This paper discusses the lowering of the high vowels /i,u/ to [e,o] in final syllables in palestinian arabic. such a shift is actually well documented in the literature with regard to final unstressed cvc syllables. yet whereas previous studies have not stated a clear prosodic condition on the shift, we bring evidence that in the galilean dialects discussed the shift is clearly conditioned and occurs in the pre-pausal position only. in addition, we show that the same shift occurs in two other positions, viz. final cv and cvcici, and that it is not merely qualitative but quantitative as well, so that lowering of /i,u/ to [e,o] often involves certain lengthening, yielding [eː,oː] respectively. in the last section of this article we provide a uniform account for vowel lowering and lengthening in the three types of syllable mentioned. we show that in the pre-pausal environment an additional templatic position is added to the last vowel; and since this position cannot be realized through vowel lengthening, a lowering agent is inserted instead. 1. Vowel Lowering and Vowel Lengthening in Levantine Arabic many syro-palestinian dialects exhibit restrictions on the occurrence of /i,u/, which regularly alternate with [e,o] in final cvc position (garbell 1958: 315; grotzfeld 1965: 12; levin 1994: 34; watson 2011: 3010). such cvc syllables are never stressed, in accordance with the stress patterns of most syro-palestinian dialects (fischer and Jastrow 1980: 182), in which the last, non-final vowel of the word is only stressed if in a heavy syllable (i.e. cvc or cvcc). in this study we checked the lowering environment cvc in context and pre-pausally among speakers of three galilean dialects, 1 at the time of the writing of this paper, the second author was also affiliated with the hebrew university of Jerusalem. 99781_JOSS_62-2_2017_05_Shachmon Faust.indd 371 7/08/17 12:41 Downloaded from https://academic.oup.com/jss/article-abstract/62/2/371/4364791/The-Spread-of-Prepausal-Lowering-in-Galilean by Central Lib Agricultural Science user on 07 October 2017 the spread of prepausal lowering in galilean arabic 372 namely those of nazareth, Kafr Qaraʽ and sakhnin.2 The data presented below were all elicited and measured by us. according to our findings, apart from the lowering of /i,u/ to [e,o] in final cvc syllables, vowels in this position are also clearly longer than their nonlow counterparts and other short vowels, whether stressed or not, by around 40ms. They are nevertheless far inferior in length to lexically long vowels, which can go up to 200ms. in what follows we use the symbol [ː] to represent this half-length of [e,o] in final cvc position. in this position [e,o] are in complementary distribution with [i,u], the former exhibit additional — though restricted — length. (1) a. [kútoːb] ‘books’ c. [míse:k] ‘he held’ [kutúbhoːm] ‘their books’ [misíkhoːm] ‘he held them’ b. [údroːs] ‘study!’ d. [íkteːb] ‘write!’ [udrúshoːm] ‘study them!’ [iktíbliː] ‘write to me’ it is worth mentioning that gutturals may also induce vowel lowering, as in the examples /xliʼna/>[xleʼna] ‘we were born’; /iḥna/>[eḥna] ‘we’; or /biḥku/>[beḥku] ‘they speak’. however, such lowering depends purely on the consonantal environment and is not a positional effect. such cases of environmental lowering are not marked in our text samples, and we therefore transcribed /šawāriʽ/ and /daʼāyiʼ/ even when realized [šawāreʽ] and [daʼāyeʼ]. 2. The Prosodic Prominence of the Final Position it is by no means surprising to find lengthened vowels in a final, preconsonantal position. The final position is one of prosodic prominence in many languages, and final closed syllables very often do not behave like non-final ones, where one would not expect length. we submit that lowering in this position is related to the extra length typical of (pre-pausal) final closed vowels (hardcastle et al. 2013: §2.2.2). Final (pre-consonantal) position > lengthening + lowering 2 some of the data for this paper were collected with the support of a grant from the israel science foundation (grant 1192/14). 99781_JOSS_62-2_2017_05_Shachmon Faust.indd 372 7/08/17 12:41 Downloaded from https://academic.oup.com/jss/article-abstract/62/2/371/4364791/The-Spread-of-Prepausal-Lowering-in-Galilean by Central Lib Agricultural Science user on 07 October 2017 the spread of prepausal lowering in galilean arabic 373 according to the autosegmental framework of government phonology (Kaye et al. 1990), length can be achieved through spreading, i.e. the propagation of a segment into a following slot. lowenstamm (1996) proposes that such spreading is only possible under ‘licensing’ from a vowel in the following syllable. for instance, in Qaraqosh neo-aramaic (Khan 2002), vowels are long in open syllables, whether stressed or not, e.g. [gdaːríːlɪ] ‘he is putting it’. This is because they are ‘licensed’ by the following vowel. in contrast, vowels in medial closed syllables, even stressed ones, are short, e.g. [ʼúxlu] ‘eat! (pl)’ because ‘licensing’ cannot hold above a consonant cluster. vowels in unstressed final closed syllables [yíːxul] ‘eat! (sg)’ are also short, because there is no following vowel to license their length. in the galilean dialects which we have checked we found that the last vowel before the pause tends to be lengthened. in order to realize this additional length, the vowel must spread. however, just as in Qaraqosh neo-aramaic, in the final cvc position there is no following vowel to license spreading. we submit that in such cases, the lowering agent /a/ of element Theory (Kaye et al. 1985, backley 2011) is inserted as a repair strategy to realize the additional length.3 The combination of the underlying high vowel /i/ or /u/, the additional length and the inserted /a/ agent yields the mid [e:] or [o:] respectively. in other words the insertion of the /a/ agent is an alternative to length through spreading.4

中文翻译：

---

伽利略阿拉伯语中的经前降低的传播

本文讨论了巴勒斯坦阿拉伯语最后音节中高元音 /i,u/ 到 [e,o] 的降低。这种转变实际上在关于最终非重读 cvc 音节的文献中有详细记载。然而，虽然之前的研究没有明确说明这种转变的韵律条件，但我们提供的证据表明，在讨论的伽利略方言中，这种转变显然是有条件的，并且只发生在停顿前的位置。此外，我们表明在其他两个位置也发生了相同的转变，即。final cv 和 cvcici，而且它不仅是定性的，而且也是定量的，因此将 /i,u/ 降低到 [e,o] 往往涉及一定的延长，分别产生 [eː,oː]。在本文的最后一节中，我们对提到的三种音节中元音的降低和延长提供了统一的解释。我们表明，在暂停前的环境中，一个额外的模板位置被添加到最后一个元音；由于这个位置不能通过元音延长来实现，所以插入了降低剂。1. 黎凡特阿拉伯语中的元音降低和元音延长 许多叙利亚-巴勒斯坦方言对 /i,u/ 的出现有限制，这些 /i,u/ 在最后的 cvc 位置经常与 [e,o] 交替 (garbell 1958: 315; grotzfeld 1965: 12 ；莱文 1994：34；沃森 2011：3010）。根据大多数叙利亚-巴勒斯坦方言的重音模式 (fischer and Jastrow 1980: 182)，此类 cvc 音节从不重读，其中单词的最后一个非最终元音仅在重音节中重读（即cvc 或 cvcc）。在这项研究中，我们在三种伽利略方言的发言者之间检查了上下文中的降低环境 cvc，1 在撰写本文时，第二作者也隶属于耶路撒冷希伯来大学。99781_JOSS_62-2_2017_05_Sachmon Faust.indd 371 7/08/17 12:41 下载自 https://academic.oup.com/jss/article-abstract/62/2/371/4364791/The-Spread-of-Loweringalpaus -in-Galilean 由 Central Lib 农业科学用户于 2017 年 10 月 7 日在伽利略阿拉伯语 372 中传播，即 nazareth、Kafr Qaraʽ 和 sakhnin。2 下面显示的数据均由我们引出和测量。根据我们的发现，除了在最后的 cvc 音节中 /i,u/ 降低到 [e,o] 之外，这个位置的元音也明显长于它们的非低元音和其他短元音，无论是否有压力，大约 40 毫秒。尽管如此，它们的长度远不如词汇长元音，最长可达 200 毫秒。在接下来的内容中，我们使用符号 [ː] 来表示 [e,o] 在最终 cvc 位置的半长。在这个位置 [e,o] 与 [i,u] 呈互补分布，前者表现出额外的——尽管受到限制——长度。(1) 一个。[kútoːb] '书' c．[míse:k] '他拿着' [kutúbhoːm] '他们的书' [misíkhoːm] '他拿着他们' b．[údroːs] '学习！' d. [íkteːb]“写！” [udrúshoːm]“研究他们！” [iktíbliː] '写信给我' 值得一提的是喉音也可能导致元音降低，例如 /xliʼna/>[xleʼna] '我们出生'；/iḥna/>[eḥna] '我们'; 或 /biḥku/>[beḥku] '他们说话'。然而，这种降低完全取决于辅音环境，而不是位置效应。我们的文本样本中没有标记这种环境降低的情况，因此我们转录 /šawāriʽ/ 和 /daʼāyiʼ/，即使在实现 [šawāreʽ] 和 [daʼāyeʼ] 时也是如此。2. 韵母在韵母位置上的韵律突出 在韵母前的韵母位置上发现加长的元音并不奇怪。在许多语言中，结尾位置是韵律突出的位置之一，结尾的闭音节通常不像非结尾的音节那样表现，人们不会期望长度。我们认为这个位置的降低与（停顿前）最终闭元音的典型额外长度有关（hardcastle et al. 2013: §2.2.2）。最终（前辅音）位置 > 延长 + 降低 2 本文的一些数据是在以色列科学基金会的资助下收集的（资助 1192/14）。99781_JOSS_62-2_2017_05_Sachmon Faust.indd 372 7/08/17 12:41 下载自 https://academic.oup.com/jss/article-abstract/62/2/371/4364791/The-Spread-of-Loweringal -in-Galilean 由 Central Lib 农业科学用户于 2017 年 10 月 7 日根据政府语音学的自动分段框架（Kaye et al. 1990）在伽利略阿拉伯语 373 中传播前降低，长度可以通过传播来实现，即传播一个段进入下一个槽。Lowenstamm (1996) 提出这种传播只有在以下音节中的元音“许可”下才有可能。例如，在 Qaraqosh 新阿拉姆语 (Khan 2002) 中，元音在开音节中很长，无论是否重读，例如 [gdaːríːlɪ] 'he is put it'。这是因为它们由以下元音“许可”。相比之下，中间闭音节中的元音，即使是重读音节，也很短，例如 [ʼúxlu] 'eat! (pl)' 因为 'licensing' 不能超过辅音群。未重读的最后闭音节中的元音 [yíːxul] '吃！(sg)' 也很短，因为没有后面的元音来许可它们的长度。在我们检查过的伽利略方言中，我们发现停顿前的最后一个元音趋于加长。为了实现这个额外的长度，元音必须扩展。然而，就像在 Qaraqosh 新阿拉姆语中一样，在最后的 cvc 位置上没有跟随元音来许可扩展。我们认为，在这种情况下，元素理论（Kaye et al. 1985, backley 2011）的降低剂 /a/ 被插入作为实现额外长度的修复策略。 3 基础高元音 /i/ 的组合或/u/, 额外的长度和插入的 /a/ 代理分别产生中间 [e:] 或 [o:]。换句话说，/a/ 代理的插入是通过传播的长度的替代。 4