Abstract Recent analyses of sluicing focus on the underlying structure of the sluiced clause, i.e. sluicing as deriving from full-fledged wh-questions, or from reduced clefts (Ross 1969, Guess who? In Robert I. Binnick, Alice Davison, Georgia M. Green & Jerry L. Morgan (eds.), Proceedings of the Fifth Regional Meeting of the Chicago Linguistic Society, 252–286. Chicago, IL: Chicago Linguistic Society, University of Chicago; Merchant 2001, The syntax of silence. Oxford, NY: Oxford University Press; Craenenbroeck, Jeroen van. 2010b. The syntax of ellipsis: Evidence from Dutch dialects. Oxford, NY: Oxford University Press., inter alia). In this paper, we investigate two sluicing strategies in Spoken Tamil, namely case-marked (CM) and non-case-marked (NCM) sluicing. In addition to the morphological distinction with respect to the presence/absence of grammatical case on the wh-sluice, we argue that the two types of sluicing differ in the configuration of the underlying embedded CP. For CM sluicing, the sluiced clause is derived from a full-fledged interrogative CP at the underlying level, whereas the bare wh-sluice undergoes leftward wh-scrambling to the CP-initial position followed by TP-domain deletion at PF. While we contend that most A/A’-diagnostics are uninformative of the type of operation wh-scrambling in Tamil involves (contra Sarma 2003, Non-Canonical word order: Topic and focus in adult and child tamil. In Karimi Simin (eds.), Word order and scrambling, 238–272. Malden, Oxford: Blackwell), various properties of the CM wh-sluice (e.g. scope, negation, adverb placement, multiple sluicing) can still be described by postulating that the wh-sluice involves A’-scrambling. For the second type of sluicing (NCM sluicing), the sluiced clause involves a biclausal structure formed by a normal sentence and a null copular question. We claim that the NCM wh-sluice is derived from Spad (Sluicing Plus A Demonstrative), since the null copular question can be accompanied by a demonstrative, cf. English ‘John met someone, who is that?’ and Dutch spading (Van Craenenbroeck 2010b). Spad is not derived from a full-fledged interrogative CP, and therefore its wh-sluice does not involve any scrambling operation. The present analysis of Tamil sluicing refutes the claim that reduced clefts are one underlying sluicing source in Dravidian languages, and moreover invites an inquiry of whether Dravidian as a language family in the historical sense always receives a homogeneous analysis, given the immense parametric variation among branch languages. In the same vein, we contend that any claim about the ‘principles’ of Dravidian syntax must be supported by strong cross-linguistic evidence at the microscopic level.

中文翻译：

---

泰米尔语中两种类型的水闸的语法

摘要最近对闸门的分析主要集中在闸门子句的基础结构上，即闸门是从成熟的wh问题或减少的裂缝中引出的（Ross 1969，Guess who？in Robert I.Binnick，Alice Davison，Georgia M. Green＆Jerry L. Morgan（ed。），《芝加哥语言学会第五次区域会议记录》，252-286。伊利诺伊州芝加哥：芝加哥大学芝加哥语言学会；商人2001，沉默的语法，牛津，纽约：牛津大学出版社； Craenenbroeck，Jeroen van。2010b。省略号的句法：荷兰方言的证据，纽约牛津：牛津大学出版社，等等。在本文中，我们研究了口语泰米尔语中的两种泄密策略，即案例标记（CM）和非案例标记（NCM）泄密。除了关于wh水闸上是否存在语法大小写的形态学区别外，我们认为这两种水闸的类型在底层嵌入式CP的配置方面有所不同。对于CM闸，闸子句是从底层的完整询问式CP派生而来的，而裸wh闸则向左扰流到CP初始位置，然后在PF处删除TP域。虽然我们认为，大多数A / A'诊断都无法说明泰米尔语所涉及的操作类型（contra Sarma 2003，非规范单词顺序：成人和儿童泰米尔语的主题和重点.Karimi Simin（ed。 ），字序和加扰（238-272）。Malden，牛津：Blackwell），CM Wh水闸的各种属性（例如，作用域，否定，副词位置，仍然可以通过假设wh-水闸涉及A'加扰来描述多次水闸）。对于第二种类型的闸门（NCM闸门），闸门从句涉及由正常句子和无效的居题组成的双月形结构。我们声称NCM闸闸是从Spad（闸门加A指示词）派生而来的，因为空问题可以伴随一个指示词，参见。英语“约翰遇到了一个人，那是谁？” 和荷兰人的吐槽（Van Craenenbroeck 2010b）。Spad并非来自完整的疑问句CP，因此它的水闸不涉及任何加扰操作。目前对泰米尔语水闸的分析驳斥了所谓的裂隙是德拉维语中潜在的水闸源之一的说法，此外，考虑到分支语言之间的巨大参数差异，还询问了Dravidian作为一种语言家族在历史意义上是否始终受到同质分析。同样，我们认为，任何有关Dravidian语法“原理”的主张都必须得到微观层面强有力的跨语言证据的支持。