5.7 The theoretical phonological analysis

The reason why there must be a move away from a universal hierarchical initial organisation is because current approaches to phonology and syntax fail when attempting to unify initially in one universal initial tree, two apparently differently organised linguistic constructions. As we have shown the final order in a sequence is different in English, and in Spanish, yet these languages although reaching different sequencing, use the same combinatorial, integrative and stabilising mechanisms.

The nonperipheral Consonant in the Svarabhakti construction, the flap, behaves as a Neither- Onset- nor -Coda as discussed in Garcia-Bellido 1999, 2000, 2003. Ambisyllabicity fails to capture the fact that middle Consonants are subject to deletion under identity with the preceding or following Vocoid. According to non-linear representations as (37), middle consonants, which are multilinked, cannot be delinked by a phonological process (Shein and Steriade1986). Those middle consonants which can undergo processes are left unrepresented and therefore unexplained by the autosegmental theory.

On the other hand a svarabhakti is a vocoid but does not function like a "Nucleus", yet it is integrated to produce meaning differences, (9i) vs (9vi), and establishes mirror relations in the same construction. The syllable representation cannot capture this simple fact because first, it has only two types of categories: Vowels and Glides, and secondly the template, which is an all-purpose-already-made-structure, can accommodate neither an interconsonantal vocoid which is neither a Vowel nor a Glide, nor it can accommodate an intervocoidal consonantoid which is neither an Onset nor a Coda..

(37a)	Ambisyllabicity must have two nuclei (Khan 1979)	(37b)

Conclusion: autosegmental phonology can integrate neither middle consonants nor middle vocoids using its universal syllable representation.

A Minimalist approach following minimalist principles also fails.

(38)	Time Phrase recursion and Local Movement (García-Bellido 1999)

I will give first the good results for this theory:

• Obstruent moves from CP to check in Spec CP where the Rhotic is its Sonorant compatibility (locality check).
• Rizzi's Relativised Minimality Goldsmith's "lines cannot cross": the vowel cannot move across the glide to V2 since it would go across G2 which is phonetically filled.

However there are problems produced by the theory:

• There is no explanation for why the moved Vocoid shoud move at all and leave a Glide copy [j].
• There is no explanation for why the trace-Vocoid should leave a bigger trace than the moved element. The big copy is c-commanded by the small one.
• There is no explanation for why the Vocoid leaves copy but not the consonantoid.
• There is no explanation for why the Obstruent should move out from CP to the top since it cannot check anything there.
• The Glide has a non identity relation in Spanish which is now NON local since there is a trace in between.

(39)	Leurbost

Problems:

• The moved Vocoid is smaller than its copy trace. The moved small copy is c-commanded by the big copy but movement is operating from top to bottom violating the main principle of trace theory.
• The movement of the C goes from top to bottom, violating the principle that traces must be c-commanded by their antecedent.

A recursive syllable approach using Levin's hierarchical structure fails as well.

(40)	The recursive syllable approach (Smith 1999)

Problem:

• There is no way to stop recursion of an Onset N''-N' or a Coda N-N''.

Índice general  I  Siguiente