1. GRS LX 700 Language Acquisition and Linguistic Theory
Week 10. The Trouble With Principle B

2. Binding Theory Binding Theory Constraints on assignment of reference.
Reflexives (himself, herself, themselves, …)
Pronouns (he, she, they, him, her, …)
Names (inherent reference)

3. Binding Theory
Principle A A reflexive (herself) must be bound within its governing category.
A binds B iff
A and B are coindexed,
A c-commands B.
Governing category ≈ IP or DP.
Mary saw herself in the mirror.
Mary said John saw herself in the window.
John stole [Mary’s pictures of herself].
Mary stole [John’s pictures of herself].

4. Binding Theory
Principle B A pronoun (her) must be free (=not bound) within its governing category.
Mary saw her in the mirror.
Mary said John saw her in the window.
John stole [Mary’s pictures of her].
Mary stole [John’s pictures of her].
Governing category ≈ IP or DP.

5. Binding Theory
Principle B A pronoun (her) must be free (=not bound) within its governing category.
A pronoun can be bound
Accidentally (by reference)
John2 said he2 eats fish.
By a quantifier
Every boy2 said he2 eats fish.

6. Binding Theory
Principle C A name/r-expression (Mary) must be free (altogether).
She saw Mary in the mirror.
She said John saw Mary in the window.
Mary stole [his pictures of John].
He stole [her pictures of John].
He said that Mary believes Sue stole my pictures of John.

7. The trouble with acquiring constraints on meaning
Every bear is washing her face.
Bunch of bears washing Goldilocks’ face.
Bunch of bears cleaning their own faces.
Every bear is washing her.
Based on what evidence would kids conclude that the second context is not described by the second sentence?

8. Ordering
In many cases, it might look like the results from Binding Theory are just about order—but for adults, there’s more to it. Binding Theory is abstract, about structure.
He said that Mickey won.
Mickey said that he won.
Before he went to school, Mickey ate a sandwich.
No c-command, no problem.

9. Binding Theory
The principles of Binding Theory seem to be universal, represented in all languages.
They prohibit certain interpretations (that is, are unlearnable from positive evidence)
The principles of Binding Theory are part of Universal Grammar, not learned.

10. Binding Theory
Yet… Experiments seem have shown that sentences ruled out by Binding Theory seem to be accepted by kids.
Do kids take a while to learn Binding Theory (even supposing it is learnable)?
When do they know it?

11. C. Chomsky (1969)
Tested Principle C with kids and proposed that kids go through three stages:
Stage 1.
Coreference is unconstrained.
Stage 2.
Linear order strategy for pronominalization (linear order; antecedent must precede pronoun)
Stage 3.
Principle C is obeyed.

12. C. Chomsky (1969) “He found out that Mickey won the race.”
“Who found out?”
Kid points to someone, maybe Mickey.
“After he found out, Mickey left.”
“Pluto thinks he knows everything.”
Stage 2: Some kids never picked Mickey.
Is backward pronominalization disallowed in these kids’ grammars?

13. Linear order strategy
Do kids go through a stage where they have a strategy for pronouns instead of Binding Theory?
Lust (1981): When asked to repeat, kids repeated forward pronominalizations much more accurately than redundant (name…name) sequences or backwards pronominalizations.

14. Linear order strategy
But this doesn’t tell us that there aren’t grammatical principles governing their use of pronouns and/or reflexives.
If it tells us anything, it only tells us that, of the grammatical options, forward pronominalization is preferred.

15. “Preference parameter”?
Lust in fact elevates this to the status of a parameter: head-final languages prefer backwards pronominalization, head-initial languages prefer forwards pronominalization.
Lust claimed there was a difference in preference between English and Japanese; O’Grady failed to replicate the difference between English and Korean.
This is not a good parameter anyway. Languages do not differ in what they allow, just in how much they like a type of sentence.

16. Truth Value Judgment
One way we can get a judgment (and not a preference) is with the truth value judgment task. Another advantage to the TVJ task is that it is not very cognitively taxing.
Reminder of the TVJ task:
Show the kid a little story.
A puppet says “I know what happened… X”.
At which point the kid either feeds the puppet a cookie or rag, depending on whether the puppet told the truth.

17. Crain & McKee (1985)
Crain & McKee (1985) tried again with Principle C, this time with a TVJ task, and found nothing particularly non-adult about kids’ use of Principle C. Not 100-0, but definitely systematic.
When hei was playing guitar, Pinocchioi was dancing. (73% yes; mean age 4;2)
When hei was playing guitar, Pinocchioj was dancing. (81% yes)
*Hei washes Goofyi. (88% no)

18. Principle C
Results about Principle C have been rather all over the map, but probably the appropriate synthesis of what’s out there is:
Kids know and obey the constraints of Principle C on their interpretations (from 3 or so).
Application of this knowledge in an experimental setting is highly dependent on the demands of the task and the context.

19. Onset of Binding Theory?
If Binding Theory is part of UG, not learned, we’d expect that kids start out already knowing it. (or maybe it matures, but let’s hold off on that possibility until we need it)
Caveat: Of course, the kids need to know what is a pronoun and what is a reflexive before they can use Binding Theory.
However: We expect to find that the first available evidence should show that kids know Binding Theory.

20. Onset of Binding Theory
But it doesn’t seem to turn out as we’d expect…
Several experiments seem to show that while kids show early evidence of knowing Principle A/C, they (appear to) consistently fail to observe Principle B—even up to (and beyond) 6 years old.

21. Chien & Wexler (1990)
Explored the question of whether kids know Principles A and B from the outset or not.
First three experiments show:
Kids correctly require local antecedents for reflexives (Principle A) early on
Kids are significantly delayed in requiring non-local antecedents for pronouns (Principle B).

22. C&W90: Experiment I
Tests Principle A (reflexives require a local antecedent) by providing sentences with two possible antecedents (one local, one not). “Simon says” act-out task. (156 kids, mean 4;6)
Kitty says that Sarah should point to herself.
Kitty says that Sarah should point to her.
Kitty says that Adam should point to her.

23. C&W90: Experiment II
Checking the effects of finiteness and also setting up a gender control on reflexives. (142 kids; mean 4;5)
Kitty wants Sarah to point to herself.
Kitty wants Sarah to point to her.
Kitty wants Adam to point to her
Snoopy wants Sarah to point to herself.

24. C&W90: Experiment III
Increased the number of conditions to test for pragmatic strategies and to replicate the results with a different task. (174 kids; mean 4;5)
(Previous task was “Simon [Snoopy/Kitty] says…”, this task was “Party game” which involved giving objects to people/puppets sitting at a table. This might, if anything, introduce a self-bias, because it’s fun to get toys. Kitty says that Sarah should give herself a car.).

25. C&W90: Experiments I-II
Kids from 2.5 to 6 showed a steady increase (from about 13% correct to about 90%) in requiring herself to take a local antecedent.
G1=2;6-3;0
G2=3;0-3;6 …
G8=6;0-6;6

26. C&W90: Experiments I-II
For some reason, kids seemed to perform better with nonfinite verbs (want); C&W have no particular explanation.

27. C&W90: Appendix I reflexives

28. C&W90: Experiments I-II
Kids showed no significant development in requiring her to take a non-local antecedent (about 75% across the board). Most of the errors treated her as taking a local antecedent.
Kitty says that Sarah should point to her.

29. C&W90: Appendix I pronouns

30. C&W90: Experiments I-II
Gender cues for non-local pronoun brought kids’ performance up to near-perfect. Had little effect on reflexives.

31. C&W90: Experiment III results
Previous results replicated for new task.
Young kids did better (operated at chance) for Principle A (meaning that they don’t have a systematic non-local coreference principle they are following—cf. Experiment I result showing them at 13% correct). Who knows what it was, but it wasn’t grammar.

32. C&W90: Possibilities so far…
Kids have to learn Principle B it takes a while.
But how on positive evidence alone?
Her is harder to learn than herself.
But kids use pronouns first (I saw him sentences indicate that they’re pronouns).
Principle B matures (constraints enforcing coreference before those prohibiting coreference?)
*UG-constrained maturation
“Principle B errors” aren’t Principle B problems.

33. Chien & Wexler (1990)
Kids do know the difference between pronouns and reflexives (they aren’t treating them all as reflexives).
E.g., I saw him, *I saw himself. Kids say sentences like I saw him often enough, but they do seem to know that reflexives need a local antecedent.

34. So what’s wrong with Principle B?
Chien & Wexler (1990): Nothing is wrong with Principle B. Kids know and respect Principle B all along.
Consider what adults can do:
That must be John—or at least he looks an awful lot like him.
So do adults violate Principle B?

35. Coindexation
Principle B says that coindexation between a pronoun and an antecedent is prohibited if the antecedent is too close.
Assuming adults obey this, that previous sentence must have been:
That must be John—or at least hei looks an awful lot like himj.
…where i and j are accidentally coreferent.

36. Coindexation
If two noun phrases share the same index, they necessarily share the same referent. Coindexation implies coreference.
If two noun phrases do not share the same index, does this mean they can’t share the same referent? Does contraindexation imply non-coreference?

37. Coindexation
The idea behind the Chien & Wexler account of the Principle B “delay” is that adults know the pragmatic Principle P, but kids are unable to use it right away.
Principle P Contraindexed NPs are non-coreferential unless the context explicitly forces coreference.

38. Coindexation So, when a kid agrees that…
Mama Bear is pointing to her.
…meaning ‘Mama Bear is pointing to herself’, what the kid really agreed to was
Mama Beari is pointing to herj.
…ok by Principle B, but violating Principle P (by allowing i and j both to refer to Mama Bear).

39. How could we ever tell?
But how can we tell if it’s Principle P that kids don’t obey and not Principle B, given that they both seem to allow Mama bear is pointing to her ‘…herself’?
Answer: Principle B also governs the use of bound pronouns, which Principle P has nothing to say about.

40. Bound pronouns A bound pronoun is like his in:
Every boyi is looking for hisi keys.
…and these are subject to Principle B, but they do not have a fixed referent, so accidental coreference is not an option here.
*Every boyi admires himi.

41. Prediction So, if found that kids accept …but refused to accept
Mama bear points to her (her = Mama Bear)
…but refused to accept
Every beari points to heri. (her = each bear in turn)
…then kids know Principle B (and what they lack is probably Principle P).

42. Chien & Wexler (1990)
First three experiments established that Principle B appears to be delayed with respect to Principle A.
Fourth experiment establishes that kids obey Principle B when coindexation would be forced by a bound variable interpretation.

43. C&W90: Experiment IV
Principle B (but not Principle P) applies also to bound pronouns—if the kids know Principle B and not Principle P, we expect to see kids getting bound pronouns right (unlike referring pronouns, as previous three experiments showed).

44. C&W90: Experiment IV items
Name-reflexive
Is Mama Bear touching herself?
Name-pronoun
Is Mama Bear touching her?

45. C&W90: Experiment IV items
Quantifier-reflexive
Is every bear touching herself?
Quantifier-pronoun
Is every bear touching her?

46. C&W90: Experiment IV controls
Name-name
Is Mama Bear pointing to Goldilocks?
Every-name
Is every bear pointing to Goldilocks?
All-name
Are all of the bears pointing to Goldilocks?

47. C&W90: Experiment IV control results
Kids under 5 did poorly on the mismatch (“no”) condition for every and all; they did less poorly on the mismatch condition for names.
Kids under 5 haven’t quite mastered quantifiers. (So we can’t test Principle B with them) (with this task)
G1=<4(48); G2=4-5(45); G3=5-6(44);G4=6-7(40)

48. C&W90: Experiment IV reflexive results
Kids over 5 did near-perfect with respect to Principle A (name-reflexive and quantifier-reflexive match/mismatch).

49. C&W90: Experiment IV name-pronoun
Kids did badly on the name-pronoun mismatch cases, steadily rising from about 70% wrong to about 25% wrong between 4 and 7.

50. C&W90: Experiment IV quantifier-pronoun
Under 5, kids were operating around chance (they don’t understand how quantifiers work yet)
Over 5, they were at 80% correct and above—in particular, better than on the name-pronoun condition; they seem to know Principle B.
(G3 went from 50% to 80%)

51. C&W90: Appendix I E4: name-pron & quant-pron

52. Chien & Wexler (1990) overall results
By the time kids understand quantifiers like every and all, pronouns, and reflexives, they apply Principle B.
Where accidental coreference is possible (despite violating Principle P), kids will allow it about half of the time.

53. Thornton & Wexler (1999)
What pragmatic knowledge do children lack? Broadly speaking, children appear to have difficulty evaluating other speakers’ intentions… As speakers, children fail to distinguish between their knowledge and that of listeners… [c]hildren use pronouns without first ensuring that a referent has been introduced into the conversational context… As listeners, children appear to assign interpretations to other speakers’ utterances that require special contextual support to be felicitous for adults… (pp )

54. Thornton & Wexler (1999)
Replicated Chien & Wexler (1990) and also tested VP ellipsis cases—another case where a pronoun can be bound (and so Principle B can be unambiguously tested).
Papa Bear wiped his face and Brother Bear did [wiped his face] too.
His = Papa Bear’s (strict—coreference)
His = Brother Bear’s (sloppy—bound)

55. Parallelism
VP ellipsis is subject to a parallelism constraint (parallelism between the overt and elided material). There are actually two parts to parallelism:
NPs in the elided and antecedent VP must
Both be bound variables or both be referential pronouns (structural parallelism)
If the pronouns are referential, they must have the same referent (referential parallelism).

56. Parallelism PB wiped his face and BB did [wiped his face] too.
his in the first clause is bound by PB. His in second must also be bound by the subject, there BB.
His in first clause is referential. It refers to GB. His in second clause must be referential, and must also refer to GB.
Kids are expected to obey structural parallelism; grammar (not pragmatics)

57. Four possible interpretations

58. Truth value judgment task
Experimenter 1 tells a story, moves the toys.
Experimenter 2 plays a puppet, who has to report what’s just happened.
The kid decides, based on whether the puppet told the truth about what happened, to either give the puppet a cookie or make it do pushups. If the puppet gets it wrong, the puppet asks the kid “What really happened?”
19 kids, 4;0 to 5;1

59. Replicating the basic result
Bert and 3 reindeer have a snowball fight and get all covered in snow. They go inside, Bert asks the reindeer to brush the snow off of him. 2 reindeer refuse, and commence brushing themselves off; the third helped a little, but mainly concentrates on brushing the snow off himself.
Every reindeer brushed him. (No: 92%) √G2
WRH? “Only one of them helped him”
Every reindeer brushed himself. (Yes: 88%) √G2
WRH? Other stuff too.
Bert brushed him. (No: 42%) (group 1: No)
Brushed hisself? Him? Wiped him? Bert??

60. Testing VP ellipsis
The caveman kissed the dinosaur and Fozzie Bear did too. (Correct: 100%)
IH brushed someone else’s hair, trolls brushed their own hair.
The Incredible Hulk brushed his hair and every Troll did too. (Yes[*SP]: 3%) √G2
WRH? Only the IH did. (First conjunct consistently controls structural parallelism).

61. Testing VP ellipsis
Lizard man and the ugly guy for some reason opt to lift up some other characters. Lizard man lifts the Smurf, ugly guy lifts Mickey.
The lizard man lifted him and the ugly guy did too. (No: 79%)
21% overriding referential parallelism? Pragmatic? W&T say “probably”.
Same kids as allow MB points to her[=MB]? Well, a subset. Pattern was: every kid who allowed ref. parallelism violations alowed MBpth, but not vice-versa.

62. Testing Principle B
Everyone is covered with glitter, Batman and 2 turtles refuse to help Smurf out because they are cleaning themselves. One turtle briefly helps Smurf, but then returns to cleaning himself.
Batman cleaned him and every turtle did too. (No: 86%)
Batman cleaned himself and every turtle did too. (Yes: 95%)
Kids can accept a sloppy reading when √Pr.B.

63. Testing Principle C He dusted the skeleton. (No: 92%)
The kiwi bird cleaned Flash Gordon and he did too. (No: 54%—adults 83%!)
What’s going on? Stress (even implicit due to the ellipsis)? ?

64. Thornton & Wexler (1999)
Conclusions: Kids seem to know and obey Principle B, Principle C, and structural parallelism.
Kids seem to have more trouble with referential parallelism and the contexts for constructions of “guises”.

65. However. Elbourne (2005). Compare: Every reindeer brushed him.
Well, the whole story revolves around Bert and his efforts to get the snow off himself. So, seems like Bert is a pretty good guess.
These are the bears; this is Goldilocks. Is every bear touching her?
There seems to be a confound with salience.

66. Grolla (2005/6) *Goofy brushed him
Cf. Goofy brushed himself
*This is the elephant that he’s skating
Cf. This is the elephant that’s skating
John loves his dog
Cf. *John loves hisself dog (?)
This is the frog [that the swan laughed [when he fell]
Cf. *This is the frog [that the swan laughed [when _ fell]
The game’s the thing.
Chance on 1-2, high acceptance for 3-4. Only chance on: Every spider washed her. (Vs. C&W90 84% reject). Blame: Salience. Cf. Is every dog touching her hat? No: 70%.

67. 
  
 
 

68. Preparatory comments for next week’s readings
Borer & Wexler (1987): They are out to challenge the idea that kids start off with the entire grammatical system, in favor of a maturational view of syntactic development.
Their famous proposal is that children cannot initially construct “A-chains”, and they use evidence primarily from passives.

69. Passives John kicked the ball (active)
The ball was kicked (by John) (passive)
Standard analysis: the ball starts off as complement of V in both; in the passive, the agent is suppressed and the verb is deprived of its ability to assign Case. Thus, the ball moves into SpecIP to get Case.
The balli was kicked ti.

70. Passives
The chain between the ball and t created by moving the ball into SpecIP is an A-chain (a chain whose top is in a position where you can only find arguments).

71. Intransitives There are two kinds of intransitive verbs:
Unergative
Unaccusative (or sometimes “ergative”)
The unergative verbs have an external argument— just like a transitive verb.
The unaccusative/ergative verbs have only an internal argument, which moves to subject position—just like in a passive.

72. Unaccusatives ≈ passives
An unaccusative is structurally like a passive:
The traini arrived ti.
An unergative is not.
The baby giggled.
So we expect kids to have the same troubles with unaccusatives and passives.

73. Verbal and adjectival passives
In English at least, it seems like there are two kinds of words with passive morphology:
Verbal: The suspect was seen.
Adjectival: His hair seems combed.
Borer & Wexler adopt an analysis under which adjectival passives do not involve syntactic movement (lexicon vs. syntax).

74. Verbal and adjectival passives
Generally, non-action verbs make poor adjectival passives (while action verbs are fine):
*The suspect seems seen. The seen suspect (fled). Seen though the movie was, John went to see it again.
The cloth seems torn. The torn cloth (is useless). Torn though the cloth was, John used it anyway.
Conclusion: It should be possible for kids to say passive-like things as long as they’re adjectival passives.

75. Babyonyshev et al. (1998)
Babyonyshev et al. (1998) extend the discussion begun in Borer & Wexler (1987), also arguing for maturation of A-chains.
They consider two possible reasons why A-chains in passives would not be allowed:
Kids can’t build A-chains.
Kids can’t “dethematize” the external argument.

76. UTAH
The Uniformity of Theta Assignment Hypothesis (UTAH) essentially says that the syntactic position in the structure to which any given q-role is assigned does not vary within or across languages.
So, the patient q-role is always assigned to the complement of V position, for example.

77. Pesetsky and movement
Languages can differ in whether they perform overt movement (before SS) or covert movement (after SS, headed to LF).
Usual example: Wh-movement (Bulgarian: all wh-movements overt; English: one overt wh-movement, the rest covert; Japanese: all wh-movements covert).

78. Pesetsky and movement
If we assume that all languages move all of their wh-words to (Spec)CP by LF (only some languages save some/all of these movements until after SS), then at LF there is always a chain like:
Wh-wordi …ti .
One way to think of “covert movement” is as “pronouncing the bottom of the chain” (in a model in which you both interpret and pronounce LF).

79. Pesetsky and movement
This idea of “pronouncing the bottom of a movement chain” comes up in part of the discussion in Babyonyshev et al. concerning pronunciation in A-chains (like those in unaccusatives and passives) as well as A-bar chains (like wh-movement chains).

80. Babyonyshev et al. (1998)
Babyonyshev et al. conduct an experiment with Russian kids to determine whether kids who cannot represent adult unaccusatives (due to the inability to represent A-chains) instead parse them as unergatives.
“S-homophone”: A different syntactic structure (e.g. an unergative) which sounds like another (e.g. an unaccusative).

81. Russian genitive of negation
There is a fairly elaborate discussion of the “genitive of negation” construction in Russian. Basically, a non-specific noun phrase in the same clause as negation will be pronounced with genitive (instead of accusative) case. Some verbs (e.g., existential be) in fact require genitive.

82. Russian genitive of negation
There is evidence that the genitive argument of an unaccusative remains inside the VP at SS.
In English, this argument would raise to subject position (SpecIP).
In Russian, it turns out that there is evidence that the genitive argument raises covertly (between SS and LF) to subject position.

83. Evidence for covert movement of the genitive argument
Negative constituents (e.g., any kind of boy) need to co-occur with negation in the same clause.
Where negative constituents participate in A-chains we can see (e.g., raising), the top of the A-chain has to be in the same clause as negation.
Genitive negative constituents with raising verbs appear in the lower clause at SS but require negation in the higher clause.
Conclusion: Genitive arguments move too, creating an A-chain, and the negation requirement is verified at LF.

84. Babyonyshev et al. (1998)
Testing the idea from Borer & Wexler (1987) that unaccusatives are analyzed as if they are unergatives by kids in the pre-A-chain stage of life.
Turns out that Russian provides a nice test of unaccusativity/unergativity with the “genitive of negation” so we can directly check to see how kids are analyzing their intransitives.

85. 
  
 
 