Camera Model Performance 63

(41a) and (41b) above are unary and ternary branching respectively and they violate the Extension Condition (EC) which stipulates that applications of merge can only target root syntactic objects. However, (41b) could be configured as (50) below:

Endocentricity suggests that when merge combines two elements (e.g. X and Y) as demonstrated above, the output, which is often a phrase is free to take its syntactic category label from either X or Y such that the new SO (Phrase) could either be XP or YP respectively. Lastly, the relationship of merge with word order points to the fact that lexical items on the numeration are not combined anyhow; for a derivation to converge, certain structural building rules and principles (e.g. linearity) which are relevant for a successful merger operation must be observed. In summary, merge as a source independent operation, satisfies the selectional properties of LIs and captures the recursion property within MP.

demonstrated thus: the V kidnapped and the DP (PRN) THEY³⁶ were merged to form VP kidnapped THEY. The structure is further merged with the tense auxiliary BE³⁷ to derive the structure below:

In the structure, the T-Aux BE is the higher element as well as the probe and it searches for a suitable goal in its c-command domain and then locates the PRN as the only potential goal. Consequently, its unvalued [Pers and Num] ϕ-features are valued by the goal. Conversely, the unvalued (Case) ϕ-feature on the goal is valued by the probe. This process is called Feature valuation. Following these observations, Radford (2009:301) supposes that agreement is characterised as involving two feature-valuation sub-operations sketched thus:

(52) Agreement

When a probe (like T) agrees with a goal in its local domain

(i) the unvalued (person/number) ϕ-features on the probe will be valued (i.e. assigned a value which is a copy of that on the goal)

(ii) the unvalued case feature on the goal will be valued (i.e. assigned a value dependent on the nature of the probe e.g. nominative if the probe is a finite T).

In relation to this, MP assumes that LIs (functional heads inclusive) enter the derivation with their features already specified and the generative procedure (GP) determines

36 In the literature, it is believed that noun and pronoun expressions enter the derivation with their person and number ϕ-features already valued and their case ϕ-feature unvalued. Hence, THEY is written in capital letter to indicate that its case feature is yet to be determined.

37 Conversely, finite T auxiliaries enter derivations with valued T feature and unvalued person and number features.

DP THEY [3^rd Pers]

[PL-Num]

[u-Case]

VP V

kidnapped T^I

T BE [PST-Tns]

[u-Pers]

[u-Num]

(51)

(Radford 2009: 284)

whether a given expression X is licit in a given derivation by checking the features of X against the features of an appropriate head. The program also assumes that LIs bear three sets of grammatical features-head, specifier and complement features. Head feature accounts for the grammatical properties of LI that marks it as distinct from others; complement feature determines the type of complement LI would select; while specifier features determine the kind of specifier a syntactic head takes.

Based on these observations, linguists posit that crosslinguistic word order variations are consequent upon the strength of features of the functional categories and the process of feature valuation. Functors are morphological due to their inflectional nature. They can bear strong or weak features, weak features are not visible at PF, and hence, their presence does not cause a derivation to crash. In contrast, strong features are interpretable at PF. There visibility can cause a derivation to crash because its presence violates the principle of full interpretation (FI). Therefore, they need to trigger the overt movement of the item which they enter into a checking relationship with, for valuation and elimination. Categorial and ϕ-features are +interpretable at LF and are assumed to have semantic import in the interpretation of SDs. On the other hand, specifier and complement features are weak and –interpretable. Generally, MP assumes that the computational system cross checks the features of LIs via feature checking operation in order to ensure compatibility of features borne by LIs.

There are two main checking configurations associated with Minimalism: Spec-Head and Head-Head relations. The former is based on the specifier feature of a head which attracts the feature of another syntactic object from its original theta-marked position into Spec-Head for feature checking purpose. This follows from the assumption that every type of structural case is checked in a Spec-Head configuration. For instance, the subject of a clause is said to be attracted from its VP-internal θ-position to Spec/I or T where it values the Spec-feature of I or T and consequently values the nominative case-feature of the moved subject. Hornstein, Nunes and Grohmann (2005: 116-121) point out that MP assumes a unified Spec-Head approach to case where the object is not expected to check accusative case in its base position but moves to some spec-position.

Better still, the assumption follows that accusative case checking could be covert or overt object movement to the case checking position because covert movement is allowed in the minimalist syntax. Similar approach is assumed for oblique case which

is said to be checked under a spec-head configuration rather than the head-complement approach. The unified analysis suggests that DPs check their case feature in a position higher than the one where they establish a probe-goal relationship with some head often overtly. This assumption differs from the GB approach where the case and θ-positions are one and the same. For instance, if an object values its case covertly, only the case feature moves while the object itself does not move to the [Spec, VP].

Conversely, Head-Head checking configuration does not involve specifiers and complements. In this case, a lexical or functional head moves to another head position because the new position bears an unvalued feature which attracts the unvalued feature of the earlier position. This involves a type of movement called head to head movement, where a head X of a phrase XP moves from its position, into another head Y of a phrase YP in a higher position (see also Radford, 2009: 155). This kind of movement is constrained by a locality principle of UG called Head Movement Constraint (HMC) stated by Radford (2009: 157) as follows: “Head Movement is only possible between a given head and the head of its complement”. In English, V(erb) to T(ense) movement and T to C(omplementiser) movement are instances of head movement. These checking configurations account for the agreement phenomena observed in languages. Therefore, syntactic derivations depend on feature checking and valuation.