A well known problem…BBRRAACCKKEETTIINNGG PPAARRAADDOOXXEESS NELS
35. UConn. 2004. Heather Newell. McGill University.
[email protected]
A well known problem…. and a proposal.
Words like unhappier or nuclear physicist or ungrammaticality
embody a phenomenon that has received much discussion in the
literature (Pesetsky (1979,1985), Sproat (1988, 1985), Stump
(1991), Nissenbaum(2000), Hoeksema(1987), Light (1993), Spencer
(1988) etc..), that of the Bracketing Paradox. The defining
property of bracketing paradoxes is that the structure implied by
the morpho-phonological properties of the construction are at odds
with the structure implied by the semantic scope of a suffix/prefix
pair. What I will argue here (c.f. Nissenbaum 2000) is that these
paradoxes can be solved by appealing to the independently motivated
mechanism of Late Adjunction. Late Adjunction, in conjunction with
the Distributed Morphology (Halle & Marantz (1994), Marantz
(1997)) maxim of Syntax All The Way Down leads us to expect late
adjunction to occur within words.
Late Adjunction…
Late Adjunction, first proposed by Lebeaux (1988) and later by
others, has been argued to account for the adjunct/complement
asymmetry with regards to Condition C effects, VP adverbs in
English (Ochi 1999), Parasitic Gaps (Nissenbaum 2000)….. Condition
C 1) a. *Shei wants the picture of Seonaidi.
b. *Which picture of Seonaidi does shei want? 2) a. *Shei wants the
picture that Seonaidi likes.
b. √Which picture that Seonaidi likes does shei want? The (a)
examples show typical Condition C violations. In (1b) A′ movement
does not rescue the sentence, yet in (2b) it does. This is argued
to be due to the fact that the argument of Seonaid, but not the
adjunct that Seonaid likes, must be merged with picture before
movement. The adjunct is merged after movement, and therefore
Seonaid in (2b) is never c-commanded by she. In (2a) the adjunct is
also merged late, but to a position c-commanded by the pronouns,
causing ungrammaticality.
Morphological Late Adjunction Let us assume that the
above-mentioned argument gives us ample reason to believe in late
adjunction. Given the premise of Distributed Morphology, that
morphemes are concatenated in the syntax in a manner
indistinguishable from that of phrases, we should expect to find
word internal anti-cyclic merger of adjuncts. What is a
Morphological Adjunct? For the purposes of this paper I will assume
that those morphemes that have no Category Features to project are
Adjuncts. Examples from English can be seen in (3). Examples of
morphemes that cannot be adjuncts are given in (4).
(3) unhappy, reapply, misalign, nuclear physicist, up chuck….. (4)
enrage, destruction, refusal, man eater, happier…..
The bolded morphemes in (3) do not check any argument features, are
not assigned case, and project no features to the root node. The
bolded morphemes in (4) either project category features, or are
arguments.
Some Non-Adjunct Morphemes Some morphemes appear to be adjuncts,
but are not….. IN vs. UN In does not behave as an adjunct should.
It has a phonological closeness to the root, it bleeds the –er
comparative, and it is only found on adjectives. In DM terms, the
difference is captured by appealing to root vs. word level
morphology. Alternately, in- is the realization of an a0 head IN,
not UN, is phonologically ‘close’ to the root. (5) intolerable vs.
impolite untrue vs. unpopular IN, not UN, bleeds the comparative
suffix. (6) politer *impoliter (anonymous reviewer) happier
unhappier IN, not UN, is retricted to adjectival
environments.
(7) inept vs. *inaid unattractive vs. undo
UN, not IN, is a morphological adjunct. (8) a (9) a or a 2 2 2 a √
un a a √ in polite 2 2 a √ un √ happy happy Most prefixes are
adjuncts, but some are not….prefixes that project category feature
are not adjuncts. EN-, BE- are not adjuncts. They project and are
therefore merged cyclically. (10) v (11) v 2 2 en a be n g g able
jewel
Linear Edge Condition (Nissenbaum 2000)
Both Syntactic (phrasal) and Morphological Adjuncts are constrained
by the LEC.
Linear Edge Condition (LEC): For any syntactic object SO accessed
in an array, merge of new material is possible inside SO only at
the linear edge. (Nissenbaum 2000: 201)
Here the Linear Edge is assumed to be defined over the last phase
sent to PF. Syntactic Adjunction and the LEC: “…some element
(either overt or covert) at the right periphery of a vP blocks the
appearance there of an extraposed relative clause.” (Nissenbaum
2000:207)
(12) a. [vP[vP I saw a man yesterday][DPa [NP[NPman] [who was very
tall]] ]
b. *[vP[vP I saw a man yesterday] [DP a [NP very tall [NP man]] ]
(Nissenbaum 2000:209) Morphological Adjunction and the LEC: Some
element, here overt, at the periphery of an X0, blocks the merger
of a morphological adjunct, the particle up.
(13) a. * n n 2 2 v er v er 2 2 throw up up chuck (c.f. Nissenbaum
2000:206)
Spell Out: Derivation vs. Inflection
Throwing up (V) vs. Thrower up/Throw upper (N)
Claim: (1)Derivational or ‘little x0’ heads trigger Phases (Marvin
20021). The heads of these phases are sent to PF along with their
complements. (the strong vP phase (as well as DP/CP) is not a
member of this group)
(2) (strong) Inflection triggers raising. Derivation does not –
derivational morphemes undergo morphological merger.
(14) a. Seonaid is eating up her broccoli. (15) a. *Seonaid is an
eater up/eat upper of broccoli.
1 Marvin explicitly claims that even derivational heads spell out
only their complements. The facts are more complicated, as she
admits, and will be discussed further below.
(16) b. Part (17) b. n 2 2 Part VP n VP 2 g er g eat ing v v 2 2 v
√ v √ 2 2 eat up * eat-er up
In (16b) the particle adjoins in accordance with the LEC. In (17b)
either the adjunct or the lowered morpheme is in violation of the
LEC/its selectional restrictions. The lexicon must encode the
idiomatic reading of ‘eat up’, but the intervening derivational
morphology impedes access to this stored form. Alternately, the
adjunct is merged early, but then the nominalizer is violating the
LEC by lowering to the position between the verb and the particle.
This begs the question of why *eat upper is bad. I propose here
that the former account of *eater up is correct, and that late
adjunction of the particle is mandatory (see Stepanov 2001).
Derivational Phase Heads, and Spell Out…
What Leads Us To The Conclusion That The Heads Of Weak Phases Are
Spelled Out With Their Complements?
1. The Phonology… Marvin (2002) shows that the weak phase head
needs to be phonologically present in some manner when its
complement is spelled-out. Marvin’s Word-Relativised PIC: “ H and
its edge are spelled out at the next (strong) phase. The domain of
H is spelled out at the phase of HP. A head h adjoined to H is in
the domain of H.” In the following example (Marvin 2002:53), the
phase head al spells out its complement.
(18) line 1 * * line 0 (* * (* ] * govern ment al góvernmént
góvernméntal(next phase)
In (18) the RLR Edge Marking and Main Stress Rules of Halle (1998)
are used to explain the phonological behaviour of governmental.
These phonological rules must have access to the phonological shape
of the phase head, indicating that it has been sent to the MS/PF
components2. If this were not the case, we would expect the
following unattested form, as line 1 and 2 markings are subject to
the PIC.
(19) line 1 * line 0 (* * * govern ment góvernment
góvernmental
2. The Morphology…
Strong Phase Heads (e.g. v(oice)0) have the following properties.
i) They may move out of their phase.
(20) Ils onti souvent [vP ti manger]. (this does not necessarily
force the conclusion that these heads are not spelled out with
their phase (c.f. Fox and Pesetsky (2004))
2 That a phase head may be influenced phonologically from outside,
e.g. governmentálity, indicates that the facts are yet more complex
than how they are presented here.
ii) They may show allomorphy sensitive to more peripheral morphemes
which
are outside of their phase.
(21) a. He [PRES] [vP goes to the store]. b. He [PAST] [vP went to
the store].
(If the verb were spelled out in the vP phase, it would not be able
to show allomorphy conditioned by the features of the T0 head,
without the assumption that lexical items are merged with
phonological/tense features in the syntax)
Weak Phase Heads (e.g. n0, a0, v0) do not have the above
properties. (22)*Tioni was given a [motivate ti] intended: A
motivation was given. (23) [[[motiv√]atv]orn] but
*[[[motiv√]izev]ingv] where v0 shows sensitivity to the syntactic
category features of the outer morpheme (that I am aware of). This
lack of derivational outwards-sensitivity is unexpected if the
derivational heads are not spelled-out with their complements,
although this could be an accidental gap. (On outward sensitivity
see Bobaljik (2000) and references therein.)
AAAllllll BBBrrraaaccckkkeeetttiiinnnggg
PPPaaarrraaadddoooxxxeeesss CCCooonnntttaaaiiinnn AAA
MMMooorrrppphhhooolllooogggiiicccaaalll LLLaaattteee
AAAdddjjjuuunnncccttt...
(24) UNHAPPIER contains UN (25) NUCLEAR PHYSICIST contains NUCLEAR
(26) UNGRAMMATICALITY contains UN (although not a telling example,
c.f.
Halle & Vergnaud 1987) The morpho-phonological requirements
force us to assume the following structures: 2 2 2 un 2 nuclear 2
un 2 happy er physic ist grammatical ity The semantic requirements
force us to assume the following structures: 2 2 2 2 er 2 ist 2 ity
un happy nuclear physic un grammatical
A happier unhappier derivation… Numeration 1 (NS) a <happy,
a> 2 a happy
(MS/PF) HAPPY
Numeration 2 (NS) X <X(a phase head), Deg> 2 This is assuming
that the X Deg Degree head is not a phase 2 head. Deg a 2 a
happy
(MS/PF) Deg Undergoes Morphological Merger, as its complement is of
the right phonological shape (Embick & Noyer 2001) The -er
allomorph is inserted and spelled-out.
HAPPIER Numeration 3 (NS) Deg <un…> 2 Where … indicates er a
further lexical items, 2 if any. a √ 2
(MS/PF) un happier UNHAPPIER
N.B. If un- were to be numerated in the previous phase, the
environment for –er would be bled, and we get the also grammatical
‘more unhappy’.
A note onUngrammaticality and a Nuclear Physicist in a
nutshell…
How ungrammaticality is grammatical… In Halle and Vergnaud (1987),
and again in Light (1993), it is noted that ungrammaticality can
have a derivation where both the semantic and phonological
requirements are satisfied by the same structure:
[[[un]grammatical] ity]
Here, assuming that un- is not cyclic, and –ity is, we only have to
assume that it is linear and not structural proximity that allows
–ity to affect the stress of the root. As this is the case, this
example does not give positive evidence of late adjunction.
Assuming, following Stepanov, that adjuncts must be the last
elements merged in a numeration, I propose the following
derivation.
n n 2 2 a ity a ity grammatical 2 un grammatical PF grammaticálity
PF ungrammaticality Though this is a possible derivation, nothing
is stopping the adjunct –un from merging in the next phase. This
could be proposed in order to unify this derivation with that of
unhappier, which a Halle and Vergnaud treatment cannot capture. But
a Nuclear Physicist cannot fit in a nutshell! Assuming that
allomorphy is conditioned locally, the affix –ist must merge with
the root before compounding occurs. i.e. physic[k]s vs.
physic[s]ist is not a phonological rule, c.f. cyst [sIst] (Selkirk
1982).
[nuclear[physicist]]
But, we then predict the semantics of the above to be ‘a nuclear
(very small) physicist’ and not ‘a practicioner of nuclear
physics’.
This paradox is solved by the present analysis in the following
way; Numeration 1 (NS) n <physic, ist> 2 physic ist
(MS/PF) the root physi[s] is conditioned by –ist. PHYSICIST
Numeration 2 <X(a phase head), nuclear> X 2 X n 2 √ ist 2
nuclear physic
(MS/PF) NUCLEAR PHYSICIST
Some Previous Analyses…. PESETSKY (1985)…. Quantifier Raising. Deg
Deg 2 2 un Deg Deg eri 2 2 happy er un Deg 2 happy ti
A problem: As QR leads to scope ambiguities, we expect them here,
but they are not found (Hoeksema 1987). STUMP (1991)…. Paradigm
Functions. A root that has been affixed by a non-category changing
affix is still visible to paradigm functions. Therefore those roots
will be inflected on their heads. E.G. unhappy has the same
category features as happy, and therefore the paradigm slot happier
is selected in the scope of the degree-comparative head. A problem:
The extra machinery of paradigms (contra paradigms see Bobaljik
2004) is not necessary, as argued here. See also Kiparsky (1982),
Falk (1991), Spencer (1988), Sproat (1992), Light (1991)…
Conclusions. 1. LATE ADJUNCTION (Lebeaux (1988) and others) OCCURS
TO X0 AS WELL AS XP. 2. ALL BRACKETING PARADOXES INVOLVE A
MORPHLOGICAL
ADJUNCT.
3. ADJUNCTS (Stepanov 2001) HAVE NO UNINTERPRETABLE FEATURES AT
THEIR ROOT NODE. THEY DO NOT PROJECT.
THIS ENTAILS THAT THEY ARE MERGED POST-CYCLICALLY.
4. MORPHEMES THAT PROJECT CATEGORY FEATURES ARE PHASE HEADS
(Marantz 2001, Marvin 2002). 5. ADJUNCTS MUST BE MERGED IN THE
POSITION WHERE THEY WILL BE INTERPRETED AT LF.
6. LITTLE x PHASE HEADS SPELL OUT WITH THEIR COMPLEMENTS.
7. THE LINEAR EDGE CONDITION (Nissenbaum 2000) HOLDS. LATE ADJUNCTS
ARE CONSTRAINED TO A PHONOLOGICAL EDGE.
The Analysis Goes On…. The Particle Verb Conundrum.
First, the longstanding debate over the structure of particle verbs
stems from a structural paradox: the particle in a particle verb
behaves both like a phrase (in that it is separable from the verb
(a)) and like a head (in word formation (b) ) (Den Dikken 1992,
1995, Svenonius 2002, Zeller 1999, Wurmbrand 2000 to mention a
few). (27) John flechtet den Buchstaben ein John braid the letter
in ‘John inserted the letter’ (28) die Einflechtung des Buchstaben
‘the insertion of the letter’
Secondly, there are bracketing paradoxes involving particle verbs
on par with the English cases above, where the particle forms a
semantic unit with the verb, but is not phonologically adjacent to
it (Müller 2003).
(29) herum-ge-renn-e ‘acts of aimless running’ NOT ‘aimless acts of
running’ Here, just as in the case of unhappier, it appears that
two separate structures are needed in order to account for the
phonological and semantic structures implied by the construction.
(30) Phonological Structure Semantic Structure N N 2 2 P N V ge- -e
herum 2 2 V ge- -e P V renn herum renn (Müller 2003:249) As implied
above with the inclusion of the English throw up and up chuck
examples, the solution to the typical Particle Verb paradox in (4)
is amenable to the proposal herein. Here I will show how this
follows from the assumptions above. I will also briefly show how
the atypical paradox in (1) and (2) is also captured within the
theory presented here.
Explaining the paradox… Particles Are Late Adjuncts
Assumptions to keep in mind: 1. (Strong) Inflection triggers
raising. 2. Derivational Heads Lower at MS/PF, and are spelled out
with their complements. 3. Adjuncts are merged where they are
interpreted. Ancillary Assumptions: 1. The circumfix ge- -e in (4)
consists of the inflectional participial prefix ge- and the
nominalizer –e. These are separate heads. 2. Head Movement
reconstructs at LF. A Run Around the Herumgerenne Paradox….
Numeration 1 (NS) v <v, renn> 2 v renn
(MS/PF) the null v and the root are spelled out. RENN Numeration 2
(NS) n <ge,e> 2 Part e ge triggers 2 raising of the t Part
verb v 2 2 ge renn v renn
(MS/PF) the derivational e lowers to Part0 and spells out with its
complement GERENNE
Numeration 3 (NS) X <X (a phase head), herum> 2 X n 2 Part
e
2 t Part
v 2 2 ge renn v √ 2 herum renn
(MS/PF) herum is spelled out with gerenne HERUMGERENNE The above
derivation captures the Phonological Form of the nominalized
Particle Verb, and its Semantic Scope.
A Non-Violation of Lexical Integrity
The non-nominal derivations where the verb is separated from its
particle (27, repeated below as 31a) can also be accounted for by
assuming raising of the verb and late, low merger of the adjunct.
(31) a. [CPJohn [Cflecht] [TPden Buchstaben [VP [Vein ti]]]] John
braid the letter in ‘John inserted the letter’ b. [TP John [vP is
[PartPthrowing [VP it [V ti up]]]] Here the verb raises before the
particle is merged. It later reconstructs to be interpreted. No
special structure is needed to explain the apparent violation of
Lexical Integrity. Particle Verbs are Complex Heads.
SELECTED REFERENCES Bobaljik, J. (2004) Paradigms, Optimal and
Otherwise: A case for Skepticism. Handouts from MIT Paradigms
Workshop
(2000) The Ins and Outs of Contextual Allomorphy. In K.K. Grohmann
and C. Struijke, eds., University of Maryland Working Papers in
Linguistics, volume 10, 35- 71.
Boškovi, . and H. Lasnik (1999) How Strict is the Cycle? Linguistic
Inquiry 30, 691-703. Chomsky, N. (1999) Derivation by Phase. MIT
Occasional Papers in Linguistics, no. 18, Cambridge, MA: MIT
Working Papers in Linguistics, Department of Linguistics and
Philosophy. Dikken, Marcel den (1995). Particles: On the Syntax of
Verb-Particle, Triadic, and Causative Constructions. Oxford
University Press, New York/Oxford. (1992). Particles. Ph.D.
Dissertation, Holland Institute of Generative Lingusitics, Leiden.
Embick, D. & R. Noyer (2001) Movement Operations After Syntax.
Linguistic Inquiry 32 (4), 555-595. Falk (1991) Bracketing
Paradoxes without Brackets." Lingua 84: 25-42. 1991 Fox, D. and D.
Pesetsky (2004) Cyclic Linearization of Syntactic Structure. draft.
MIT. Halle, M. (1998) The Stress of English Words 1968-1998.
Linguistic Inquiry 29, 539-568. Halle & Marantz (1994) Some Key
Features of Distributed Morphology", MIT Working Papers in
Linguistics, 21, 275-288, 1994. (1993) Distributed Morphology and
the Pieces of Inflection," in K. Hale and S.J. Keyser, eds., The
View From Building 20, Cambridge, Mass.: MIT Press, 111-176. Halle,
M. & J.-R.Vergnaud (1987) An Essay on Stress. Cambridge MA: MIT
Press. Hoeksema, J. (1987) Relating Word Structure and Logical
Form. Linguistic Inquiry 17, 177-183. Kiparsky, P. (1982) From
Cyclic Phonology to Lexical Phonology. In H. van der Hulst and N.
Smith, eds., The Structure of Phonological Representations,
Dordrecht: Foris, 130-
175. Lebeaux, D. (1988) Language Acquisition and the Form of the
Grammar. PhD Dissertation. UMass.: MIT Press. Light, M. (1993)
Taking the Paradoxes Out of Bracketing in Morphology," in
Proceedings of the Second Formal Linguistics Society of Mid-America
Conference. University of
Michigan, 1991. Marantz, A.(2001) Words.(MSWord document) Handout
from WCCFL XX.
(1997) No Escape from Syntax: don’t Try Morphological Analysis in
the Privacy of your own Lexicon. In Dimitriadis, A., and Siegel, L.
(eds.). University of Pennsylvania Working Papers in Linguistics,
4.2. 201-225.
Marvin, T. (2002) Topics in the Stress and Syntax of Words. PhD
Dissertation. MIT. Müller, S. (2003) Solving the Bracketing
Paradox: An Analysis of the Morphology of German Particle Verbs.
Ms. Language Technology Lab, DFKI GmbH, Saarbrücken, Germany
Nissenbaum, J. (2000) Investigations of Covert Phrase Movement. PhD
Dissertation. MIT. Ochi, M. (1999) Multiple Spell-Out and PF
Adjacency. In Proceedings of the North Easter Linguistic Society
29. Amherst, Mass: GLSA. Pesetsky, D. (1979) Russian Morphology and
Lexical Theory.
(1985) Morphology and Logical Form. Linguistic Inquiry 16, 193-246.
Selkirk, E. (1982) The Syntax of Words. Cambridge MA: MIT Press.
Spencer, A. (1988) Bracketing Paradoxes and the English Lexicon.
Language 64(4), 663-682. Sproat (1992), Unhappier is not a
Bracketing Paradox," Linguistic Inquiry, 23, 347-352, 1992. (1985)
On Deriving the Lexicon. The Linguistic Review (Dissertation
Abstracts), 5, 185-196. (1984) On Bracketing Paradoxes. in M. Speas
and R. Sproat (eds.), MIT Working Papers in Linguistics, 7,
110-130. Stepanov, A. (2001) Late Adjunction and Minimalist Phrase
Structure. Syntax 4:2, 94-125. Stump, G. (1991) A Paradigm-based
Theory of Morphosemantic Mismatches', Language 67, 675-725
Svenonius and G. Ramchand. (2002) The Lexical Syntax and Lexical
Semantics of the Verb-Particle Construction. Paper presented in
April 2002 in Santa Cruz at the West Coast
Conference on Formal Lingusitics. To appear in the Proceedings.
Wurmbrand 2000 The Structure(s) of Particle Verbs. Draft. Zeller,
Jochen (1999). Particle verbs, local domains, and a theory of
lexical licensing. Ph.D. Dissertation,
Johann-Wolfgang-Goethe-Universität zu Frankfurt am Main,
Frankfurt.
(1997). Particle verbs and a theory of late lexical insertion. Ms
University of Frankfurt, Frankfurt.