic and Berber do not have tonal languages, whereas the other three subfamilies do. We refer to Down- ing (this volume) for some brief remarks about Afro-Asiatic languages spoken outside Africa and more elaborate discussion of Afro-Asiatic languages in Africa. StressTyp contains information about one Cushitic language outside Africa, Beja (spoken in parts of Egypt, Sudan and Eritrea): 2 Beja; Bedawi; Bischari [A/P] f Stress falls on heavy penultimate syllables. f If the penult is light, stress falls on the antepenultimate syllable. f In bisyllabic words stress is penultimate unless the penult is light and the nal is not. ga'na:j gazelle 'enton here 'an/alan I cursed sa'no:ku your brother We refer to Downing (this volume) for a discussion of this and other Cushitic languages. 4.1. Egyptian (<Afro-Asiatic) Genetic information ANCIENT EGYPTIAN UAncient Egyptian, UCoptic There is a written record from around 3200 to 400 BCE. A distinction is made (with further divisions) between early Egyptian (32001300 BCE) and late(r) Egyptian (1300 BCE5th century AD), the last phase being called Demotic Egyptian. Coptic, as known since the 4th century AD, the latest phase of this language, went extinct in the late 17th. (The language name Modern Egyptian refers to a form of Arabic; cf. below.). See Loprieno (1995) for a general overview. 2. This language is sometimes seen as a separate branch of Afro-Asiatic. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 618) 618 Harry van der Hulst and Sam Hellmuth Accent information Loprieno (1997, 2004) provides the following information about earlier or historical Egyptian. The accent was on the ultimate or penultimate syllable: (1) wabaa_ to become white jafdaaw four saatpaw is chosen saa:tap to choose CVC and CV syllables occur in all positions in the word, but accented penultimate vowels in open syllables are always long. Word-nally, an extra C could be present, thus allowing CVCC. Perhaps there was an earlier APU accent location which became integrated in the above pattern by loss of the PU vowel: (2) _u piraw > _u praw transformation Indeed, in later periods unaccented vowels would reduce and delete which leads to complex consonant clusters, as can be noted in Coptic. 4.2. Semitic (<Afro-Asiatic) Genetic information SEMITIC EAST: UAkkadian, UEblaite WEST: UAmorite, UUgaritic ARAMAIC: Western: UNabataean Western Middle Aramaic languages Samaritan Aramaic, UJewish Palestian Ar. Western Neo-Aramaic: Maaloula Eastern: UBiblical A., Syriac, Turoyo, UMlahso, Mandaic, Judaeo-Aramaic CANAANITE: UEdomite, UMoabite, UAmmonite Hebrew: Biblical (Tiberian) Hebrew, Mod. Hebrew UPhoenician/Punic ARABIC: See below SOUTH Western ANCIENT SOUTH ARABIAN: Sabean, Qatabanian, Hadhramautic, Minean ETHIOPIAN: North: UGeez, Tigrinya, Tigre, Dahlik South: Amharic, Harari, Gurage Eastern: Mehri, Harsusi, Bathari, Hobyot, Jibbali, Soqotri 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 619) Word accent systems in the Middle East 619 Here Arabic is seen as a separate branch of Central Semitic. Woodard (2004) has a distinction between Ancient South Arabian and Ancient North Arabian. However, the latter group concerns older forms of Arabic, which is here placed in Central Semitic. See also Bergstrasser (1983), Hetzron (1997) and Faber (1997). The living languages that are spoken outside Africa are Neo-Aramaic languages, Hebrew, Eastern South Semitic and varieties of Arabic. Accent information East Semitic UAkkadian (which splits into Babylonian and Assyrian), known from the later half of the 3rd millennium, replaced Sumerian (see above) in the beginning of the 2nd millennium BC and was then itself replaced by Aramaic during the rst half of the rst millennium (although surviving until 100 AD) (G. Gordon 1997). Buccellati (1997) subscribes to the view that accent falls on the rightmost long vowel (excluding the nal morphe- mic vowel). According to an alternative view accent is initial, but it might be the case that this perhaps obtains in the absence of long vowels. This would indicate that Akkadian had an unbounded last/rst system. This is conrmed by Huehnergard and Woods (2004: 234) who say that accent falls on the ultimate syllable if closed and has a long vowel (superheavy), or when containing a circumex vowel (a vowel resulting from contrac- tion, perhaps counting as two syllables). Otherwise the accent falls on the rightmost nonnal syllable which is either closed or contains a long vowel. If there are no heavy syllables at all, accent is initial: (3) idu # k s/he killed ibnu they built iparras s/he will cut ma # rum son nadin is given lu gods Knudsen (1980) also presents an analysis of Akkadian accent. He notes that there is no evidence for secondary accents. Primary accent falls on the nal or penultimate syllable. If the nal syllable was short and the penultimate had a long vowel, accent was penultimate. A nal closed syllable with a long vowel would have primary accent. This would also be the case if the nal syllable contained a circumex vowel. He also notes that accent would be penultimate if the last two syllables were light, even if the antepenultmate vowel were long. In case words only have light 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 620) 620 Harry van der Hulst and Sam Hellmuth syllable, he notes, that there is no evidence for accent. These observa- tions are not incompatible with those made by Huehnergard and Woods (cf. above), except that their unbounded account would suggest that in the last mentioned case (heavylightlight) accent would be ante- penultimate. Knudsens observations suggest a bounded rather than an unbounded system. No information on accent could be found for UEblaite and UAmorite (only known through proper names in Akkadian texts); cf. Gordon (1997a,b,c). West Semitic: Central StressTyp extracts Aramaic [P] Afro-Asiatic, Semitic, Central, Aramaic. East Kurdistan, Israel. Palmyra, Sinai. f Stress falls on the penultimate syllable. f Epenthetic vowels do not count for stress assignment. 'arba four cay'xa:na teashop da'qiqa ne We did not nd information on UAmmonite, UUgaritic (Gordon 1997a,b; Dennis 1997, 2004). Turning to Aramaic, Creason (2004: 400) says that in Proto-Aramaic a nal closed syllable was accented, otherwise accent was on the penult (even if the nal vowel was long). Then nal short vowels would be deleted, or lengthened. In case of deletion, the penultimate open syllable now would become a nal closed syllable, with accent. This basic pattern (nal if closed) remains constant throughout the history of Aramaic, but in some late Aramaic dialects, nal accent shifted to the penult in some or all words. In line with this Kaufman (1997: 121) states that Classical Aramaic is said to have nal accent (where he apparently ignored the penultimate location when the nal syllable is open), while the modern languages have penultimate stress. Western Aramaic: In his discussion of Neo-Aramaic languages (E.g. Maaloula) Jastrow (1997: 336) says that in Western Neo-Aramaic word accent is usually on the penultimate. The last syllable is accented if it has a long vowel or ends in two or more consonants. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 621) Word accent systems in the Middle East 621 Eastern Aramaic: Daniels (1997: 137) discussing Syriac accent refers to the view that accent is penult among the Nestorians, but on nal closed syllables and other- wise on the penult for Maronites. The latter pattern is the norm today among Chaldeans (these terms referring to dierent Christian traditions or groups). Malone (1997: 156) proposes that accent in Classical Mandaic falls on the last vowel of the word. In Modern Mandaic, according to Malone (1997: 149), accent falls on the rightmost, non-nal full vowel. This sug- gests an unbounded system, but no default clause is mentioned in case no full vowel is present. As for the Eastern Neo-Aramaic (ENA) languages Jastrow (1997: 353) reports that in Mlahso and in the North-Eastern (NENA) languages nouns are accented on the last syllable, but in the majority of ENA lan- guages accent has shifted to the penult. Accent moves to the (new) penult when suxes are added, although in verbal forms accent will stay on the original penult when further suxes are added. This penult location is, for example, found in Turoyo: (4) Turoyo: h ozeno I (m.) see h ozenole I (m.) see him He also says that perhaps in these varieties accent has become phonemic in the sense that across verb classes dierent patterns may obtain: (5) Turoyo: mal m he collects mal m collect! Whereas Jastrow says that in NENA languages accent is nal in nouns, Hoberman (1997: 330 .) reports that accent in the NENA dialects spoken in northwestern Iraq, except in verbs, uniformly falls on the penult. In verbs, where penult is still the default, accent placement is governed by the morphology. Consider the following minimal pair (representative of Modern Aramaic NENA dialects): (6) ma:lIple teach him (imp. sing.) ma:l ple that he teach him He adds the following generalizations: (7) a. In the imperative form accent is initial b. Accent is APU when a surface penult /i:/ or /u:/ is derived from an underlying non-syllabic semi-vowel. c. Certain verbal suxes cannot be accented, such as the past tense sux /wa:/. When such suxes occur accent occurs earlier in the word. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 622) 622 Harry van der Hulst and Sam Hellmuth Two word combinations in which the second word cliticizes to the rst show a reduction of the accent on the second word. Canaanite StressTyp extracts Hebrew, Modern [U;P] Afro-Asiatic, Semitic, Central, South, Canaanite. Israel. f Primary stress is mostly nal, but sometimes phonemically on the penult. f Secondary stress on alternate syllables to the left of the main stress. g a'doI big ta'am tasted mcvuga'iIm adults 'taam taste Hebrew, Tiberian (dialect of Hebrew; Ivrit) [U/P] Afro-Asiatic, Semitic, Central, South, Canaanite. Lake Tiberia (Israel). f Main stress is assigned to the nal syllable if it is closed. f Otherwise stress is penultimate (but may shift again to the nal syllable). f Secondary stress two syllables to the left of the main stress and long vowels before that. 'qa:mu: arise third.pl. Llab'lcm write scnd pl masc Additional information UPunic is a late dialect of UPhoenician through we which we get most information about earlier phases of Phoenician (Hackett 2004; Segert 1997a,b). This language disappears in the eastern Mediterranean area during the rst century BCE and in North Africa in the fth century AD. Segert (1997: 63) suggests that the position of word accent can be deter- mined on the last syllable in most cases. We have no information on UMoabite, UEdomite and UAmmonite. Turning to Hebrew, Steiner (1997: 149) says that in Biblical (Tiberian) Hebrew primary stress is on the ultimate or penultimate syllable, with an increased tendency toward the ultimate; see also Khan (1997) and Rendsburg (1997). In Modern Hebrew (Bat-El 1993, Bolozky 1997, Berman 1997, Con and Bolozky 2005) word accent follows the Sephard tradition. It is nal, with (sometimes systematic) exceptions: 3 3. We thank Shmuel Bolozky for supplying us with additional information. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 623) Word accent systems in the Middle East 623 (9) a. In general, penultimate accent may result when suxes remain neutral such that the accent remains on the nal syllable of the base. This happens in the verbal system where (except in the present participle) accent lies on the stem-nal vowel: katavti I wrote dibarta you spoke nimcenu we were found ip la she dropped (trans.) b. Also in nouns that are acronyms or frequently used, accent can be stable under suxation: mankal general manager mankalim (plural) rasar rst sergeant major rasarim (plural) tu t water melon tu tim (plural) cxo k laughter cxo kim joyous moment c. Stable accent is also found in names for residents: telav v Tel Aviv telav vi resident of . . . xolo n Holon xolo ni resident of . . . d. Certain derived (segolate) nouns like meser messenger; here, historically a vowel broke up a nal impermissible cluster. e. Insertion of /a/ causes penultimate stress in another group of words: gavo a tall (m.sg.) poteax open (m.sg.) f. Penultimate accent occurs in names of childrens games: Sh ra (name) and klatim (childrens game); compare: klaf m cards klam gard game monopo l monopoly mono pol Monopoly (game) g. Due to the inuence of Yiddish accent, we nd a penultimate location in colloquial style of proper names: Yael PYael xa m life Xaim proper name In general it would seem that in informal speech, penultimate accent may take over from the more formal nal accent. h. In nonverbal forms there are sporadic exceptions to the nal accent pattern: lama where (h)ena here i. Loan words (like akademiya, instalator) can be exceptional, 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 624) 624 Harry van der Hulst and Sam Hellmuth In words that have exceptional accent (when no suxes are involved) accent appears to be stable under suxation (10) t ras t ras im corn plural This especially applies to borrowings which keep their original accent loca- tion even if non-nal and this position is also maintained under suxation. Bat-El (1993) analyses the accent system of Modern Hebrew nouns, which involves lexical marking of stems and suxes. It would seem that primary accent location follows a LAST/LAST pattern: it falls on the last lexically accented syllable, or, if no lexical accents are present, on the last syllable. Modern Hebrew has regular alternating secondary accents going left- ward away from the primary accent. South Semitic: For the Eastern South Semitic (the modern Subarabic or south Arabian) languages, Lonnet and Simeone-Senelle (1997: 354) describe the location of accent as follows: accent falls on the nal CVC(C) or CV: (C) syllable and if there is no such syllable on the rst CVC syllable. This suggests an unbounded Last/First pattern, although no mention is made of the loca- tion in words that also lack CVC; some examples suggest that accent is initial in that case. Simeone-Senelle (1997: 386) states that in Mehri, Hobyo#t, Bat h ari and Harsu#si accent is on the last strong syllable, or on the rst syllable if all vowels are short. This conrms the Last/First pat- tern, though Janet Watson (p.c.) points out that in at least some dialects of Mehri the Last/First pattern only holds of stems (not inected words) and that in any case strings of short syllables are rare and thus initial stress is itself rare. In contrast, Simeone-Senelle (1997: 386) notes that in Jibba#li a word can have several accents, whereas in Soqot there is a general trend to have initial accent. As for the Old/Ancient South Arabian languages belonging to the Western South Semitic, Graag (1997) and Nebes and Stein (2004) provide no information on accent. 4.3. Arabic languages (<Semitic < Afro-Asiatic) Genetic information A standardized version of Arabic, the language of the Quran and early Islamic literature arose in the 8th century. This version came to be known as Classical Arabic (Fischer 1997; Owens 2006) which develops in the 19th 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:53) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 625) Word accent systems in the Middle East 625 and 20th century into Modern Standard Arabic which is used in most sectors of public life. Meanwhile, many more conservative Bedouin (rural) and sedentary (urban) varieties of Arabic develop which are usually grouped into a Western and the Middle Eastern group. The Western group (the Maghrebi or North African dialects) covers Libya, Tunisia, Algeria, Morocco, Mauritania, and the outlying Maltese variety (Borg and Azzopardi-Alexander (1997) (as well as now extinct Andalusian Arabic and Siculo Arabic). The Middle Eastern group covers the Arabian Peninsula, Iraq, Iran, Sudan, Syria-Lebanon-Palestine, Jordan, Egypt and areas in Afghanistan and Uzbekistan, as well as an outlying variety in Cyprus (Alexander (1997). Most entries in StressTyp (nearly all of which are analyzed in Hayes 1995) come from the Middle Eastern group which can be geographically subdivided into a southern group (Arabian Penin- sula), a northern group (Levant and Mesopotamia) and a central group (including Egypt and Sudan). Linguistic studies on Arabic have a rich tradition (cf. Owens 2006; Versteegh et al. 2009) and in more recent times the variety of accentual systems has attracted wide attention (cf. Watson, to appear for an over- view of dierent approaches and some case studies). Kaye (1997) and Kaye and Rosenhouse (1997) provide general discussions of the phonology. Studies of specic accentual systems can be found in Birkeland (1954), Janssens (1972), Angoujard (1990), Hayes (1995) and Kager (2009). StressTyp extracts Arabic, Classical [L/F] f Stress the rightmost non-nal syllable that has a long vowel or closing consonant. f If there are no such syllables, stress the rst. ki'taabun book 'mamlakatun kingdom 'baIahalun date ta'mamtumaa you both completed Middle East, Southern Arabic, Bedouin Hijazi (dialect of Arabic) [U-P/A] f Superheavy (long vowel coda, or double coda) nal syllables carry stress. f In other cases, stress falls on the penult if it is heavy, otherwise on the antepenult. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 626) 626 Harry van der Hulst and Sam Hellmuth mak'tu:b written a'iabl I hit 'ma:lana our property mak'tu:fah tied f.s. ga:'bIIna meet us m.s. 'kitab he wrote 'bgaiah cow 'aza he raided Arabic, Gulf [U/P] f If the vowel of the nal syllable is long, it bears stress. f In all other cases stress is on the penult. ha'jaallIn their life jjuu'!uun they see 'beettum their house mus'laj!a hospital 'qa]bIlj your heart mIlhaav'jIIn having quarrelled with each other Middle East, Northern: Arabic, Beirut/Lebanese (dialect of Arabic, North Levantine) [U-P/A] 4 f Superheavy (long vowel coda, or double coda) nal syllables carry stress. f In other cases, stress falls on heavy penults, otherwise on the antepenult. 'daiab hit da'iabna he hit sa'aialun tree (Classical) 'daiabu they hit \aI'Iam na we teach '\aIIamu they teach Arabic, Damascene (dialect of Arabic, North Levantine) [U-P/A] f Superheavy (long vowel coda, or double coda) nal syllables carry stress. f In other cases, stress falls on the penult if it is heavy, otherwise on the antepenult. ma'da:ics schools dai'iasl I/you m.s. taught mut'lahIdc united \aI'Iamna teach pl. Arabic, South Levantine; Palestinian Arabic [L (CNT)] f Superheavy (long vowel coda, or double coda) nal syllables carry stress. 4. StressTyp has two dierent entries for Beirut and Lebanese Arabic which we have here collapsed because they represent the same dialect. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 627) Word accent systems in the Middle East 627 f Otherwise, stress penultimate syllables if heavy, and in bisyllabic words. f Heavy antepenults followed by two light syllables carry stress. f Four syllable (Classical) words with only light syllables have initial stress. f All other words ending in three light syllables have antepenultimate stress. da'iasl I studied 'maLlab oce 'Lalab he wrote 'ba:iaLo he blessed him 'sajaratun a tree (Classical) saja'ialuhu his tree (Classical) Arabic, Negev Bedouin [L (CNT)] f Stress the last syllable with a long vowel or vowel cluster (VVC or VCC). f Otherwise stress the second syllable. f There is some variation in four syllable words. ana'ma:l sheep (several) a'nam sheep !a'iasah his horse an'LIlaIav he was killed Arabic, Bani Hassan [U-P/A/P] 5 f Stress a nal superheavy syllable. f Stress the penultimate light syllable if not preceded by a heavy syllable. f Stress penultimate or antepenultimate heavy. jI'ba:b youth \aI'Iaml I taught maL'lablI my library mI'La:lIb oces 'vaIad boy 'smILah a sh \aIIa'maluh she taught him ba:ia'Laluh she blessed him 5. This system is coded in StressTyp as [L/F] but no examples are given and we nd the following verbal description: f Stress the rightmost syllable with a long vowel or coda, otherwise the rst. f Long vowels preceding the main stress have secondary stress. The generalisation above is based on the data in Irshied & Kenstowicz (1984). Hayes (1995) sees Bani Hassan Arabic as similar to Palestinian Arabic with the dierence that in a nal HLL] sequence the accent is on the H rather, as in Palestinian Arabic, on the penultimate light. He also says that accent is penultimate in HLLL]. We discuss an analysis below. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 628) 628 Harry van der Hulst and Sam Hellmuth Middle East, Central Arabic, Egyptian Radio (dialect of Arabic, Egyptian Spoken) [L (CNT)] f Superheavy (long vowel coda, or double coda) nal syllables carry stress. f Otherwise, stress penultimate syllables if heavy, and always in bisyllabic words. f Three syllable words with only light syllables have antepenultimate stress. f Four syllable words with only light syllables have antepenultimate or pre-antepenultimate stress in free variation. f If the antepenult is heavy and the penult and nal light, either it or the penult is stressed. f If the pre-antepenult is heavy, stress is always antepenultimate. f Rhythm is assigned to non-nal heavy syllables and odd light ones from left to right. sa'la:m peace Lala'bahu or 'Lalabahu he wrote it qad'damna we presented '?abadan never 'maIIL king mux'laII!a dierent Arabic, Egyptian; Cairene Arabic (classical 6 ) [L (CNT)] f Superheavy (long vowel coda, or double coda) nal syllables carry stress. f Otherwise, stress penultimate syllables if heavy. f In all other cases, stress the penult or antepenult, whichever is separated from the last heavy syllable, or the left word edge, by an even number of syllables (or zero). 'buxaIa misers \a'maIlI you f.s. did mux'laII!a dierent f.s. mai'laba mattress saLa'LIIn knives jajaiala'humaa their (dual) tree nom. (Classical) Western group (Maghrebi) Arabic, Libyan; Cyrenaican Bedouin [L (CNT)] f Bisyllables of which the rst syllable is open and the vowel is short have nal stress. 6. Note that this refers to Classical Arabic as pronounced by speakers of Cairene Arabic; of the examples given only the last word is exclusively Classical. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 629) Word accent systems in the Middle East 629 f Superheavy (long vowel coda, or double coda) nal syllables carry stress. f Otherwise, stress heavy penults or heavy antepenults followed by two light syllables. f Otherwise, stress the penult or the antepenult, whichever is separated from the closest preceding heavy syllable or the left word edge by an odd number of syllables. lI'iu:!Igun they accompanied 'maLlab oce LI'lab he wrote LI'lablan you wrote ma\iI'LIlla her quarrel !Ina'I:I cups Maltese [U/P] f Stress falls on the nal syllable if it has a long vowel or is closed by two consonants. f In all other cases stress is penultimate. bci'qu:q apricot a'zaql I have dug 'lI!Ia girl Additional information Middle East, Southern Sanaani Arabic is spoken in the Old City of the capital city of Yemen (Watson 2002, to appear), and has a complex stress system, summarized in the following algorithm from Watson (2002: 82): a) stress the rightmost non-nal CVV or CVG syllable (['ha:Laaha:] like this, [mIl'axxIia:l] late f.pl.), otherwise, b) stress a nal CVVC/CVCC syllable ([ba'na:l] girls), otherwise c) stress the rightmost non-nal CVC syllable up to the antepenult (['madiasIh] school), otherwise d) stress the leftmost CV sylla- ble (['iagabalIh] his neck). However, as Watson (2002: 81) points out, aside from the unusual fact that a CVV/CVG can attract stress away from a nal superheavy, the Sanaani stress system is in other respects similar to a Last/First pattern: a) stress a nal superheavy CVCC or CVVC syllable, otherwise b) stress the rightmost non-nal heavy syllable (up to the antepenult), otherwise c) stress the leftmost light syllable. Watson (2002: 98121) analyses the variable behaviour of CVG/CVV vs. CVC syllables in Sanaani by appeal to a two-layer metrical grid (Hayes 1995). She notes that there is considerable uctuation in stress position in connected speech, particularly in pre-pausal and post-pausal position, and that secondary stress is observed in Sanaani in words containing two or more feet. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 630) 630 Harry van der Hulst and Sam Hellmuth The dialect that McCarthy (1979: 461) refers to as Yemen Plateau Arabic is a cover term for a grouping of dialects spoken in the high plateau (Hochebene) regions of northern Yemen, taking in the following geographical areas of Yemen (Diem 1973: 127): Sanaa and environs, the plateau areas to the north and south of the capital, the North East and al-Jawf, and the South East (Al-Bayda and Hari:b). Diem (1973: 11) gives the following basic algorithm for all Yemeni dialects: a) stress a nal superheavy CVCC or CVVC syllable, otherwise b) stress the rightmost non-nal heavy syllable, else c) stress the rst syllable of the word. Diem (1973: 11) notes particularly that in the high plateau areas (only) stress is not conned to the last three syllables of the word but can occur further forward in the word, as in the example ['hamaIalch] she carried it. This is the evidence which causes McCarthy (1979: 461) to cite Yemen Plateau Arabic as one of the few contemporary spoken dialects which still show the Classical Last/First stress assignment pattern (beyond the last three syllables of the word), but contemporary speakers in fact vary in their accentuation of words of this type between the initial and penult (Janet Watson p.c.). Prochazka (1988) generalizes over all of the dialects of Saudi Arabia (including North Arabian, Najdi and Hijazi) and claims that all display a rightmost heavy else antepenult pattern, matching the pattern described above for Bedouin Hijazi Arabic. Middle East, Northern The stress patterns of Baghdadi Arabic are described in general terms in Erwin (1963: 4042) and can be paraphrased as: a) stress a nal super- heavy, else b) stress the rightmost heavy syllable, else b) stress the ante- penult (penult in disyllables). This would place Baghdadi in the U-P/A category. McCarthy & Raouli (1964: 1011) state that in words contain- ing only light syllables stress will fall on the initial syllable, but give no ex- amples of words which are longer than 3 syllables (e.g. they give ['LcIcmc] word). It is possible that the three syllable window does indeed apply, but that McCarthy & Raoullis generalisation holds because, as in Pales- tinian Arabic (Kenstowicz 1983), any 4 syllable word containing only light syllables will undergo syncope and be reduced to 3 syllables. De Jongs (2000) survey of North Sinai Bedouin Arabic dialects includes a number of distinctive stress patterns as potential dialectal markers. A small number of dialects in the survey are reported to have initial stress in words containing 4 light syllables (CaCaCaCv, without syncope/ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 631) Word accent systems in the Middle East 631 resyllabication to e.g. CCvCCv), which suggests that they retain the Classical Last/First pattern. Picketts (2006) survey of North Arabian Arabic dialects spoken by present-day nomadic groups in Syria and Lebanon, 7 are all reported to show the Classical pattern: a) stress a nal superheavy (CCVVC, CVVC, CVCC) e.g. [ljI'0I:i] many, else b) stress the rightmost heavy syllable (CCVV, CVC, CVV, CCVC) e.g. ['madiasa] school, else c) stress the initial syllable e.g. ['vasal] medium (sized). The Arabic dialect described in the literature as Bani Hassan Arabic is spoken by a Bedouin-origin community now settled in the north of Jordan (Kenstowicz 1983, Irshied & Kenstowicz 1984). The dialect displays trisyl- labic vowel deletion, a common marker of Bedouin-origin varieties (e.g. /samaL-ah/ a sh is realised as ['smILah], Irshied & Kenstowicz 1984: 137). The accent system no longer retains the Classical pattern however, since stress is conned to the last three syllables of the word, with a non- initial accent appearing in words containing 4 syllables: e.g. [\aIIa'maluh] she taught him. Note that this pattern is not due to the special status of the 3f.s. sux, as it would be in Egyptian Arabic and Lebanese Arabic in which the 3f.s. sux always attracts stress, regardless of the syllabic struc- ture of the word; in BHA the sux does not automatically attract stress: LA: [ ja:'!Ilu] she saw him/BHA ['ja:!aluh] she saw him (Irshied & Kenstowicz 1984: 129). Middle East, Central Egyptian Saiidi Arabic is spoken along the Nile Valley in Upper Egypt (south of Cairo as far as Aswan), and McCarthy (1979: 461) mentions it as a spoken dialect which retains the Classical Last/First stress assignment pattern. Khalafallah (1969) describes the stress patterns as follows: a) if there is a long vowel in the word it will bear stress (no words contain more than one long vowel) e.g. [xaIa'ga:lu] his clothes/[xuz'na:] we kept it, else b) stress the rightmost closed syllable (non-nal CVC or nal CVCC) e.g. [\aI'Iaml] I taught, [ma'saLlu] I caught him, else c) stress the rst syllable ['Lalab] he wrote. Crucially, c) holds in words contain- ing 4 light syllables such as ['Lalabalu] she wrote it, conrming that this is in essence a Last/First pattern (though a nal CVV attracts stress). 7. Dialects of the Syro-Mesopotamian, Anazi, and Shammari groups in John- stones (1967) classication. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 632) 632 Harry van der Hulst and Sam Hellmuth Nishio (1994) reports the same Classical style pattern in the dialect spoken in the Nile Valley village of Qift (near Qena). Hamid (1984: 37) also describes the stress patterns of Sudanese Arabic in terms of a Classical style Last/First pattern, which reduces to: a) stress the rightmost heavy syllable, else b) stress the initial syllable (cf. Kensto- wicz 1984: 129). However, no sample words are given that contain more than 3 light syllables, and there are a number of counterexamples to the initial stress rule, which Hamid explains by appeal to interac- tion with segmental phenomena, which might be amenable to a L(CNT) analysis. Western group (Maghrebi) Boudlal (2001: 107.) describes the stress patterns of the dialect of Moroccan Arabic spoken in Casablanca, for words in isolation, as follows: a) stress the nal syllable if it is heavy e.g. [II'mun] oranges, otherwise, b) stress the penult (regardless of weight) e.g. ['iomIa] sand. Only CVC syllables con- taining full vowels are heavy; CoC is treated as light. For isolation forms then, Moroccan Arabic displays a U/P system similar to that observed in Maltese; in connected speech Boudlal found stress was invariably word- nal. Guella (m.s.) describes the stress patterns of the dialect of Algerian Arabic spoken near Tlemcen as follows, barring some morphologically con- strained exceptions: a) stress the leftmost long vowel (e.g. [moj'du:dI:n] tied up m.pl.), otherwise b) stress the penultimate syllable (e.g. [loq'IcbLum] she will overturn you pl.). Talmoudi (1980) describes the stress patterns of the dialect of Tunisian Arabic spoken in the Old City of Sousa as follows: a) stress the leftmost syllable containing a long vowel, otherwise b) stress the initial syllable. Closed syllables, both CVC and CoC, are treated as light. It is not clear whether the patterns described hold for isolation forms or in connected speech. Abumdas (1985) gives an account of the dialects of Libyan Arabic spoken in Tripoli, Ben Ghazi and Zliten (a Bedouin-origin variety). He notes that there are stress minimal pairs (as also reported by Mitchell (1960) for Cyrenaican Bedouin Arabic), e.g. ['xaIaq] creating P[xa'Iaq] he created, but states that stress is nonetheless predictable in many cases, giving rules parallel to those put forward for Cyrenaican Bedouin Arabic (Eastern Libyan Arabic) by Mitchell (1960) and Owens (1984). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 633) Word accent systems in the Middle East 633 Generalizations The following table summarizes the accentual types that we have men- tioned in this chapter: (11) An overview of accentual types in Arabic Classical Arabic [L/F] Middle Eastern Central Western group Southern group Northern Group Central Group Gulf: [U/P] Beirut/Lebanese: [U-P/A] Radio Egyptian: [L(CNT)] P[U-P/A] Cyrenaican Bedouin (East. Libyan): [L(CNT)] Sanaani: [L/F] and [U-P/A] Damascene: [U-P/A] Cairene: Trochaic [L(CNT)] Libyan (Tripoli): [L(CNT)] Yemen Plateau: [L/F] South Levantine/Palestinian: Trochaic [L(CNT)] or [L/F] and [U-P/A] Egyptian Saiidi: [L/F] Moroccan: [U/P] Bedouin Hijazi: [U-P/A] Negev Bedouin: Iambic [L(CNT)] Sudanese: [L/F] Nigerian: [U-P/A] Saudi Arabic dialects: [U-P/A] Bani Hassan: [U-P/A/P] Chadic: [U-P/A] Urban Hijazi Arabic: Trochaic [L(CNT)] Baghdavi: [U-P/A] Tunisian: [F/F] North Sinai Bedouin: [L/F] Algerian [F/P] North Arabian dialects: [L/F] Maltese: [U/P] Because the Arabic languages show such an interesting variety of accen- tual systems, their proper analysis has been the subject of both descriptive and theoretical work. Here, following Hayes (1995), Kager (2009) and Watson (to appear), who oer detailed analyses, we will briey discuss the variety of systems and their possible relations, using some ingredients of the accentual theory proposed in van der Hulst (in prep.). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 634) 634 Harry van der Hulst and Sam Hellmuth The table in (12) summarizes the various accentual types: (12) A summary of accentual types in Arabic [L/F] Classical; Yemen Plateau; North Sinai Bedouin; North Arabian Dialects; Egyptian Saiidi; Sudanese [F/F] Tunesian [F/P] Algerian [U-P/A] Bedouin Hijazi; Saudi Arabic dialects; Beirut/Lebanese; Damascene; Baghdavi; Nigerian; Chadic [U-P/A/P] Bani Hassan [L/F] and [U-P/A] Sanaani [U/P] Moroccan; Gulf; Maltese [L(CNT) tr] EM(syll): Palestinian; /EM (syll): Radio Egyptian EM(syll): Cairene (~[U/P]) EM (syll): Urban Hijazi (no post-heavy strong light) [L(CNT) ia] Negev Bedouin; Cyrenaican Bedouin Extrametricality It seems that all Arabic languages treat nal CVXC as heavy and nal CVC as light. The status of nal CVV, which is rare or absent, is not clear, but apparently not uniform across the dialects. Final C is always invisible, but on top of that we need syllable EM. In nal position, if CVV is either absent, or present and invisible (like CVC and CV), we can simply say that syllable-EM applies. If nal CVV is stressed, we have to say that there is (a) nal C-extrasyllabicity (to cover nal CVC) and, additionally, nal LIGHT syllable EM (to cover nal CV). Final CVC, then, is (harmlessly) doubly invisible. Extrasyllabicity of C makes it light and as such extrametrical together with nal CV. However, Cairene does not have nal syllable EM, while it does ignore nal C, because nal CVC acts as light. In U/P languages we also do not have nal syllable EM, but still nal C is invisible, except in Moroccan where nal CVC is heavy. In this language CoC is light together with nal CV. To see a separate role for consonant-EM we can look at bisyllabic words. In most languages, in bisyllabic words, one would suppose that 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 635) Word accent systems in the Middle East 635 syllable-EM is suppressed to guarantee word-minimality. Most of these CVCVC (LH) words have initial accent, suggesting that the nal C does not contribute to weight, but in Negev Bedouin such words have nal accent, which suggest that there is no nal C-EM. This may be connected to the rise of so-called iambic patterns. Watson discusses the case of Sanaani Arabic in which the rightmost non-nal CVV or CVG (syllables closed by a geminate consonant) attract accent away from nal superheavy syllables and CVV syllables (and of course CVC syllables). This language then treats CVXC and CVV as monosyllabic and as such they fall under syllable-EM. Bounded systems All Arabic languages agree in certain patterns: (13) a. Accent is U if the U syllable is VXC b. Accent is P if the P syllable is heavy and the U syllable is light (A nal CVC counts as light) Systems start to dier once we look at nal sequences of light syllables. Here we see a rich variety. The standard Latin like pattern adds the following third clause to the two clauses in (13) (14) c. Accent is A if the two nal syllables are light This creates the [U-A/P] pattern of Damascene Arabic and many other varieties. However Cairene Arabic has P accent if a word ends in two light sylla- bles preceded by a heavy syllable or two (or rather an even number of) other light syllables (counting from the word beginning or the rightmost heavy syllable): (15) HLL ] HLLL ] XLLLL ] (X # or H) This is a count system. In van der Hulst (in prep.) a count system is treated as having two bounded accentual domains, one on the left (which in this case is right-headed if the right hand syllable is heavy, otherwise left-headed) and one on the right that is not headed: (16) x >> (ss)sssss(ss)*s An alternating trochaic pattern of rhythmic beats echoes rightward, away from the initial accent (shown in 17 as Perfect Gridding-trochaic (Left to Right)) and invades the right-hand domain as follows: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 636) 636 Harry van der Hulst and Sam Hellmuth (17) Cairene Arabic x x x ER(R) x x x x x x x PG-tr (LR) (HL)L ] H(LL)L ] XL(LL)L ] Note that a rhythmic beat is assigned to a post-heavy light syllable (since rhythm in this language is moraic, as shown in Hayes 1996). In the rst case the domain has two beats (one by weight and one by rhythm) and the rightmost wins, which is shown in (17) as End Rule (Right)). Palestinian Arabic is minimally dierent from Cairene: (18) Palestinian Arabic x x x ER(R) x x x x x x x PG-tr (LR) (HL)L ] H(LL)L ] XL(LL)L ] The accent is further to the left, which Hayes (1996) in his metrical account handles by imposing foot extrametricality). We suggest an alterna- tive. To account for the HLL] pattern we need to assume that the rhythmic beat on the penultimate light syllable is deleted; in other words: within the accentual domain a heavy syllable prevails over a rhythmically strong light syllable. 8 However, we also need to account for the pre-antepenultimate pattern in case a word ends in a string of four light syllables which is claimed to exist because speakers pronounce quadrisyllabic light-syllabled words with initial accent. To us, this looks like a Classical application of the Classical Arabic unbounded stress rule (specically its default clause which applies if all syllables are light). We therefore submit that Palestinian Arabic embodies a hybrid aspect: in case of a long nal string of light syllables it applies the unbounded initial default that is characteristic of Classical Arabic; this is indicated in table 11 by specifying that the system is both [L/F] and [U-P/A]). 9 Urban Hijazi Arabic is like Palestinian Arabic, although it misses its hybrid character; it is simply [U-P/A]: (19) Urban Hijazi x x x ER(L) x x x x x x x PG-tr (LR) (HL)L ] H(LL)L ] XL(LL)L ] 8. We cannot say that the ER is Left, because in case of (HH)L] stress is penultimate, in all Arabic dialects. 9. Radio Egypt Arabic vacillates between the Cairene and the Palestinian pattern. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 637) Word accent systems in the Middle East 637 There is further variation among the count systems. Negev Bedouin has iambic rhythm rather trochaic rhythm: (20) Negev Bedouin x x x ER(L) x x x x x x x PG-ia (LR) (HL)L ] H(LL)L ] XL(LL)L ] As in Cairene and Palestinian, the location of accent is not constant within a window that has two light syllables. This means that Negev is also a count system. However, the rhythmic pattern is iambic rather than trochaic. Finally we look at two non-count patterns: (21) Bani-Hassan 10 x x x ER(R) x x x x x x Def (R) (HL)L ] H(LL)L ] XL(LL)L ] This system diers from Damascene Arabic in that a domain with two light syllables is right headed (indicated by Default (Right)), whereas the default is Left in Damascene: (22) Damascene x x x ER(R) x x x x x Def (L) (HL)L ] H(LL)L ] XL(LL)L ] The dierences between these various systems are small, but real, although the diagnostic sequences may not be so easy to obtain, because of the fact that (nal) sequences of light syllables are rare. It is interesting to note that count systems come very close to being weight-sensitive right-edge systems which we realize when we compare the languages discussed above: (23) SH ] HL ] HLL ] HLLL ] XLLLL ] Damascene U P A A A Bani-Hassan Arabic U P A P P Negev U P A P A Cairene Arabic U P P A P Palestinian Arabic U P A A I / PA Urban Hijazi U P A A P 10. This pattern possibly also occurs in Riyadh Arabic as analyzed in Halle and Kenstowicz (1989). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 638) 638 Harry van der Hulst and Sam Hellmuth As for the rst three sequences, we only see a dierence in the HLL case which seems to reect a dierence in extrametricality (no in Cairene). But if we add the fourth sequence (HLLL]), syllable-EM (no) can no longer be correct for Cairene, and this is where we have to resort to the count analysis. For Palestinian we do not have to do that yet. We can treat Bani-Hassan and Negev alike in that both treat a domain with two light sylables as right-headed, whereas the others have a left-headed choice. Only by adding the fourth sequence of syllables (four nal light syllables) can we dierentiate all the systems, fully enforcing a count analysis of Negev, Urban Hijazi and Palestinian. The point of this exercise is to demonstrate that the crucial evidence for deciding the precise nature of the system (especially whether the system is a count system or not) lies in sequences of light syllables that are not frequent in the data. Unbounded systems There is a third type of system that occurs with some frequency in the Arabic languages and is exemplied by Classical Arabic which has an unbounded LAST/FIRST system with some form of extrametricality. In unbounded systems the domain of accent assignment comprises the whole word. If the word contains heavy syllables, one of these (in Classical Arabic the rightmost non-nal one) attracts accent; if there is no (non-nal) heavy syllable in the word, the default option is a light syllable at one of the edges (in Classical Arabic it is the rst). This LAST/FIRST pattern is reconstructed for Classical Arabic and as such it is controversial. An alternative interpretation of the comparative evidence is that accent never falls further leftward than the antepenultimate syllable (cf. Angoujard 1990 and Kager 2009 for discussion). However, as we have seen, Classical Arabic is not unique in having an unbounded system in the Arabic family. McCarthy (1979: 461) remarks that the Classical pattern is only preserved in a few modern dialects (Egyptian Saiidi, Yemen Plateau). Our survey here above suggests that it is also found in Bedouin-origin North Arabian dialects in Syria, Lebanon and Sinai, and possibly also in Sudanese Arabic. Among the unbounded systems we also see cases that dier from Classical Arabic in the choice of the leftmost heavy syllable to bear accent. Algerian Arabic has a FIRST/FIRST system (with the default clause as in Classical Arabic, i.e. rst), while Tunisian Arabic chooses the penultimate syllable as a default and hence has FIRST/PENULT [F/P] 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 639) Word accent systems in the Middle East 639 Historical change If the unbounded L/F pattern can indeed be attributed to Classical Arabic (or perhaps proto-Arabic), two kinds of systems have developed from this unbounded system. We nd count systems in which the primary accent is on the right side of the word, but computation starting on (i.e. counting from) the left side, and then we nd right-edge systems of the Latin type with penultimate or antepenultimate accent. McCarthy (1979) sees the count system type (of, for example, Cairene Arabic) as a later development than the Latin type (found in, for example, Damascene Arabic). In van der Hulst (1997), it has been suggested that the Cairene (count) system might be a transitional phase, forming the link between the initial (default) accent of the unbounded Classical Arabic and the bounded Latin-like accent rule of Damascene Arabic. Interpreted within the accentual theory in van der Hulst (in prep.), the transitions that have taken place can be seen as follows: (24) a. Unbounded domain (s-extrametricality) [(.............)<s>] L/F: ER(R), Def (L): Classical Arabic F/F: ER(L), Def (L): Tunesian F/P: ER(L), Def (R): Algerian b. Count system (s-extrametricality, except in Cairene which means that nal C-EM emerges) [(ss)ssss(ss)*<s>] L(CNT)-trochaic: Cairene, Palestinian L(CNT)-iambic: Negev Bedouin c. Right-edge bounded system (s-extrametricality) [.............(ss)<s>] U-A/P: ER(R), Def(L): Damascene Place accent on last heavy or rst light in the rightmost bisyllabic domain d. Right-edge bounded system (s-extrametricality, but C-EM except in Moroccan.) [.............(ss)] U/P: ER(R), Def(L): Gulf Place accent on last heavy or rst light in the rightmost bisyllabic domain Classical Arabic is unbounded. The count systems display a fracturing of the unbounded domain into two polar bounded domains, with the right- most domain being the strongest. It would seem that count systems really 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 640) 640 Harry van der Hulst and Sam Hellmuth are intermediate systems in that on the one hand the whole word domain (at least both edges) is relevant, while, on the other hand, accent is con- ned to a two-syllable window on the right-edge. This provides for the re-interpretation in terms of a bounded right-edge system, one variety of which (namely the one that displays antepenultimate accent) maintains syllable-extrametricality, while the other does not (Gulf, Moroccan and Maltese Arabic). Final accent in [LH] words in, for example, Negev Bedouin Arabic, points to a loss of consonant extrametricality in specic cases, which may be connected to the rise of so-called iambic patterns. Throughout, all sys- tems maintain weight-sensitivity. 4. Conclusions and Generalization Very little, if anything can be said about the isolate extinct languages mentioned in section 3. The pattern for Egyptian suggests a weight- sensitive right-edge system which ts with the general pattern found in the Semitic languages where accent is either unbounded or conned to the right edge: SEMITIC EAST: Last/First (UAkkadian) WEST: CENTRAL: ARAMAIC: Latin-type CANAANITE: Final (Hebrew), Latin type (Tiberian Hebrew) ARABIC: Last/First / L (count) / Latin type SOUTH Western (not discussed here) Eastern: Last/First Knudsen (1980) in his analysis of Akkadian accent concludes (p. 15): viewed in a historical perspective, most features in the Akkadian system of stress as outlined above are common to Akkadian, Hebrew, Aramaic, and Medieval Arabic. It is tempting to see the Last/First system as fundamental with, as suggested earlier, count systems and Latin-type systems having derived from that. The loss of weight-sensitivity and extrametricality as exemplied by Modern Hebrew seems a nal step in the direction of xed right-edge accent. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 641) Word accent systems in the Middle East 641 References Abumdas, A.H.A. 1985 Libyan Arabic Phonology. University of Michigan. Abu-Salim, I.M. 1987 Vowel Harmony in Palestinian Arabic: A Metrical Perspective. Journal of Linguistics, 23, 1: 124. Angoujard, Jean-Pierre 1990 Metrical structure of Arabic. Dordrecht: Foris. Bergstraer, G. 1983 Introduction to Semitic Languages. Winona Lake, Ind.: Eisen- brauns. (Translation by P.T. Daniels of the German original, einfu hrung in die semitischen Sprachen, Munich: Max Hueber, 1928; 2nd. ed. 1963. Berman, Ruth 1997 Modern Hebrew. In: Robert Hetzron 1997 (ed.), 312333. Birkeland, Harris 1954 Stress patterns in Arabic. Oslo: Jacob Dybwad. Bolozky, Shmuel 1997 Isreali Hebrew Phonology. In: Alan S. Kaye 1997 (ed.), 287 312. Borg, Albert and Marie Azzopardi-Alexander 1997 Maltese. London and New York: Routledge. Borg, Alexander 1997 Cypriot Arabic Phonology. In: Alan S. Kaye 1997 (ed.), 219 244. Borg, Alexander 1997 Maltese Phonology. In: Alan S. Kaye 1997 (ed.), 245286. Boudlal, A. 2001 Constraint interaction in the phonology and morphology of Casa- blanca Moroccan Arabic. Universite Mohammed V, Rabat, Morocco. Buccellati, Giorgio 1997 Akkadian and Amorite Phonology. In Alan S. Kaye 1997 (ed.), 338. Con, Edna A. & Shmuel Bolozky 2005 A Reference Grammar of Modern Hebrew. Cambridge: Cam- bridge University Press. Comrie, Bernard, Maria Polinsky, and Stephen Matthews 2003 The Atlas of Languages. Second Edition. London: Quarto Pub- lishing Inc. Creason, Stuart 2004 Aramaic. In: Roger D. Woodard 2004 (ed.), 391426. Daniels, Peter T. 1997 Classical Syriac Phonology. In: Alan S. Kaye 1997 (ed.), 127 140. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 642) 642 Harry van der Hulst and Sam Hellmuth de Jong, R. 2000 A Grammar of the Bedouin Dialects of the Northern Sinai Littoral Bridging the Linguistic Gap between the Eastern and Western Arab World. Leiden, Brill. Diem, W. 1973 Skizzen Jemenitischer Dialekte. Wiesbaden, Kommission bei Franz Steiner Verlag. Edzgard, Dietz Otto 2003 Sumerian Grammar. Leiden: Brill. Erwin, W. 1963 A short reference grammar of Iraqi Arabic. Washington, George- town University Press. Faber, Alice 1997 Genetic subgrouping of the Semitic languages. In: Robert Hetzron 1997 (ed.), 315. Fischer, Wolfdietrich 1997 Classical Arabic. In: Robert Hetzron 1997 (ed.), 187219. Gordon, Cyrus H. 1997a Ugaritic Phonology. In: Alan S. Kaye 1997 (ed.), 4954. Gordon, Cyrus H. 1997b Amorite and Eblaite. In: Robert Hetzron 1997 (ed.), 100115. Gordon, Cyrus H. 1997c Eblaite Phonology. In: Alan S. Kaye 1997 (ed.), 3948. Gordon, Giorgio 1997 Akkadian. In: Robert Hetzron 1997 (ed.), 6999. Graag, Gene 1997 Old South Arabian Phonology Phonology. In: Alan S. Kaye 1997 (ed.), 161168. Guella, N. ms. On syllabication, stress and intonation in an Urban Arabic dia- lect. King Saud University, Riyadh. Hackett, Jo Ann 2004 Phoenician and Punic. In: Roger D. Woodard 2004 (ed.), 365 385. Halle, M. and M. Kenstowicz 1989 On Cyclic and Non-Cyclic Stress. Ms. Dept. of Linguistics, M.I.T., Cambridge, MA. Hamid, A.H.M. 1984 A descriptive analysis of Sudanese colloquial Arabic phonology. University of Illinois. Hayes, B. 1995 Metrical stress theory: principles and case studies. Chicago, Uni- versity of Chicago Press. Hayes, John 1997 Sumerian phonology. In: Alan S. Kaye 1997 (ed.), 10011020. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 643) Word accent systems in the Middle East 643 Hetzron, Robert (ed.) 1997 The Semitic languages. London: Routledge. Hoberman, Robert D. 1997 Modern Aramaic phonology. In: Alan S. Kaye 1997 (ed.), 313 336. Huennergard, John and Christopher Woods 2004 Akkadian and Eblaite. In: Roger D. Woodard 2004 (ed.), 218 287. Hulst, H.G. van der 1977 Primary accent is non-metrical. Rivista di Linguistica 9/1: 99 127. Hulst, H.G. van der 2000 Issues in foot typology. In: M. Davenport & S.J. Hannahs (eds Issues in phonological structure, 95127. Amsterdam: John Benjamins Publishing Company. Irshied, O. & M. Kenstowicz 1984 Some phonological rules of Bani-Hassan Arabic: a Bedouin dialect. Studies in the Linguistic Sciences 14: 109148. Janssens, Gerard 1972 Stress in Arabic and word structure in the modern Arabic dialects. Leuven: Peeters. Jastrow, Otto 1997 The Neo-Aramaic languages. In: Robert Hetzron 1997 (ed.), 334377. Kager, Rene 2009 Stress. In: Kees Versteegh et al (eds.), The Encyclopedia of Arabic Language and Linguistics, vol. 4, 344353. Leiden: Brill. Kaufman, Stephen A. 1997 Aramaic. In: Robert Hetzron 1997 (ed.), 114130. Kaye, Alan S. 1997 Arabic Phonology. In: Alan S. Kaye 1997 (ed.), 187204. Kaye, Alan S. (ed.) 1997 Phonologies of Asia and Africa. Volume 1 and 2. Winona Lake: Eisenbrauns. Kaye, Alan S. and Judith Rosenhouse 1997 Arabic dialects and Maltese. In: Robert Hetzron 1997 (ed.), 263311. Kenstowicz, Michael 1983 Parametric variation and accent in the Arabic dialects. In Chukerman, A., M. Marks, & J.F. Richardson (eds.) Papers from the nineteenth regional meeting of the Chicago Linguistic Society. CLS 19, 205213. Chicago, Chicago Linguistic Society. Khalafallah, A.A. 1969 A Descriptive Grammar of Sa9i:di Egyptian Colloquial Arabic. The Hague, Mouton. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 644) 644 Harry van der Hulst and Sam Hellmuth Khan, Georey 1997 Tiberian Hebrew Phonology. In: Alan S. Kaye 1997 (ed.), 85 102. Knudsen, Ebbe Egede 1980 Stress in Akkadian. Journal of Cuneiform Studies 32/1: 316. Lonnet, Antoine and Marie-Claude Simeone-Senelle 1997 La phonologie des langues subarabiques modernes. In: Alan S. Kaye 1997 (ed.), 337372. Loprieno, Antonio 1995 Ancient Egyptian: A linguistic introduction. Cambridge: Cam- bridge University Press. Loprieno, Antonio 1997 Egyptian and Coptic phonology. In: Alan S. Kaye 1997 (ed.), 431460. Lorpieno, Antonio 2004 Ancient Egyptian and Coptic. In: Roger D. Woodard 2004 (ed.), 160217. Malone, Joseph L. 1997 Modern and Classical Mandaic Phonology. In: Alan S. Kaye 1997 (ed.), 141160. McCarthy, John 1979 On stress and syllabication. Linguistic Inquiry 10; 443466. McCarthy, R.J. & F. Raouli 1964 Spoken Arabic of Baghdad. Beirut, Librarie Orientale. Michalowski, Piotr 2004 Sumerian. In: Roger D. Woodard 2004 (ed.), 1959. Mitchell, T.F. 1960 Prominence and syllabication in Arabic. Bulletin of the School of Oriental and African Studies 23: xxx269. Nebes, Norbert and Peter Stein 2004 Ancient South Arabian. In: Roger D. Woodard 2004 (ed.), 454 487. Nishio, T. 1994 The Arabic dialect of Qift (Upper Egypt Tokyo, Institute for the Study of the Languages and Cultures of Asia and Africa. Owens, J. 1984 A short reference grammar of Eastern Libyan Arabic. Wiesbaden, Harrassowitz. Owens, Jonathan 2006 A linguistic history of Arabic. Oxford: Oxford University Press. Pardee, Dennis 1997 Ugaritic. In: Robert Hetzron 1997 (ed.), 131144. Pardee, Dennis 2004 Ugaritic. In: Roger D. Woodard 2004 (ed.), 288318. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 645) Word accent systems in the Middle East 645 Pickett, I. 2006 Some aspects of dialect variation among nomads in Syria and Lebanon. SOAS. Prochazka, T. 1988 Saudi Arabian dialects. London, Keegan Paul International. Rendsburg, Gary A. 1997 Ancient Hebrew Phonology. In: Alan S. Kaye 1997 (ed.), 6584. Ruhlen, Merritt 1991 A guide to the worlds languages. Volume 1: Classication. Stan- ford, CA: Stanford University Press. Segert, Stanislav 1997 Old Aramaic Phonology. In: Alan S. Kaye 1997 (ed.), 115126. Segert, Stanislav 1997a Phoenician and Punic Phonology. In: Alan S. Kaye 1997 (ed.), 5564. Segert, Stanislav 1997b Phoenician and the Eastern Cannanite languages. In: Robert Hetzron 1997 (ed.), 174186. Simeone-Senelle, Marie-Claude 1997 The Modern South Arabian languages. In: Robert Hetzron 1997 (ed.), 378423. Steiner, Richard C. 1997 Ancient Hebrew. In: Robert Hetzron 1997 (ed.), 145173. Stolper. Matthews W. 2004 Elamite. In: Roger D. Woodard 2004 (ed.), 6094. Versteegh, Kees et al (eds.) 20069 The Encyclopedia of Arabic Language and Linguistics. Leiden: Brill. Watson, J.C.E. 2002 The phonology and morphology of Arabic. Oxford, OUP. Watson to appear Arabic word stress. In: Colin Ewen, Marc van Oostendorp and Keren Rice (eds.) Phonological Compendium. Blackwell. Wilhem, Gernot 2004a Hurrian. Roger D. Woodard 2004 (ed.), 95118. Wilhem, Gernot 2004a Urartian. Roger D. Woodard 2004 (ed.), 119137. Woodard, Roger D. (ed.) 2004 The Cambridge Encyclopedia of the Worlds Ancient Languages. Cambridge: Cambridge University Press. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 (V9 27/8/10 15:54) WDG (155mm230mm) TimesNRMT 1201 Goedemans pp. 615646 1201 Goedemans_11_Ch11 (p. 646) 646 Harry van der Hulst and Sam Hellmuth