Scarica Syntactic Structures e più Appunti in PDF di Morfologia e Sintassi solo su Docsity!
Week 1
Lesson 2
The building of the sentence
Syntax is about combining different units ( words and affixes ) together, then combine the bigger units ( phrase ), until all of the units have been used, and a sentence^1 has been constructed. It is kind of instinctive for a native speaker to know how to combine these elements ( Chomsky 1986, Knowledge of Language → generative grammar^2 ; Darwin talked about human instinct toward language).
Two fundamentals of functioning of the syntax:
- Merge : the computational operation to combine two entities to build up a bigger one;
- Recursive character of Merge : Merge can be iterated and reapplied, to form more complex outputs.
Syntactic Structures
Sentences are made up of organized sets of words and affixes; words combine (this operation is called Merge ) and this gives rise to Syntactic Structures. Since the organization of syntax is hierarchical , and not linear^3 , we say that words are organized in structures , not strings.
Merge
Merge optimally shows the syntactic hierarchy: two units are combined to form a new element, which can be combined again with another element and so on. (^1) In italiano sintagma. (^2) Human language is shaped by a set of basic principles that are part of human brain; this Universal Grammar comes from our innate language faculty. (^3) Linear is just the shape of the oral or written expression of language, but its fundamental organization is hierarchical.
“a + b” and “c + [a b]” are the horizontal notation with square brackets of the vertical trees. Terminology Head → Word or affix that drives the operation Merge (External Merge). The head determines the category of that phrase. All phrases have to have a head. Phrase → a group of words which belong together as a unit, created by the Merge. Phrases do a single job in the sentence (also called constituents ). Example: After the inspiring lecture, the students thanked the visiting teacher from an overseas university. After the inspiring lecture (phrase 1: when?), the students (phrase 2: who?) thanked the visiting teacher from an overseas university (phrase 3: did what?). Peter is my father. That tall man walking down the street is my father. In this last example, Peter and that tall man walking down the street are both doing the same job: to specify who is the father, so both are considered as phrases. Complement → Position of the phrase that combines with the head in the first application of Merge. Specifier → Position of the phrase that combines in a further application of Merge. The phrase Traditional standard schema of a phrase (X’-schema): Where X is the closest head to the phrasal level, so the label of the phrase is XP. Specifier and Complement are Phrasal positions (they are phrases, not heads ).
Some general background:
I-Language^5 and Universal Grammar : the language faculty is a human cognitive capacity that responds to a natural instinctive capacity to acquire any language by any child, internal to the mind/brain. UG is the name given to this human language capacity, which is shown by neuroimaging techniques. Linguists have indirect access to UG through:
- Language description
- Language comparison
- The study of forms of different types of linguistic data (language acquisition; language pathologies…)
- Experimental data Language faculty is characterized by invariable and (compatible with the former) variable properties. Rules and internal grammar In internal grammar, rules are not seen as regulating instruction, as in normative grammar, but instead they are internal computational mechanisms: explicit instructions can be learned, but they do not activate the language areas of our brain (unless they reflect the internal mechanisms). Neuroimaging evidences the different status of the 2 different kinds of rules ( internal grammar and normative grammar ). Impossible rules: it is impossible to set rules on a linearity level, because syntax is built in a hierarchical way. Language Acquisition, Natural Language & Language Faculty Data → LAD^6 → Competence What is a natural language?
- A combination of sounds and meaning (Aristotle)
- A computational mechanism/a generative (explicit) procedure (Chomsky)
- A sensory motor system (externalization through sounds or signs)
- A conceptual/Intentional system The basic property of human language → a finite computational system which can yield to an infinity of expressions (recursion). Plus, we make a creative use of our faculty of language, in the sense that when we combine all the possible expressions, we build sentences that we’ve probably never heard or said before. → UG. (^5) Internal-Language (^6) Language Acquisition Device (Universal Grammar is a model of LAD). With “language” we indicate Natural Languages.
Lexical vs Functional head Lexical heads vehicle a descriptive content, they include N (noun), V (verb), A (adjective) Functional heads vehicle a grammatical content, as D (determiner); P (preposition); C (complementizer^7 ); T (tense^8 ) (^7) For example, the word “that”, complementizer of a complement clause and not a nominal clause. (^8) All those features that are express
The head parameter is one of the first parameters that gets naturally acquired by children. In fact, children do not do mistakes in this sense: they immediately recognize and instinctively understand the parameter that regulates the head-complement relation, and they do not have doubts in recognizing the correct order of the two elements with respect to the language they are acquiring. Ex: LATIN: Caesar [omnem agrum Picenum] percurrit (O-V) ITALIAN: Cesare percorre [tutto il territorio dei Piceni] (V-O) GERMAN: ...dass Peter [das Buch] gelesen hat (O-V) ITALIAN: …che Pietro ha letto [il libro] (V-O) In German, as we can see in embedded clauses (subordinate clauses^9 ), the object goes before the verb, and also, the auxiliary of the verb ( hat ) that is the carrier of the tense, goes in final position. In Italian, in subordinate clauses we have the order VO, so we find no difference in order between roots (main clauses) and embedded clauses (subordinate clauses). English follows the same scheme as Italian (head goes first). JAPANESE: John-ga Mary-wo butta (OV) In Japanese, the head is always put in final position (if it’s a verb), both in main and subordinate clauses. Japanese is, in fact, a head-final language, and all the verb’s complements that are present in the sentence precede the verb. Comparing Japanese and German in their head-complement relation, we see that while Japanese has postpositions (Mary- to ), German, as English and Italian, has prepositions ( mit Mary, with Mary, con Mary). So, it is immediately possible for a child acquiring first language to internally distinguish those parameters that regulate the head-complement relation in the data he gets in contact with. From this explanation we understand that in VO languages, head is higher than the complement phrase in the hierarchical structure of the sentence, while in OV languages it is the opposite case: through the manipulation of the structure, the head ends up in a lower position with respect to the complement part.^10 UG and Parameters Parameters can be considered as the primitive ingredients of the computational system of language faculty: they regulate the UG. In fact, Parameters are not phenomena: they are formal properties that manifest themselves in linguistic phenomena, and their value varies by language according to their presence (+) or absence (-) from that particular language. Our job is to find which is the formal property that triggers this process. Language variation occurs within limits: variation of a language cannot (of course) exceed the limits of UG, that are universally valid. For example, a language cannot be structured in a non-hierarchical way because the hierarchic structure of the language is common to all existing languages. It is (^9) Introduced by a complementizer, in this case dass (that). (^10) Let’s remember that syntax is not developed in a linear way, but in a hierarchical structure, and that precedence in linearity means higher in structure.
impossible that a language could vary in this domain: UG is in fact the internal grammar of any speaker, it includes all the properties that all languages have (even the invented ones). Parameters Parameters can be individuated by observation of data that can be acquired regarding syntactic phenomena (language description). Let’s compare two languages keeping the relevant structures as similar as possible to each other: the eventual differences found will constitute a minimal pair. ENGLISH
- Did Mary go out?
- No, she just came back
- Mary says that she wants to talk with John
ITALIAN
- Maria è uscita?
- No, ( pro )^11 è appena tornata
- Maria dice che ( pro ) vuole parlare con Gianni As we see from the above examples, while in English we must always repeat the subject through the use of a pronoun, in Italian we find a crucial property of personal pronouns, that is called null- subject. Null-subject means that it is not necessary to use a pronoun to underline who the referent is, unless we want to emphasize it for some reason (for example, if we specify a context for sentence
- in which no one wants to talk with Gianni anymore. Ex: Maria dice che LEI vuole parlare con Gianni , ma altri non vogliono neppure considerare l’idea di incontrarlo. Here the emphasis is put on LEI because it underlines a comparison between “others” that do not want to talk with him, and Maria that does, instead). In fact, Italian also has pronominal subjects that are possible option, if the subject must be marked for a particular purpose, as in the example in the brackets. So, if we use a pronoun in Italian, it will have a slightly different meaning of the English one, because it will imply the will to underline something that is not considered in the English equivalent sentence. The direct corresponding Italian translation to the English sentence 3. would be involving a null-subject. A fundamental distinction: Lexical vs Functional Heads Parameters are properties that concern the functional lexicon (as pronouns^12 ). LEXICAL HEADS N (noun) V (verb) A (Adjective)
FUNCTIONAL HEADS
D (determiner) P (preposition) C (complementizer) T (tense) (^11) In Italian we have a “silent pronoun”. It is called Null-subject. (^12) Because they express grammatical features, as genre, number…
Example 2: Probably the man who will talk to the girl will go out of the room In this case, “probably” is referred to “will go out of the room”. In this case we have an external merge. Difference between external and internal merge : external merging is combining two units to build a bigger unit including both of them; internal merge is taking something from an already build structure and remove it from its position to move it in another place. Internal merge is just another way to refer to movement. Phrases Each phrase has a head. The head is either a lexical word (noun, verb), a functional word (determiner, preposition) or an affix/feature^16 (tense, number). NP → N VP → V PP → P AP → A DP → D D Each V in internal lexicon has an argument structure associated with it, composed of roles that are carried by nominal arguments in the syntactic structures [Theta roles θ1, θ2, θ3, which correspond to 3 verbal classes]. These arguments represent the participants in the event represented by the verb. In languages with determiners, an NP (noun phrase) is always introduced by a determiner (“una/la/quella ragazza compra il/un/quel libro” is correct; “ragazza compra libro” is incorrect), with exception for generic plurals and mass nouns. Another parameter (related to functional head): a language can openly have determiners (D) or not. The head of a phrase with an NP as complement is necessarily a D, if the language has determiners. (^16) Expressing a grammatical content
In languages that have no determiners, a process occurs that moves the N in the D position, so that, the position of the D, that in those languages remains available, gets occupied by N. So we can say that D is always the head of the phrase whose complement is an NP. DP As we see, there is always an introductive functional part (D) introducing the lexical part (N). In this case, the functional spine is the part in blue, while the part in black is the lexical spine. “A” is part of the functional spine^17 , because it expresses a type of modification. Some languages distinguish adjectives, so it is debatable weather they belong to the lexical or to the functional part. But in this case and in general, if they express a type of modification that is not subjected to inflection (as color adjectives), they are part of functional lexicon. Structures: iterating the same operations Let’s look at some more common noun phrases:
- Il bel viaggio nella città della ragazza/Maria
- The beautiful trip in the girl’s/Mary’s town (^17) Sometimes the lexical spine is called “extended projection”.
In this case, α c-commands β; and β c- commands α. There is no dominance between α and β, but both α and β are immediately dominated by the same node A. In this case, α c-commands B and β; and B c- commands α, but β does not c-command α, because α is not contained in the sister node of β (the sister node of β is empty). Here α is higher than β. β is contained in the sister node of α, which is B, and both α and B are dominated by A. In this case A dominates B and C. B and C c- command each other, and B also c-commands β, which is contained in its sister node C. C c- commands α and γ, beyond B, because they are both contained in its sister node B. B dominates α and γ, while C dominates β. α c- commands γ and γ c-commands α. But nor α or γ do c-command β, because β is not included in the sister node of α or γ. Similarly, β does not c- command anything, because its sister node is empty. Lesson 6 7. C-command and binding The domain of binding is where c-command relation becomes crucial: we distinguish three fundamental principles in nominal expressions: anaphor , pronoun and r-expression.
1. Principle A: an anaphor is bound (in the relevant domain).
Maria ( si^19 ) critica sé stessa Maria criticizes herself In this case, sé stessa (herself) is the anaphor^20. We say that the anaphor is bound to the main clause (the relevant domain) because it must agree with the referent (Maria). It would be impossible to say “Maria critica sé stesso”, or “Maria criticizes himself”, because the relevant domain (in this case Maria) would not c-command the anaphor. Anaphor is always bound to the referent. C-command occurs in the smallest domain, so, for example: *Mary says that John criticizes herself (wrong: John does not c-command the anaphor, and Maria is outside the relative clause) Mary says that John criticizes himself (right: c-command is local, John c-commands the anaphor)
2. Principle B: a pronoun is free – not bound – in the relevant domain.
Mario lo conosce bene Mario knows him well Here we see the difference between anaphor and pronoun: lo refers to someone else, so it is not bound to Mario (which is the relevant domain). La sorella di Gianni lo conosce bene John’s sister knows him well In this case is even more evident that the pronoun does not agree with the referent (the sister) in the relevant domain. Moreover, lo and him are not necessarily referring to Gianni (or John), they could be also referring to someone else. If the pronoun is referring to Gianni, then it must be bound to the embedded clause, otherwise there is no necessity to bound the pronoun to any of the present referents.
3. Principle C: an R-expression is free (in any domain)
Lui( pro ) conosce Gianni He knows John As we see, R-expressions are completely free in any domain. Let’s see the example: Lui dice che Maria conosce Gianni He says that Mary knows John It is evident that Lui/he is not bound to Maria/Mary or to Gianni/John, because its referent is not present in the sentence. So, the r-expression appears to be free both from the relevant domain (main clause) and from the embedded clause. (^19) We call this kind of clitic proclitic , because it precedes the verb. (^20) Reflexive pronoun = anaphor.
Principle B: ENGLISH: In “Mary criticizes her”, her cannot be referred to Mary, because, as we saw, pronouns are not bound to the main domain. Mary c-commands her. ITALIAN: In Italian, personal pronouns are ( pro- ) clitics. This property does not affect their binding properties as pronouns, which remain constrained by principle B. But, because they are proclitics, they are moved in a higher position inside the tree, if compared with an English translation of the phrase. Principle C: Clearly, “she” c-commands Mary. So, Mary and she are necessarily two different people, because the requirement of principle c is for r-expressions to be free across all the domains. It is similar to principle b, but stronger. Furthermore, in principle C, the subject of the DP c-commands the whole clause. The case of Null-Subject Mary says that she wants to talk with John Maria dice che vuole parlare con Gianni We said that in this case, she could be referring to Mary or to someone else, it depends on the context. So, the relevant property of pronouns here, is precisely the fact that the pronoun is free in the given domain (the embedded clause), so it can pick any referent, including Mary. This is because the principle B has this negative requirement in the given domain. In this domain, she is free, there
is no binder. So Mary can be the referent, or somebody else can be the referent. Moreover, we said that the null subject, in null-subject languages like Italian, is a pronoun, is not an anaphor. So, implicitly, it behave following principle B. And indeed, respecting to the interpretation, it can have features (genre, number) that accord to the main domain referent, or to an external referent. Pronouns are part of the functional category, in fact they express features like number, person, genre. Towards sentence structure: V… C… T //…V… Now we are putting our attention on the fact that c-command is relevant even in manipulation (moving elements inside the trees). Movement always occurs from a lower to a higher position (to a c-commanding position).
Complementizer: di/to (infinite), che/that (finite)
Let’s consider these examples: Ho deciso [che (-) partirò] Ho deciso [di (-) partire] It would be incorrect to say: Ho deciso [di partirò] Ho deciso [che partire] This happens because partirò and partire are introduced by two different complementizers ( che , di ). So, complementizers influence the morphological inflexion of the verb (tense): the complementizer expresses the type of verbal inflection. The selection of the type of clause is made by the verb (declarative, interrogative…). Moreover, normally, the verb chooses the type of complementizer (ex: I wonder → IF). So, there is a relation between V and C (V selects C), then there is a relation between C and T (C determines T). 22 Che , introduces a finite morphology, while di introduces an infinite morphology in Italian. The same is in English with: I decided [that (I) should leave] I decided [to (-) leave] (^22) V → verb; C → complementizer; T → tense.
Complement : it is a phrasal or clausal category that is necessary to complete the meaning of a given expression, and it is selected by the head of a phrase. It is daughter to the X’ level and sister to the X (head). The map of Canada (C) The best in town (C) She becomes a doctor (C)
Constituent
A constituent is a group of words that function together as a unit. They are embedded one inside another to form larger constituents. Ex: the student loved his syntax assignments. We can distinguish them by adding brackets in the linear sequence: [TP[NP[D the][N student]][VP[V loved][NP[D his][AdjP[Adj syntax][N assignments]]]].
Week 3
Lesson 7
Toward a cartography of the clause structure
The overall functional spine of the clause contains:
- Heads expressing grammatical content typically connected to the verbal morphology or to the tense of the expressed clause.
- Modifiers of such content. They can be adverbs, adjectives (they express (T)ense. (M)ood, (A)spect).
- Syntactic positions of the discourse content (they can also be morphologically marked). All these positions belong to a certain area of the clause. One crucial aim is to design a detailed map of the functional spine. This program is often called Syntactic cartography.
The functional spine of the clause
T:
Abbreviation for a series of inflectional features , heads corresponding to different functional positions in the syntactic structure. T is typically realized on the verbal inflection ( tense ), and it is often expressed with affixes, for example: ITA: Ride- va - mo root of the verb- past tense – 1 st^ person plural
C:
Abbreviation ( complementizer ) for different positions in the immediate (left) periphery of the clause. The C head that characteristically hosts the element introducing a subordinate clause, indicates a series of positions in the left periphery of the clause where syntactic constituents (DPs, PPs…) with special discourse values are moved (such as questions, focus, topic, CLLD).
