Transformations (Primarily Movement) Grammatical Rules Are All Structure-Dependent, ↓ As Discussed in Myth 12 Early in the Semester, in S-Structure the Textbook (Pp
Announcements LNGT0101 Introduction to Linguistics LAP typed proposal is due by e-mail or as hard copy in my mailbox this Friday by 3pm. Mention the language of your choice, the interesting linguistic or nonlinguistic inspected that you hope to study in your project. List at least two references that you’ll be using.
Lecture #15 HW4 will be the final homework, so you Nov 5th, 2012 can focus on the LAP.
2
Announcements A visual puzzle
On Wed, please sit in ‘dialectal’ groups, as in the clip we saw on HW3: http://www.youtube.com/watch?v=fGxlxOcS-tE
I will send a doodle link for possible times for the screening of The Linguists.
http://www.magicmgmt.com/gary/oi_pac_tr i/#
3 4
Another visual illusion (just for fun) Summary of Syntax so far
Syntax is the study of sentence structure. They key notion to understanding sentence structure in human language is “constituency.” Constituency of a string of words can be determined by objective diagnostics: the http://www.youtube.com/watch?v=hPCoe- substitution, movement, clefting, and 6RRks&feature=player_embedded#! stand-alone tests.
6
1 Summary of Syntax so far Summary of Syntax so far
Constituents are phrases. A phrase is a string of words composed of a syntactic head, its The syntactic categories we talked about so far complement (if needed), and its specifier (if any). are: NP, VP, PP, AP, AuxP, and CP. All phrases follow the X'-schema: Our grammar thus far has two types of rules: XP (i) Phrase structure rules (PSRs) of the form ru A B C, and Specifier X' (ii) Lexical insertion rules, which insert words ru into syntactic structures generated by PSRs. X Complement
7 8
Summary of Syntax so far Today’s agenda
The grammar we have constructed so far It remains to account for sentence can account for the following aspects of relatedness. We do this with regard to the speakers’ syntactic knowledge: relationship between statements and
- Grammaticality questions (a variation on Linda’s question
- Recrusiveness from last time).
- Ambiguity We also need to explain why languages differ in their syntax. We do this with regard to word order (a variation on Danielle’s question from last time).
9 10
Sentence relatedness revisited Sentence relatedness revisited
As we said before, some sentences are Here’s the mini PSG again: intuitively “felt” to be related, e.g., 1. CP C AuxP a. Your friend can play the piano. 2. AuxP NP Aux' b. Can your friend play the piano? 3. Aux' Aux VP We know that a phrase structure grammar can 4. VP V (NP) (PP) generate the (a) sentence, but the question now 5. VP V (CP) is: Can it also generate the sentence in (b)? 6. VP V (AP) 7. NP (Det) N (PP) 8. PP (Deg) P NP 9. AP (Deg) A (PP) 11 12
2 Sentence relatedness revisited Sentence relatedness revisited
The answer then is probably not. There is no The additional rule can help, but at a high cost: PSR that will allow the Aux “can” to appear at Now, we simply have no direct explanation for the beginning of the sentence. why a statement and a corresponding question But why should this be a problem? Can’t we are felt to be related. simply add a rule that allows us to have an Aux In essence, while a phrase structure grammar head at the beginning? After all, this is a mini- can account for grammaticality, ambiguity, and grammar, not an exhaustive grammar. recursiveness, it fails to account for sentence Yes, we sure can. Here’s one possible rule: relatedness in a straightforward manner, which AuxP Aux NP VP is not a good result. To solve this problem, we need to enrich our Can this rule help? grammar.
13 14
Transformational rules Transformational rules
A solution, first proposed by Chomsky in the 1950s, is to include another What is a transformational rule? component in the grammar in addition to A transformational rule is a syntactic the phrase structure component: a operation that takes one structure as input transformational component that and operates on it producing a modified consists of a set of transformational syntactic structure as output. rules.
15 16
Deep and surface structure Deriving English yes-no questions
For this purpose, a fundamental distinction in So, let’s now get back to the yes-no question the grammar has to be made between two “Can your friend play the piano?” and see how separate levels of structure: we can implement a transformational analysis. (a) a pre-transformational structure, which is Now, instead of drawing a tree for the yes-no called deep structure (or D-structure) and is question directly, we actually draw a tree for the derived by phrase structure rules, and corresponding statement “Your friend can play (b) a post-transformational structure, which is the piano.” called surface structure (or S-structure) and The only difference is that such structure will be is derived through the application of marked as interrogative. We can do that, say by transformational rules. adding a [+Q] feature on C in the tree.
17 18
3 Now, a transformation moves Aux to C, thereby deriving: Your friend can play the piano. Can your friend play the piano? CP ei CP C AuxP +Q ei eo NP Aux’ C+Q AuxP ru ei can eo Det N Aux VP NP Aux' your friend can ru ru ei V NP Det N Aux VP play ru Aux-to-C Det N Movement your friend ru the piano V NP play ru D-structure Det N the piano (Note: [+Q] indicates this sentence is interrogative. After all, we do not want to say that both sentences are identical. They obviously are not.) S-structure 19 20
Evidence for Aux-to-C movement Deriving wh-questions
But how do we prove that there is actually Ok, let’s try another kind of question, the Aux-to-C movement in English yes-no so-called wh-questions, e.g., questions? What evidence is there that What will your friend play? ‘invisible triangles’ actually exist in syntax? Since “what” is interpreted as the object of Well, consider: “play,” we assume that this is where it He asked if your friend could play the piano. starts at D-structure: *He asked if could your friend play the piano. your friend will play what
21 22
Now, since this is a question, we apply Aux-to-C We apply PSRs to derive the D-structure: movement to derive the S-structure: CP CP ei ei C AuxP C AuxP +Q +Q will eo eo NP Aux' NP Aux' ru ei ru ei Det N Aux VP Det N Aux VP your friend ru your friend will ru V NP V NP play | N play | what N But does that give us the desired sentence? what 23 24
4 CP Where do wh-phrases end up? ei NP C' To get the desired surface structure, we | ei what C AuxP need to move the wh-phrase “what” to the +Q front of the sentence. will eo NP Aux' The question now is: Where does the wh- ru ei phrase move to? Det N Aux VP There is a restriction, however. It’s called your friend ru structure perseveration: Phrases can V NP move only to specifier positions, and play | heads can only move to head positions. Wh-movement t
(“t” stands for trace of the moved phrase, another invisible triangle.) 25 26
A puzzle: wanna-contraction A puzzle: wanna-contraction
Who do you want to kiss? Provide a principled explanation for the Who do you wanna kiss? contrast for 4 points of extra credit. Deadline: Monday Nov 12th in class. Who do you want to kiss Mary? Assume that ‘do’ starts under Aux. *Who do you wanna kiss Mary? Assume also that verbs like ‘want’ subcategorize for an AuxP complement Compare: I want to kiss Mary. I wanna kiss Mary. headed by the Aux element ‘to.’ So, a VP rule in that case is VP V AuxP And think triangles.
27 28
Parameters of question-formation Parameters of question-formation
In other languages like Japanese, Chinese, and Notice that not all languages are like Egyptian Arabic, the wh-word appears where English when it comes to wh-questions. other nouns normally appear: Japanese Some languages like English form a John-ga dare-o butta ka? question by fronting the wh-word: John-Subj who-Obj hit Q-particle What did you see _? “Who did John hit?” Egyptian Arabic These are typically referred to as wh- /inta Suft miin? fronting languages. you saw who “Who did you see?” This type is called wh-in-situ languages. 29 30
5 Syntax: The grammar model Universal Grammar: Principles and Parameters
Phrase structure grammar (X'-theory) Languages differ because UG (Universal ↓ Grammar, remember?) includes two components: D-structure principles and parameters. The principles are ↓ invariant; they hold in all languages. For example, Transformations (primarily Movement) grammatical rules are all structure-dependent, ↓ as discussed in Myth 12 early in the semester, in S-structure the textbook (pp. 157-60). Parameters are also universal, but unlike But if this language model is universal, why do principles, they come in the form of (usually) languages differ then? binary options, and this is where the locus of cross-linguistic variation exists.
31 32
UG: principles and parameters UG: principles and parameters
As Chomsky notes: Or, we can represent this graphically as follows: “We can think of the initial state of the faculty of language as a fixed network connected to a switch box; the network is constituted of the principles of language, while the switches are the options to be determined by experience. When the switches are set one way, we have Swahili; when they are set another way, we have Japanese. Each possible human language is identified as a particular setting of the switches—a setting of parameters, in technical terminology.” UG Japanese English
33 34
UG: principles and parameters UG: principles and parameters
Under this approach, a child’s job is to “set” the value of each parameter on the basis of the PLD We can think of UG as an initial state S that 0 in the linguistic environment around her. gets mapped onto a final state SF, through exposure to primary linguistic data (PLD). This should explain the role of the environment in language acquisition: If you’re born in Beirut, S0 + PLD SF then your PLD are different from the PLD of
S0 is the general system that we are born with, someone born in Moscow, hence the acquired system will be different. and SF is what we end up referring to as English, Finnish, Tiwa, Khmer, etc. Under this approach, language acquisition is the result of interaction between nature (principles and parameters) and nurture (PLD).
35 36
6 UG: principles and parameters Variation in basic word order
So, if I haven't confused you already with Even though languages may allow several word orders in sentences, each language typically has all these invisible triangles, you must be one order that is used in “neutral” contexts. This wondering when I’ll start giving you some is what is called “basic word order.” examples of parameters. Consider English, for example: Which of these Let me start with a parameter that should do you think represents the basic word order in English? help us explain variation in basic word Seafood I like. (OSV) order across languages. Believe you me. (VSO) John plays the piano. (SVO)
37 38
Basic word order Basic word order
If we confine ourselves to transitive SVO: English clauses with three elements: Subject, Verb John loves Mary. and Object (S, V, O), then we should expect six possible basic word orders in human language: SOV: Japanese SVO, SOV, VSO, VOS, OVS, OSV John-ga Mary-o butta Do we find these attested in natural John-SU Mary-OB hit languages? “John hit Mary.” Actually, we do.
39 40
Basic word order Basic word order
VSO: Welsh OVS: Hixkaryana (Carib) Darllenais I y llyfr Kanawa yano toto read I the book canoe took person “I read the book.” “The man took the canoe.”
VOS: Malagasy (Austronesian) OSV: Nadëb (Maku) manasa ni lamba ny vihavavy samũũy yi qa-wùh wash the clothes the woman howler-monkey people eat “The woman is washing the clothes.” “People eat howler-monkeys.”
41 42
7 Distribution of basic word order Distribution of basic word order types in the world’s languages types in the world’s languages
As it turns out, typological studies reveal Word order # of Languages % preferences for certain word orders than SOV 180 45 others. SVO 168 42 VSO 37 9 VOS 12 3 Consider the frequencies reported in Tomlin’s (1986) language sample, for OVS 5 1 example: OSV 0 0
43 44
Distribution of basic word order English vs. Japanese types in the world’s languages
English: The child might think that she will show Mary’s picture With greater than chance frequency, then, SVO of John to Chris. and SOV orders indicate a clear preference for word order in natural languages. Japanese: Taroo-ga Hiro-ga Hanako-ni zibun-no But what’s even more interesting is that each of Taroo-SU Hiro-SU Hanako-to self-POSS these two common orders has a set of syasin-o miseta to omette iru correlates that go with it. To see what this picture-OB showed that thinking be “Taro thinks (literally, is thinking) that Hiro showed a means, let’s compare English and Japanese. picture of himself to Hanako.”
45 46
Phrase structure: Word order correlates English vs. Japanese
Element A Element B English Japanese V NP A precedes B A follows B How do we express the difference V PP A precedes B A follows B between English and Japanese in terms V embedded CP A precedes B A follows B of the X'-schema for phrase structure
P NP A precedes B A follows B then?
N PP A precedes B A follows B Obviously, in English, heads precede their complements; in Japanese heads C embedded AuxP A precedes B A follows B follow their complements. Aux VP A precedes B A follows B
47 48
8 The X'-schema in The head directionality parameter English vs. Japanese
The difference between English and Japanese English Japanese thus comes down to the “directionality” of the head within the phrase: heads are initial in XP XP English, but final in Japanese. This is typically ru ru referred to as the head directionality (HD) Specifier X' Specifier X' parameter: Heads occur initially (i.e., before their ru ru complements) or finally (i.e., after their Xhead Complement Complement Xhead complements) within phrase structure.
49 50
The head directionality parameter How about subjects?
Notice that the HD parameter does not say anything about the position of subjects in The head-initial setting of the HD sentences, since these are not complements parameter holds in English, Edo, Thai, (they are specifiers, remember?). Khmer, Indonesian, Zapotec and Salish, This is actually good, since English and while the head-final setting holds in Japanese are both subject-initial. We don’t want to parameterize that. Rather, in both languages, Japanese, Lakhota, Turkish, Basque, the subject is the specifier of Aux: Navajo, the languages of the Eskimos, AuxP NP Aux' and Quechua.
51 52
So, why do English and Japanese English vs. Japanese look dramatically different then? English Japanese
Now, let’s try to make things more CP C AuxP CP AuxP C interesting and see how and why English AuxP NP Aux' AuxP NP Aux' and Japanese do really look dramatically Aux' Aux VP Aux' VP Aux different on the surface. VP V (NP) VP (NP) V That’s where trees can help for sure. Here VP V (PP) VP (PP) V are some PSRs for both languages: VP V (CP) VP (CP) V PP P NP PP NP P NP N (PP) NP PP N 53 54
9 So, why do English and Japanese John said that Mary read the book. look dramatically different then? CP CP C AuxP ru AuxP NP Aux' Cdeclarative AuxP Compare English and Japanese again: Ø ru Aux' Aux VP NP Aux' VP V (NP) John said that Mary read the book. John ru Aux VP VP V (PP) +past ru VCP VP V (CP) John-ga Mary-ga hon-o yon-da-tu it-ta say ru PP P NP C AuxP declarative NP N (PP) John-SU Mary-SU book-OB read-past-comp say-past that ru NP Aux' Mary ru Aux VP Given the PSRs for both English and Japanese, +past ru VNP the structural trees will look as follows: read the book
55 56
John-ga Mary-ga hon-o yon-da-tu it-ta John-SU Mary-SU book-OB read-past-comp say-past So, …
CP CP AuxP C ru AuxP C AuxP NP Aux' declarative The principles and parameters approach ru Ø Aux' VP Aux NP Aux' accounts for word order correlates in SVO and John-ga ru VP (NP) V VP Aux SOV languages in a straightforward manner. VP (PP) V ru ta CP V VP (CP) V ru it PP NP P AuxP C Notice also how a simple difference in head ru tu directionality leads to a dramatic variation on the NP PP N NP Aux' Mary-ga ru surface, due to its cumulative effect on all heads VP Aux and complements in a language. ru da NP V hon-o yon
57 58
And … How about VSO languages?
In addition, since the HD parameter does not Remember that 9% of the languages in apply to specifiers, it follows that both English and Japanese will behave the same with regard Tomlin’s sample are VSO. Why do these to the position of subjects in sentences. languages exist? Do they follow from the head directionality parameter? Finally, since the HD parameter has two settings Well, the first thing to notice is that in only, it predicts two types of languages, SOV these languages the verb comes before and SVO, which is exactly what we find in language samples: these two orders represent the object. So, they must be … about 90% of human languages. Right, head-initial.
59 60
10 Deriving VSO basic word order The subject placement parameter
But then the main difference in their word order as opposed to SVO and SOV Let’s follow Mark Baker, the author of The languages is that the subject follows, rather than precedes, the verb. Atoms of Language, and call it the Subject Placement parameter: So, how can our theory of grammar “derive” VSO orders then? “The subject of a clause is in the specifier Head directionality can’t do it. So, there of VP (as in Welsh), or in the specifier of must be another parameter involved. What AuxP (as in English).” could that be?
61 62
The subject placement parameter The English-Welsh contrast
The subject placement parameter then has to do with the phrase structure rule that introduces CP CP ru ru subjects : C AuxP C AuxP English: ru ru AuxP NP Aux' NP Aux' Aux VP Subject ru ru Aux' Aux VP Aux VP NP V' Welsh: ru Subject ru V NP V NP AuxP Aux VP VP NP V' Subject position in English is high Subject position in Welsh is low
63 64
Welsh Welsh
Given the subject placement parameter, CP the structure of Welsh sentences with ru C AuxP auxiliaries becomes straightforward. -Q ru Here’s an example, followed by a tree: Aux VP (1) Naeth y dyn brynu gar Naeth ru did the man buy car NP V' y dyn ru “The man did buy a car.” V NP brynu gar
65 66
11 Welsh The verb movement parameter
Ok, but how about this other Welsh example, then? I guess it’s time for me to come clean on (2) bryn-odd y dyn gar how Aux and V eventually get together. It buy-Past the man car turns out there are two options, thereby “The man bought a car.” formulating another parameter. There’s no overt auxiliary here, so how does the “V moves up to Aux (Welsh), or Aux verb come to precede the subject? moves down to V (English).”
67 68
The verb movement parameter Welsh
So, the reason why Welsh is always verb- CP CP initial is because the Aux head has to host ru ru a verb (either an auxiliary verb, or a main C-Q AuxP C-Q AuxP verb, if an auxiliary head is absent). ru verb ru Aux VP movement Aux VP The tree structures for the Welsh example -odd ru bryn+-odd ru in (2) before and after movement takes NP V' NP V' y dyn ru y dyn ru place would be as follows: V NP V NP bryn gar gar
69 70
Welsh Parlez vous français?
VSO languages like Welsh and Irish are thus Compare the position of adverbs in English and possible because of the interaction between two French: parameters: the subject placement parameter John often kisses Mary. and the verb movement parameter. *John kisses often Mary. If you understood these syntactic gymnastics, you must be asking: How do we prove this? Is *Jean souvent embarasse Marie. there any evidence for the assumption that in Jean often kisses Marie. English Aux moves down to V? Jean embarasse souvent Marie. Luckily, there is. Let’s contrast English and Jean kisses often Marie. French.
71 72
12 Verb position in English vs. French Verb position in English vs. French
English French English French CP CP CP CP ru ru ru ru C-Q AuxP C-Q AuxP C-Q AuxP C-Q AuxP ru ru ru ru NP Aux' NP Aux' NP Aux' NP Aux' John ru Jean ru John ru Jean ro Aux VP Aux VP Aux VP Aux VP -past ru -past ru ru embarasser+[-past] ru Adv V' Adv V' Adv V' Adv V' often ru souvent ru often ru souvent ru V NP V NP V NP V NP kiss Mary embarasser Marie kiss+[-past] Mary Marie 73 74
Interim summary Interim summary
Welsh differs from both English and French in So, here’s the story: having the subject placed in the specifier of VP. English, French, and Welsh, all share the English and French subjects are in the specifier of same head-initial setting for the HD AuxP. parameter, as opposed to English differs from both French and Welsh in Japanese/Turkish/Navajo, which are head- having Aux move down to V. In French and final. Welsh, V moves up to Aux. The interaction of these parameters gives us But: English, Japanese, Welsh, and French. If I haven’t confused you by now, then why not look at German/Scandinavian? 75 76
Sprechen Zie Deutsch? German: The V2 effect
The verb is always the second constituent a. Ich las letztes jahr diesen Roman in German sentences, following the subject, I read last year this book or a fronted object, or an adverbial. b. Diesen Roman las ich letztes jahr If that is the case, then it must be that this book read I last year German, like French, has V-to-Aux c. Letztes jahr las ich diesen Roman movement. last year read I this book Unlike French, though, German can even have the verb before the subject. So, what do you notice here about the position of the verb in German? Hmmm … what’s going on here?
77 78
13 German: The V2 effect German: The V2 effect
German (sketchy since German word order is a matter of debate) If V can move up to Aux in declarative clauses (as in French and Welsh), one can imagine a CP language where V can keep moving all the way ru Specifier C' up to C, right? At least, the system of sentence Diesen Roman ru structure we’re using here does not prevent that C AuxP las ru from happening. NP Aux' And that seems to be what is happening in ich ru German main clauses. Let’s call this the V2 … VP … ru parameter. The parameter also holds in … V … Scandinavian languages.
79 80
German: The V2 effect Parameters and languages so far
Parameter English Japanese French German Welsh Is there evidence from German showing that its V moves to C? HD Head- Head- Head- ? Head-initial parameter initial final initial Yes. Consider: a. Hans schlugden Ball Subject Specifier of Specifier of Specifier of Specifier Specifier of placement AuxP AuxP AuxP of AuxP VP Hans hit the ball parameter ‘Hans hit the ball.’ Verb Aux down ? V up to V up to V up to Aux movement to V Aux Aux b. Ich denke daB Hans den Ball geschlangen hat parameter V2 No ? No Yes ? I think that Hans the ball hit has parameter ‘I think that Hans hit the ball.’ “?” indicates issues that we simply did not address in this class; it does not mean that linguists don’t know the settings of these parameters in such languages. 81 82
VOS/OVS/OSV languages
VOS languages should be derivable by a parameter for subject position. I’ll let you figure this one out on your own. Another parameter: Do you OVS/OSV languages are not that well need to ‘verbalize’ your subject? understood, but there are definitely ways to derive their word order. In the interest of time, we won’t be discussing them here. If your LAP language ends up being of either type, then let the class know what you find out.
83 84
14 The null subject parameter The null subject parameter
Consider these data from English, Italian, however, allows the subject of a French, and Italian, all of which allow SV tensed sentence to be omitted, an option (=Subject-Verb) orders: that is not available in English or French: (1) John will leave. (5) *Will leave. (2) Jean arrivera. French (6) *Arrivera. French Jean will-arrive will-arrive (3) Gianni verrá. Italian (7) Verrá. Italian Gianni will-come. will-come.
85 86
The null subject parameter Summary
This case of cross-linguistic variation is typically referred to as the null subject These are some examples of parameters. parameter. I hope the notion is clear by now. “In some languages (e.g., French, English, Edo) every tensed clause must have an overt subject. In other languages (e.g., Italian, Spanish, Romanian, Navajo, Arabic) tensed clauses need not have an overt subject.”
87 88
Next class agenda
Sociolinguistics: Chapter 10, pp. 430-452.
89
15