I argue that verb movement should be derived in nar row syntax. The derivational system of Labeling Algorithm proposed by Chomsky (2013, 2015) allows movement t o occur in narrow syntax without recourse to any movement-triggering features, contra Chomsky (2001, 2004, 2008). In this system, (both external and internal) Merge is free, even if a new semantic effect does not occur on a raised category in its raised p osition, contra Chomsky (2001). Thus, verb movement, which does not change a semantic interpre tation in the unmarked case, should be able to occur in narrow syntax without recourse to any driving features. As we see in the Icelandic data, it is actually difficult to claim t hat verb movement occurs in phonology, contra Chomsky (2001). I propose a way to carry out verb m ovement in narrow syntax in terms of feature inheritance and copy deletion. In the Romance languages as well as languages suc h a English, both φ-features and tense are inherited by T, to which R( =V)+v* moves, whereas in the V2 languages, only φ-features are inherited by T with tense remaining i n C, to which R+v* directly moves. In all of those languages, it is a m tter of morphophonology which copy of a verbal head, the one in a higher position or the on in the v* head position, is chosen to be pronounced. I also claim that do is inserted in English to phonetically embody feat ures such as [Q], [Foc(us)] and [Pol(arity)]. The feature that t he inserted do embodies is adjacent to tense and φ-features in each of the relevant constructions, wh ich accounts for why the inserted do carries tense and agreement as pointed out by the t raditional literature. ∗ Special thanks to Anders Holmberg for giving me ma ny insightful comments and suggestions and much adv ice, which helped me improve this paper in many points. Thanks also to Christer Platzack and Johan Brandtle r for giving me many helpful comments for this work and t o Halldór Ármann Sigurðsson for giving me Icelandic ata. Many thanks to two anonymous reviewers for their ve ry insightful comments, which enabled me to revise a former version of this paper significantly. Part of this p aper was presented at Keio University Linguistics Co lloquium, 30 January 2016. I would like to thank Hisatsugu Kitah ara and the audience there for their many comments for his work. I take all responsibilities for any errors.

中文翻译：

---

狭义句法中的动词移动

我认为动词运动应该从狭义的句法中推导出来。Chomsky (2013, 2015) 提出的标签算法的派生系统允许运动以狭窄的语法发生，而无需求助于任何运动触发特征，这与 Chomsky (2001, 2004, 2008) 相反。在这个系统中，（外部和内部）合并是免费的，即使新的语义效果没有出现在其凸起位置的凸起类别上，与 Chomsky (2001) 相反。因此，在未标记的情况下不会改变语义解释的动词移动应该能够以狭窄的语法发生，而无需求助于任何驱动特征。正如我们在冰岛语数据中看到的那样，实际上很难断言动词运动发生在音韵学中，这与 Chomsky (2001) 相反。我提出了一种在特征继承和副本删除方面以窄语法进行动词移动的方法。在罗曼语和英语等语言中，φ-特征和时态都被 T 继承，R(=V)+v* 移动到 T，而在 V2 语言中，只有 φ-特征被 T 继承时态保留在 C 中，R+v* 直接移动到。在所有这些语言中，选择发音词首词的副本，即处于较高位置的词首词或处于 v* 词首词首词位置的词，是形态音韵学的问题。我还声称 do 在英语中插入以在语音上体现诸如 [Q]、[Foc(us)] 和 [Pol(arity)] 之类的特征。插入 do 所体现的特征与每个相关结构中的时态和 φ 特征相邻，这解释了为什么插入的 do 带有传统文献所指出的时态和一致性。∗ 特别感谢 Anders Holmberg 给了我很多有见地的评论和建议以及很多建议，这帮助我在很多方面改进了这篇论文。还要感谢 Christer Platzack 和 Johan Brandtle r 为我的工作提供了许多有益的评论，感谢 Halldór Ármann Sigurðsson 给我冰岛语 ata。非常感谢两位匿名审稿人非常有见地的评论，这使我能够对本文的旧版本进行重大修改。这篇论文的一部分于 2016 年 1 月 30 日在庆应义塾大学语言学研讨会上发表。我要感谢 Hisatsugu Kitah ara 和那里的听众对他的工作提出的许多意见。我对任何错误承担全部责任。