You are on page 1of 5
Matching Theory “This tandout has been adapted from Andrew Radford (2004), Minimalist Ste, chapters 4 &8 (CUP) ‘The concept of “feature” was introduced in the second chapter of Andrew Radford’s Minimalist Syntax. Exploring the Stracters of English. CUP, 2004. Features ate relevant to (1) WORD MORPHOLOGY ice. the categorisation of lexical items); and to (2) SYNTAX (Le. the configuration of syntactic derivations at Surface Structure) GRAMMATICAL PROPERTIES of WORDS: FORM > morphological characteristics SYNTACTIC DISTRIBUTION * position in the sentence SUBCATEGORIAL FEATURES: Draw specific distinctions between the members of a particular category PRN: ATTRIBUTES > person / number / gender / case VALUE > 28°, 38 / Sp. Pl / masc, fem / NOM, OBJ, GEN CONTEXTUAL FEATURES > SELECTIONAL PROPERTIES of WORDS ‘Words impose restrictions as to their neighbouring words: He might to Patis > modal AUX verb (I) selects BARE INFINITIVE V Hei to Paris > aspectual/ progressive AUX verb (T) selects PRESENT PARTICIPLE V He has to Patis > apectual/perfect AUX verb (F) selects PAST PARTICIPLE V CATEGORIAL FEATURES Verb = [+V, -N] Adjective = [¥V, +N] Noun = [-V, +N] Preposition = [-V,—N] > some grammatical properties are CROSS CATEGORIAL: eg, umprefixation is a morphological property shared by V and A (unwind, unusual) 1. The sufegoria? features axe the set of head features catiied by each lexical element. Head features are alzeady valued (chrcked) when the lexical item entets a syntactic derivation. 2. Each lexical item also carries a set of contextual (or selctional) features which determine the type of constituent this item may merge with. These contextual features are unvalued (iachecked) at the moment the item enters a derivation. cee FEATURE MATRICES: orn: oP (FoTOLes cr Bere \ GER ‘The veerger and movement operations affecting the position of the subject in a sentence will he determined by Ts contextual features and by the head feanutes carried by the closest nominal c-commanded by T°, In a structure such as We have arrived in Prerto Rive, the auxiliary HAVE canties valued Present Tense head features and unwalued Agreement (person/number, ie phi) contextual features; and the nominal (DP) we has valued phi head featutes (person/number) and unvalued T and CASE contextual features. ‘Thus, in the detivational process, the T-probe will search its c-commanded domain for 2 nominal-goal (eg, we} whose features match its unvalued contextual features. Once the T-probe has found this matching nominal, feature copying will take place, i.e. the contextual features of the nominal will be copied onto the probe and subsequently valued. FEATURE COPYING and FEATURE VALUATION TP Nominative Case is assigned —by a finite T-probe~ to a nominal whose phi-features match its own contextual features, once they have been copied and valued. FEATURE DELETION and subject ATTRACTION: Once the DP-goal we has valued the unvalued phi-features of the T-probe, and T has valued the unvalued T. feature of the DP, the probe's and the goal’s contextual featutes will be defeted. Icidentally, valuation by T of an unvalued T feature on DP will naturally result in the assignment of a Nominative Case feature on DP. Furthermore, the auxiliary HAVE also casties an EPP-feature. EPP requires that every finite T have an overt Specifier (te. English sentences require overt subjects), and this requirement may be fulfilled either via (1) merger of an expletive, or (2) movement of the closest argument fron the thematic domain (ja accordance with the Attract Closest Princip). In our example, the DP we will MOVE to Spec-TP in order to satisfy the uninterpretable EPP feature of T. In non-finite clauses, movement of a covert subject PRO will fulfil the same EPP requirement. Summary and further considerations * The relation MATCH: a and match in respect of some feature {F} either if both have the same value for {F}, or if one is valued for {F} and the other is unvalued for {F}, but not if and 8 have different values for {F} © Feature Value Correlation = Interpretable features: TEINSE features on T, and g-featutes on nominals * Uninterpretable features: CASE featutes on nominals and g-features on’T 2. Interpretable features enter the derivation already valued 3. Featutes which enter the derivation unvalued ate uniaterpretable Interpretable and uninterpretable features at the semantic component(LF), Some of the features catried by lexical items are semantically interpretable (e.g. number, person, degree, fens), and some are semantically uninterpretable (eg, all contextual featutes, non,fiite features, CASE). The fact that in the passive voice an argument with a given S-role may take on different CASE features depending on the syntactic position it ‘occupies, constitutes a syntactic piece of evidence in favour of the uniterpretability of CASE features: © Borges (he [NOM] wrote The garden of forking paths and The Immortal (them [OBY)) © The garden of forking paths and The Immortal (they [NOM] were written by Borges (bire (OB]]) Chomsky claims that uninterpretable features should be deleted in the course of the syntactic derivation, in the specific sense that they are invisible in the semantic component (LF), but are visible in the syntax and in the PF component. The syntactic operation involved in deleting the uninterpretable features is called Feature Deletion: « deletes any uninterpretable person/ mumcber/ case feature(s) catried by B if a is q-complete and if the value(s) of any ¢-features cartied by 8 match those of the corresponding @-features of «. + The difference between the Feature Deletion operation and Trace Copy Deletion is that the former renders the affected features invisible to the semantic component (LF), while leaving them visible to the syntactic and phonological components; while the latter renders traces of moved constituents invisible to the phonological component (PF); while leaving them visible in the semantic component (LE). © For two items to menge in the course of a syntactic derivation it is mandatory that theit respective features match. If features match with each other, valuation will be accomplished. Head-features ate interpretable and already valued when the lexical item enters the derivation but conttotual features, being uninterpretable at LF, will be deleted once they ate valued. + The Inclusiveness Condition bars the introduction of new elements (ftature) in the course of a dexivation. ‘+ According to Torrego & Pesetzky, nominals mezging at Spec-VP enter the derivation with their person and uber features already valued, and the only uavalued feature catsied by these aominals is T, not CASE. So, ‘when the head of TP matches the nominal in phi-features, T's number and pervon contextwal features are valued, then the unvalued T' contextual feature on the nominal is valued by the T head feature of T®. Once these features have been valued, those which are uninterpretable are defeced, only then does the nominal move to Spec“TP in order satisfy the EPP requirement of T to project an overt specifier. ‘© In sentences with unaccusative VP's, the Spec-VP position is dethematised and vacant, therefore the EPP requirement may be satisfied either by moving one of the internal arguments of V to Spec-TP, or by mesging an expletive at Spec-TP. * Chomsky sees uninterpretable features at the very heart of agreement. And posits that “Probe and Goal must both be active for Agrement to apply”, and that a constituent (whether Probe ot Goal) is active only if it contains one or mote uninterpretable features. * Ina sentence with its notional subject in extraposition and an expletive if grammatical subject, such as If said that he bas taken bribes, we may wonder why BE cannot agree with the overall CP ‘that be has taken bribes” or ‘with the Av subject of the TP complement to this CP. It cannot agree with the CP because CP’s have no number properties, nor do they carry a CASE feature. And it cannot agree with the subject of the TP complement of this CP because CP’s are phases, and phases are impenetrable to any farther syntactic operation. Chomsky's Phase Impenetrability Condition argues that “any goal within the c-command domain of the phase CP is impenetrable to further syntactic operations. Once a phase has been completed, the domain of the phase (ic its complement) is unable to undergo any further syntactic operations.” ‘The syntax of Existential Sentences. Agreement does not necessarily involve a specifier-head relationship because Minimalism restricts all syntactic relations to one of e-command created by merger (theoretical prospective), and also because it is difficult to explain the agreement in existential sentences (descriptive prospective). Pesetzky’s Earliness Principle states that in derivations such as: There are students waiting far you, agreement vill apply when the VP “students waiting for you” merges with T. At this point, the probe BE, which haas its tense head features valued, but its contextual -features as yet unvalued, searches for and finds a goal “students”, which has its head ¢-features valued, but its contextual CASE features as yet unvalued, Valuation involves a Feature-Copying operation: if « is valued for some feature {F} and B is unvalued for (F} and if B agrees with o, the feature value for {F} on ais copied onto 8. Thus, in There are students waiting for jon, one xefiex of agreement is that the unvalued g-features carried by the probe BE are valued by the valued ‘g-features of the goal “students”, and another reflex of agreement is that the unvalued Case feature {U-Case} cattied by the goal “students” will be valued by the ptobe BE. Thetefore, if probe and goal match in -featutes, a Feature Copying Operation will take place, as well 2s Nom/native Case Assignment. However, the EPP-requirement for T to project an overt specifier will not be satisfied by moving the goal “Students” to Spec-TP, but by mexging the expletive there in spec-TP instead. This expletive has n0 semantic content, but it carries an uninterpretable persom feature, which is intrinsically 3“ person. There is a probe which is active by vistue of its uninterpretable 3© person feature, and finds an appropriate goal BE, whose 9- features have been deleted, but remain visible in the syntax and PF component. Since both #here and BE have the same {3% person} feature, the uninterpretable (3 Pers} on “there” will be again deleted. ‘There BE students waiting for you {2°Pers} {Pres Tas} {3° Pets} {32Pers} {PINum} {pLsum} {NemCase} {EPP} However, according to Bowers, expletive there originates in spec-VP, as 2 non thematic subject, and hence, it can only occur in intransitive VP’s which do not have an external argument. In unaccusative structures, like There have arisen several problems, {FAVE} probes for active matching goals which carry person and/or number features, and locates two such goals, there and several problems. Accordingly, T simultaneously agrees with both sere and several problems, resulting in multiple agreement (a probe and two or more matching goals). The {EPP} feature of T simultancously attracts the CLOSEST goal, so triggering movement of there to spec-TP., ‘Affix Hopping. Tis the locus of Tense properties in the sense that T is the constituent that catties the tense (finite or non-finite) features of a clause. T can be either overt, when it is headed by an auxiliary, or covert, as in the case of affizmative present and past simple clauses, e.g. (1) He dots enjoy Syntace. / (2) He enjoys Syntae Both structures contain 2 Ths affix (Tense) in T. but they differ in that the Tns affix is attached to the auxiliary de in 1, but is unattached in 2, because there is no auxiliary in’ for the affix to attach to, But how come the affix ends up spelled out 2s an -8 inflection on the main verb enjoys? Once the syntax has formed a clause structure the relevant syntactic structure is sent to the semantic component (LF) to be assigned a semantic interpretation, and to the PF component to be assigned a phonetic form. In the PF component, a morphological operation applies, known as Affix Hopping: in the PF component an unattached tense aftix is lowered onto the closest head c-commanded by the affix (provided that the lower head is a verb). So, seemingly auxiliarless finite clauses are TPs headed by a T constituent containing an abstract tense affix which is subsequently lowered onto the verb by an. gffx-bopping operation in the PF component. One argument in support of the tense affix analysis comes from coordination tests, eg 0 He enjoys Syntax and as learned a lot © He enjoyed Syntax and ks taking a follow-up cours In the above examples, what is coordinated is the T-bar as it comprises a present tense auxiliary (bas, i) with a VP complement. 0 [tis said they were arrested 0 He expected them to be arrested > Hewas shocked at their being arrested Floating Quantifiers: we must distinguish between features of Q and features of D. D catties an unvalued T feature. Therefore, the T-probe values ferson/ number with the D-goal and not with the Q-goal. (ecamph) ATTRIBUTE and VALUE In relation to a feature such as [singular number], monber is said to be an attribute (in the sense that it is the property being described), and singular is its value. To VALUE a feature is to assign it a value. For example, a finite auxiliary enters the detivation with its petson and umber features unvalued, ic. not assigned any value, and these are then valued via agreement with a nominal in the course of the detivation, Table of features affecting the different categories: category inflect for PRONOUN | number gender Case these, thove she-he he - him — his Noon ‘number ‘gender (wnprodocive) Case books, churches, oxen, mice _| widow / widower ut’ cas the man’s hat VERB ‘number and person | tense / mood / aspect | non-finite be) plas us) played: (we) thought ‘osing singing / su ADI degree JECTIVE | sey / easier / easiest degree ADVERB | et] bier Se Note: some demenis undesgo internal inflection

You might also like