This paper particularly brings Epstein, Kitahara, and Seely’s (EKS 2014) suggestion into critical consideration and tries to make an alternative to their suggestion. What is interesting in EKS’s (2014) proposal is that they argue for the unnecessity of such concepts as numeration and lexical array, and even for the elimination of phases. Though it has been taken for granted that computational complexities are resolved by transferring some stages of a derivation as chunks to the interfaces and then continuing the derivation as such, EKS (2014) apparently prove that we can use the labeling algorithm to resolve the computational complexities in the way that some chunks are constructed even without phases. In this paper, we will see that the concepts of lexical array and numeration are necessary, despite the plausibility of EKS’s suggestion. Furthermore, it will be shown that with the independent necessity of those concepts, we still need the concept of phase as the basic unit of derivation and transfer.

1. Introduction
 2. LA, MoM, and the Elimination of Phase
 3. On the Necessity of Lexical Array and Phase
  3.1. D-Structure Representation vs. Component
  3.2. Multiple Spell-out and Lexical Insertion
  3.3. Chains and Phrases
  3.4. Lexical Array as a Pool of Syntactic Objects
  3.5. On the Label φ
 4. Conclusion
 References

• Yong-Ha Kim(Andong National University)