The Morphosyntax of Upward Agreement and Downward Agreement · Running time: Number of basic steps...

transcript

The Morphosyntax of Upward Agreement andDownward Agreement

Introduction, Part IV: Differences between Upward Agreement and

Downward Agreement

Anke Himmelreichanke.assmann@uni-leipzig.de

Universitat Leipzig, Institut fur Linguistik

16 November, 2017

1 / 65

Table of Contents

1 Summary of Part III

2 Cases of Agreement

3 Possibilities for AnalyzingInstances of Agreement

4 Choosing between thePossibilities

5 Summary of Part IV

2 / 65

Table of Contents

1 Summary of Part III

3 / 65

Conceptual Aspects

Conceptual Aspects = ComplexityComplexity = running time / memory spaceRunning time: Number of basic steps depending on the size ofthe input.Memory space: Maximal memory space consumption in the timeof the application of the algorithm depending on the size of theinput.

4 / 65

Conceptual Aspects

Conceptual Aspects = ComplexityComplexity = running time / memory spaceRunning time: Number of basic steps depending on the size ofthe input.Memory space: Maximal memory space consumption in the timeof the application of the algorithm depending on the size of theinput.

4 / 65

Linguistic Theories as Algorithms

ProblemHow do we get a specific linguistic expression?

Algorithm:1. Apply rule Ti to get representation r1.2. Apply rule Tj to get representation r2.3. Apply rule Tk to get representation r3.

...n. Apply rule Tz to get representation rn.

Input: A representation, i.e., a structure, (that is evaluated ortransformed), the size of which might be the number of elementsin the representation or the number of relations (orderedpairs/n-tuples of elements)Basic step: A minimal transformation/operationMemory space: Number of elements in the representation thatone needs to look at at one point in the algorithm

5 / 65

Direction Condition as an Algorithm

C-command as an algorithmC-command relations between X and Y can be read off of a syntactictree. If trees are displays of a derivation and derivations arealgorithms, then c-command can be understood as the part of thederivation that have got X and Y into there positions.

ProblemFinding out if two nodes X and Y in a tree are in a c-commandrelation. (C-command (X,Y)?)Finding a node Y that is in a c-command with a given node X.(Search-Goal (X))

6 / 65

Worst Case Complexity and Big O

Worst-case complexity tells us something about the maximum ofresources we should plan in.When it comes to comparing (worst-case) complexity, it is moreinteresting to identify to which complexity group an algorithmbelongs than to simply compare the complexity, because the goalis to classify algorithms.The Big O notation is the notation for the growth rate (or order ) ofa function, i.e., it says something about how much complexityincreases if the size of the input increases.Since the size of the input approaches infinity, concrete constantsand coefficients can be neglected.

7 / 65

Os and the Direction Condition

The most interesting part of Agree, when it comes to growthorders, is determining the goal.The worst complexity has an Agree algorithm that uses sequentialsearch (O(n)).This is independent of whether the tree is (primarily) searchedupwards or downwards.

8 / 65

So what?

Assuming that complexity is the main design aspect under which atheory is evaluated, there is no general preference of one Agreedirection over the other.Thus, only empirical arguments do really matter.

9 / 65

Table of Contents

2 Cases of AgreementCases of Downward AgreementCases of Upward AgreementCases of Bidirectional AgreementComparison of the Cases

10 / 65

Table of Contents

11 / 65

φ-Agreement

Hindi (Bhatt (2005, 775), Boeckx (2004, 26))

(1) a. Vivek-neVivek-ERG

kitaabbook.F.SG

parh-niiread-INF.F.SG

chaah-ii.want-PERF.F.SG

‘Vivek wanted to read the book.’b. Mona

Monakutto-kodog.M.PL-ACC

dekh-naa/*niisee-INF/*INF.F.SG

chaah-tiiwant-HAB.F.SG

thii.be-PAST.F.SG‘Mona wanted to see the dogs.’

12 / 65

Table of Contents

13 / 65

Negative Concord

(2) a. ?John didn’t eat nothing.b.??Nobody didn’t eat.

Blanchette (2016), Zeijlstra (2004)

14 / 65

Tense Concord

(3) a. John said Mary was illb. Jan

Johnzeisaid

datthat

MarieMary

ziekill

waswas

‘John said Mary was ill’

Zeijlstra (2012)

15 / 65

Binding and Co-Reference

(4) a. She likes herself.b. He likes himself.c. She likes her.d. You like yourself.e. They like themselves..

(5) A: I met the most fascinating woman yesterday.B: Oh yeah? Who was she/*he?

Preminger and Polinsky (2015)

16 / 65

Table of Contents

17 / 65

Himmelreich (2017b)

Phenomenon?

How does Agree work?

What is the probe?

18 / 65

Assmann et al. (2015)

Phenomenon?

How does Agree work?

What is the probe?

19 / 65

φ-Agreement with Vs, Ns, As is Agree I

Agreement Asymmetries between verbs, nouns, and adjectives(Swahili, (Baker, 2008, 1f))

(6) a. Ni-li-kuwa1SS-PAST-be

ni-ki-som-a.1SS-CONT-read-FV

‘I was reading.’b. Ni-∅

1SS-bem-refu.CL1-tall

‘I am tall.’c. Ni-li-po-kuwa

1SS-PAST-when-beki-janaCL7-child

...now

sasa1SS-be-when

ni-li-poCL1-man

m-tuCL1-whole

m-zima, ...

‘When I was a child ... Now that I am a man ...’

20 / 65

φ-Agreement with Vs, Ns, As is Agree II

Baker (2008, ch.2):Universal bidirectional Agree at first glance

(7) A functional head F agrees with XP, XP a maximal projection,only if:a. F c-commands XP or XP c-commands Fb. There is no YP such that F c-commands YP, YP

c-commands XP, and YP has φ-featuresc. F and XP are contained in all the same phases (e.g., full

CPs)d. XP is made active for agreement by having an unchecked

case feature

21 / 65

φ-Agreement with Vs, Ns, As is Agree III

(8) a. Verbs are lexical categories that license a specifier.b. Nouns are lexical categories that have a referential index.c. Adjectives are lexical categories that have neither a

specifier nor a referential index.

(9) Any lexical category can be immediately dominated by theprojection of a functional head that matches it in grosscategorical features. Functional heads, unlike lexical heads, canmanifest agreement.

(10) The Structural Condition on Person Agreement (SCOPA)A functional category F can bear the features +1 or +2 if andonly if a projection of F merges with an NP that has thatfeature, and F is taken as the label for the resulting phrase.

22 / 65

Parametrization I

(Baker, 2008, 215)

(11) The Direction of Agreement Parametera. F agrees with DP/NP only if DP/NP asymmetrically

c-commands F, orb. F agrees with DP/NP only if F c-commands DP/NP, orc. F agrees with DP/NP only if F c-commands DP/NP or vice

versa.

23 / 65

Parametrization II

(12) a. On the table were/*was (put) some peanuts.b. On the table was/*were (put) a peanut.

(Kinande, Baker (2003))

(13) a. Omo-mulongoLOC.18-village.3

mw-a-hik-a18S-T-arrive-FV

mukali.woman.1

‘At the village arrived a woman.’b. Oko-mesa

LOC.17-tablekw-a-hir-aw-a17S-T-put-pass-FV

ehilanga.peanuts.19

‘On the table were put peanuts.’

(Burushaski, Willson (1996, 3))

(14) a. Dasıngirl(ABS)

ha-ehouse-OBL

mo-yan-umo.3SO.F-sleep-3SS.F/PAST

‘The girl slept in the house.’b. Dasın

girl(ABS)ha-ehouse-OBL

huruT-umo.sit-3SS.F/PAST

‘The girl sat in the house.’

24 / 65

Table of Contents

25 / 65

Possible Differences between Upward and Downward Agree

Type of FeatureDownward Agree Upward Agreeφ case

tensenegationindices(?)

LocalityDownward Agree Upward Agree

within phrase (spec-head)within clause within clauselong-distance long-distance

Categories of Probe and GoalDownward Agree Upward Agreeverbal verbal

nominal

26 / 65

Table of Contents

3 Possibilities for Analyzing Instances of AgreementInteraction of Movement and AgreementIndirect AgreementNo Agreement

27 / 65

Table of Contents

28 / 65

Change the position of the goal

(15) XP

... α ...

29 / 65

Change the position of the probe

(16) XP

(see Preminger and Polinsky (2015))

30 / 65

Table of Contents

31 / 65

Agree with lower element

(17) XP

... β ...

32 / 65

Agree with higher element

(18) XP

... α ...

33 / 65

Table of Contents

34 / 65

Matching is due to process other than Agree

What are alternative processes?In what sense are they different?

35 / 65

Table of Contents

4 Choosing between the PossibilitiesYes to Downward AgreementYes to Upward AgreementNo to Downward AgreementNo to Upward AgreementChoosing the Right Version of the Direction Condition

36 / 65

Table of Contents

37 / 65

LDA in Tsez

(19) a. Eni-rmother-DAT

[ uziboy.I.ABS

F-ay-ru-łii-arrive-PST.PRT-NMZL

F-iy-xo.I-know-PRES‘The mother knows that as for the boy, he arrived.’

b. Eni-rmother-DAT

[ uz-aboy-ERG

magalubread.III.ABS

b-ac’-ru-łiIII-eat-PST.PRT-NMZL

] b-iy-xo.III-know-PRES

‘The mother knows that as for the bread, the boy ate it.’

QuestionCan this be upward agreement?Can this not be agreement?

Polinsky and Potsdam (2001); Preminger and Polinsky (2015)38 / 65

Table of Contents

39 / 65

Sequence of Tense

(20) a. John said Mary was illb. Jan

Johnzeisaid

datthat

MarieMary

ziekill

waswas

‘John said Mary was ill’

QuestionCan this be downward agreement?Can this not be agreement?

Zeijlstra (2012)

40 / 65

What does BiAgree has to say about LDA and SoT?

41 / 65

Table of Contents

42 / 65

Berber I

(21) a. zri-nsaw-3PL

imhdarnstudents

MohandMohand

‘The students saw Mohand.’b. man

whichtamghartwoman

yzrin/*t-zrasee.PARTCP/3SG.FEM-saw

MohandMohand

‘Which woman saw Mohand?’c. man

whichtamghartwoman

nna-nsaid-3PL

qathat

t-zra3SG.FEM-saw

MohandMohand

‘Which woman did they say saw Mohand?’

Ouhalla (1993), Georgi (2014)

43 / 65

Analysis Georgi (2014)

(22) a. CP

... twh ...

44 / 65

Contra Movement I

(23) absence of island effects (Elouazizi (2005, 126))a. sqssa-n

ask.PERF-3PL.MAASC

[ mawhether

y-wSa3SG.MASC-give.PERF

JamalJ.

lktabbook

w-arbaCS-boy

‘They asked whether Jamal gave the book to the boy.’b. man

whichlktab2book

ixefabout

sqssa-nask.PERF-3PL.MASC

[ mawhether

Jamal1J.

*(T)it

y-wSi-nPRT-give.PERF-PRT

t1 t2 ito

w-arbaCS-boy

‘Which book did they wonder whether it is Jamal whogave it to the boy?’

45 / 65

Contra Movement II

(24) absence of weak crossover effects (Elouazizi (2005, 127))

(D)COP

[ ymasmother-his

kurevery

arba]jboy

i gwho

[vP i-texsse-nPRT-love.IMPERF.PRT

tjson-his/her

memi-sj ]

‘It is the mother of every boy who loves her son.’

46 / 65

Contra Movement III

(25) absence of reconstruction (Elouazizi (2005, 127))a. D

ixefinessi/jhimself

ixefabout

y-siwrr3SG.MASC-talk.PERF

MuhandjMuhand

agwith

OmariOmar‘It is himself that Muhand talked with Omar about.’

b. DCOP

MuhandjMuhand

i gwho

y-siwrr-nPRT-talk.PERF-PRT

agwith

OmariOmar

xabout

ixefinessi/∗jhimself‘It is Muhand who talked with Omar about himself.’

47 / 65

Berber with Upward Agree (Himmelreich (2017b))

(26) a. CP

... twh ...

wh[φ ≺ wh]

... twh ...

wh[φ ≺ wh]

48 / 65

Table of Contents

49 / 65

Standard Arabic I

(27) a. Pal-Pawlaad-uthe-children-NOM

naam-uu/*naam-a.slept-3PL/*slept-MASC.3SG

‘The children slept.’b. naam-a/*naam-uu

slept-MASC.3SG/*slept-3PL

l-Pawlaad-u.the-children-NOM

‘The children slept.’

Aoun et al. (1994), Bjorkman and Zeijlstra (2014)

50 / 65

Standard Arabic II

(28) a. t-taalibaat-uthe-student.FEM.PL-NOM

Pakal-na/*Pakal-atate-FEM.3PL/*ate-FEM.3SG

‘The students ate.’b. Pakal-at/*Pakal-na

ate-FEM.3SG/*ate-FEM.3PL

t-taalibaat-uthe-student.FEM.PL-NOM

‘The students ate.’c. naam-uu/*naam-a

slept-3pl/*slept-M.3sghum.they

‘They slept.’

Benmamoun and Lorimor (2006), Preminger (2015)

51 / 65

Analysis based on Preminger (2015) I

(29) Partial Agreement in VS-structuresa. [ ate[π: ≺ #: ≺ gen: ] ... [Intervener student[π:3, #:pl,

gen:fem] ]]b. [ ate[π: ≺ #: ≺ gen: ] ... [Intervener student[π:3, #:pl,

gen:fem] ]]c. [ ate[π: ≺ #: ≺ gen: ] ... [Intervener student[π:3, #:pl,

gen:fem] ]]d. [ ate[π: ≺ #: ≺ gen:fem] ... [Intervener student[π:3, #:pl,

gen:fem] ]]e. [ ate[π:def=3 ≺ #:def=sg ≺ gen:fem] ... [Intervener

student[π:3, #:pl, gen:fem] ]]

52 / 65

Analysis based on Preminger (2015) II

(30) Full Agreement in SV-structuresa. [ ate[π: ≺ #: ≺ gen: ] ... [Intervener student[π:3, #:pl,

gen:fem] ]]b. [ ate[π: ≺ #: ≺ gen: ] ... student[π:3, #:pl, gen:fem] ...

[Intervener tstudent ]]c. [ ate[π:3 ≺ #:pl ≺ gen:fem] ... student[π:3, #:pl, gen:fem]

... [Intervener tstudent ]]d. [ student[π:3, #:pl, gen:fem] ate[π:3 ≺ #:pl ≺ gen:fem] ...

t′student ... [Intervener tstudent ]]

53 / 65

What does BiAgree have to say about AAE and PA?

54 / 65

Table of Contents

55 / 65

Falsification

A major criteria of theories is that they are falsifiable: A theoryshould be formulated in a way that it can easily be falsified.Note: Verification of theories is never possible!Falsification: Find one empirical phenomenon that the theoryshould cover, but does not cover.Verification: Test the theory on all empirical phenomena that itshould cover.

56 / 65

Three versions of the direction condition I

1 Downward Agree: The probe has to c-command the goal.Falsified by: A case where the goal c-commands the probe.Caveat: Falsifying data can be reanalysed as an interaction of Moveand Agree, with the goal starting out in the c-command domain ofthe goal and than moving up to a position higher than the goal.Caveat to Caveat: Test whether movement plays a role.

57 / 65

Three versions of the direction condition II

2 Upward Agree: The goal has to c-command the probe.Falsified by: A case, where the probe c-commands the goal.Caveat: Falsifying data can be reanalysed as agreement betweenthe probe and and another (perhaps mediating) element that ishigher than the probe.Caveat to Caveat: Test whether there is reason to propose such amediator.

58 / 65

Three versions of the direction condition III

3 Bidirectional Agree: It must be the case that the probec-commands the goal or the goal c-commands the probe.

Falsified by: A case, where c-command between probe and goal isgiven but no Agree applies.Caveat: Falsifying data can be reanalysed as a conspiracy of other(language-specific) factors that rule out Agree.Caveat to Caveat: Test whether the factors are relevant.

59 / 65

Construing arguments

BiAgree

DAgree UAgree

60 / 65

Table of Contents

5 Summary of Part IV

61 / 65

Cases of Agreement

Apparent empirical differences between Downward, Upward andBidirectional Agree:

Downward: Almost exclusively φ-agreementUpward: Often phenomena other than φ-agreementBidirectional: Often Case (phenomena where intuitively bothelements are dependent)

62 / 65

Analyzing Cases of Agreement

For all empirical cases, all analytical possibilities can be used.Very often, it’s hard to distinguish empirically between thepossibilities.In order to distinguish them, it is useful to check for:

The existence of movementThe existence of covert elementsSpecial morphological markers signalling elements or relations

63 / 65

References I

Aoun, Joseph, Elabbas Benmamoun and Dominique Sportiche (1994): ‘Agreement, Word Order, and Conjunction in SomeVarieties of Arabic’, Linguistic Inquiry 25, 195–220.

Assmann, Anke, Doreen Georgi, Fabian Heck, Gereon Muller and Philipp Weisser (2015): ‘Ergatives Move Too Early: On anInstance of Opacity in Syntax’, Syntax 18(4), 343–387.

Baker, Mark (2003): Agreement, dislocation, and partial configurationality. In: A. Carnie, H. Harley and M. A. Willie, eds, Formalapproaches to function in grammar. Amsterdam edn, John Benjamins, pp. 107–134.

Baker, Mark C. (2008): The Syntax of Agreement and Concord. Cambridge University Press.

Benmamoun, Elabbas and Heidi Lorimor (2006): ‘Featureless expressions: when morphophonological markers are absent.’,Linguistic Inquiry 37, 1–23.

Bhatt, Rajesh (2005): ‘Long distance agreement in Hindi-Urdu’, Natural Language and Linguistic Theory 23, 757–807.

Bjorkman, Bronwyn and Hedde Zeijlstra (2014): Upward Agree is superior. Ms., lingbuzz/002350.

Blanchette, Frances (2016): ‘Subject-Object Asymmetries in English Sentences with Two Negatives’, Penn Working Papers inLinguistics 22(1), 41–50.

Boeckx, Cedric (2004): ‘Long-distance agreement in Hindi: some theoretical implications’, Studia Linguistics 58, 23–36.

Elouazizi, Noureddine (2005): Anti-Agreement Effects as (Anti-)connectivity. Proceedings of WCCFL 24.

Georgi, Doreen (2014): Opaque Interactions of Merge and Agree: On the Nature and Order of Elementary Operations. PhDthesis, Leipzig University, Leipzig.

Himmelreich, Anke (2017a): Big Os in Linguistics. Ms., Leipzig University.

Himmelreich, Anke (2017b): Case Matching Effects in Free Relatives and Parasitic Gaps: A Study on the Properties of Agree.PhD thesis, Leipzig University.

Ouhalla, Jamal (1993): ‘Subject-Extraction, Negation and the Antiagreement Effect’, Natural Language and Linguistic Theory11, 477–518.

Polinsky, Maria and Eric Potsdam (2001): ‘Long-Distance Agreement and Topic in Tsez’, Natural Language and Linguistic Theory19, 583–646.

64 / 65

References II

Preminger, Omer (2015): Upwards and onwards: On the direction of valuation in phi-feature agreement. Talk at New YorkUniversity.

Preminger, Omer and Maria Polinsky (2015): Agreement and semantic concord: a spurious unification. Ms., lingbuzz/002363.

Willson, Stephen (1996): ‘Verb agreement and case marking in Burushaski’, Work Papers of the Summer Institute of LinguisticsNorth Dakota 40, 1–71.

Zeijlstra, Hedde (2004): Sentential negation and negative concord. PhD thesis, University of Amsterdam.

Zeijlstra, Hedde (2012): ‘There is only one way to agree’, The Linguistic Review 29, 491–539.

65 / 65

The Morphosyntax of Upward Agreement and Downward Agreement · Running time: Number of basic steps...

Documents