Download Roots and Domains. Some remarks “On the Identity

yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Roots and Domains. Some remarks “On the Identity of Roots”
Elena Anagnostopoulou, University of Crete
As summarized in the concluding section 5, Harley investigates three issues: (i) the
nature and properties of roots, (ii) the syntactic behavior of roots, i.e. whether they
undergo Merge with phrasal constituents or not and whether they project or not, and
(iii) the demarcating head for special interpretation of roots, whether this is the first
categorizing head or the higher head Voice. In this commentary, I will concentrate on
the third question, addressing it from the point of view of previous work on Greek
participles which discusses this very same question (Anagnostopoulou & Samioti
2013, to appear).
2. Harley’s argument that Voice is the phasal head determining inner vs. outer
attachment behavior
Harley takes as a starting point Marantz’s (2001)/ Arad’s (2003) hypothesis
summarized in (1):
The Marantz / Arad Hypothesis
Roots are assigned an interpretation in the context of the first category
assigning head/ phase head merged with them, which is then fixed throughout
the derivation
Harley (p. 32) argues (with Borer 2008, 2009) that the evidence from words based on
series of derivational affixes does not suggest a clear division between (i) irregular,
idiosyncratic, idiomatic interpretation inside the first categorizing head vs. (ii) regular,
productive, compositional interpretation outside this head. Since roots never occur in
isolation and cannot be interpreted unless they combine with a categorizing head, they
are expected to be assigned an idiosyncratic interpretation at the first categorizing
affix. However, this meaning does not have to be fixed for subsequent computation,
and idiosyncratic semantics can also be assigned above the first catergorizing head, at
later derivational cycles. She brings examples like the following as an argument for
her point:
a. nature
b. class
‘made natural’
‘become a citizen by
residing in a country’
# ‘things which have been classified
‘small newspaper advertisements’
This research has been supported by an Alexander von Humboldt Bessel award (2013). I would like
to thank Alec Marantz for comments.
c. nation
‘make national’
‘government takeover of business’
Not antonym: private
(Antonym: privatize)
d. √domin
# ‘woman who dominates’
‘woman who performs
ritualized sexual domination’
In these examples, multiply derived words have new interpretations not inherited by
the constituents they contain. Moreover, in (2b) and (2d), the compositional meaning
contributed by their subpart is unavailable. Harley mentions a number of other cases
where the entailments of the words they contain do not contribute compositionally to
the meaning of the outer-derived words, e.g. universe-university, hospital-hospitality,
sanitary-sanitarium, auditory-auditorium, relate-relation-relationship. She concludes
that the slippery and gradient judgments concerning differences in compositionality
are different from the classic examples of inner vs. outer derivational morphology,
and that the boundary delimiting the inner vs. outer domain in words is Voice, the
head introducing the agent argument (as originally suggested by Marantz 1997):
syntactic causatives, eventive passives, eventive nominalizations, -able formations all
involve outer affixation, which must be understood as attachment above Voice and
not as attachment above v, i.e. Voice and not v is a phase head in the vP domain. As
for the observation that root-derived words are more idiosyncratic than word-derived
words, she takes this to reflect a difference in degree rather than kind: the more
structure is involved the less frequent idiosyncratic non-compositionality becomes
(just as it has claimed to be the case with non-word idioms).
3. Adjectival participles present evidence that v and not Voice demarcates the
inner vs. outer domain for the distribution of participial morphology in English
and Greek
Two of the examples cited by Harley (p. 34) as evidence for the distinction between
‘inner’ vs. ‘outer’, ‘lexical’ vs. ‘productive’ occurrences of the same affixes are
stative vs. resultative participles in English (Embick 2003; Embick 2004) and stative
vs. resultative participles in Greek (Anagnostopoulou 2003; Alexiadou &
Anagnostopoulou 2008). These are adjectival participles which split into two main
types, depending on whether they denote a state resulting from a prior event or they
lack event implications (see Kratzer 1994, 2000). The former are called “resultative”
and the latter “stative” by Embick. Both types are adjectival in English and Greek, i.e.
they involve an (abstract) adjectival head above the head hosting participial
In English, we always find regular -ed participle morphology on participles
with event implications while we find irregular morphology with participles without
event implications. For the root √OPEN, “opened” has event implications and “open”
not. With √CLOSE, the same form “closed” is ambiguous. As a result, “open” but not
“opened” is licit after verbs of creation, which can only combine with participles
lacking event implications. “Closed”, which is ambiguous, is licit in this context (see
Embick 2004):
Not universally though; Dubinsky & Simango (1996) show that Chichewa statives behave like
adjectival passives but are verbal.
The door was built open/ *opened
The door was built closed
In Greek, the suffix –menos always occurs on participles with event implications, and
the suffix –tos always occurs on participles without event implications (see
Anagnostopoulou 2003):
“opened”/ “open”
Only –tos forms are licit under verbs of creation which can only combine with
participles without event implications (compare to (3) above for English):
To jiaurti
pix-to / *pig-meno
The yoghourt was made
thick-to/ thick-meno
‘The yoghourt was made thick/*thickened’
Several diagnostics have been argued to differentiate adjectival participles with event
implications from adjectival participles without. Except for the verb of creation test
illustrated in (3) and (5), the two types differ e.g. with respect to the licensing of
manner adverbs as well as restitutive/repetitive ambiguities (von Stechow 1998).
Within DM it has been argued that these differences can be accounted for if
participles with event implications contain a little v head introducing an event variable,
while participles without event implications lack such a head (see e.g. Embick 2004
for English and Alexiadou & Anagnostopoulou 2008, Anagnostopoulou & Samioti
2013, to appear for Greek). On this analysis, -menos attaches above v and –tos
attaches to the Root:
attachment of -menos above v
attachment of -tos to the Root
Embick (2003) argues that the irregular vs. regular distribution of participial
morphology in English can receive a satisfactory treatment in terms of the hypothesis
that there are two points for Vocabulary Insertion (VI) in English participles. VI is
divided into (i) the Root Cycle for Root-attached ASP suffixes 3 and (ii) an Outer
Cycle for non-Root attached ASP suffixes:
Spell-Out of ASP: Root Cycle
ASP ↔ -en / ______ {√ROT, √SHRINK,.....}
Embick assumes that participial morphology is an exponent of ASP, a stativizing head.
ASP ↔ -0 / _______ {√OPEN, √EMPTY, ...}
ASP ↔ -t /________ {√BEND,.......}
ASP ↔ -èd/ _______ {√BLESS, ALLEGE, AGE, .....}
ASP ↔ -ed/ ______ {√CLOSE, OBSTRUCT,.....}
Spell-Out of ASP: Outer Cycle (above v)
ASP ↔ -en / _____ { √BREAK, √SPEAK,.....}
ASP ↔ -0 / ______ { √HIT, √SING, √SHRINK, ....}
ASP ↔ -t / ______ {√BEND, √BUY,.......}
ASP ↔ -ed
As a result, √ROT is on a list in the Root Cycle for –en, but not in the Outer Cycle
(where default –ed surfaces): rotten vs. rotted, √ SHRINK is on a list for –en in the
Root Cycle and on a list for 0 in the Outer Cycle, and so on.
Importantly for present purposes, it is clearly the little v head introducing an
event variable and not Voice that delimits the inner vs. outer domain in English and
Greek adjectival participles. First, unaccusatives may yield adjectival participles with
event implications in examples like “a gradually/dangerously frozen lake”, “suddenly
fallen leaves”, “slowly melted snow” (taken from Bruening, to appear), and they
clearly do not contain Voice. Moreover, the issue of whether adjectival participles
ever contain Voice in English is highly controversial. Kratzer (1994, 1996) and
Embick (2004) argue that Voice is not possible in English adjectival participles. Note
that by-phrases, instruments and agent-oriented adverbs, standardly taken to be
licensed by Voice, are ill-formed in contexts like (9) where participles are
unambiguously adjectival:
The door seems broken/ opened/ painted (*by Mary).
Alexiadou, Anagnostopoulou & Schäfer (to appear) building on – and modifying –
McIntyre (2013) and Bruening (to appear) argue that Voice can be included in
English adjectival participles (contra Kratzer 1994, 1996 and Embick 2004), but its
presence is severely restricted.
Unlike English, Greek adjectival participles productively employ Voice, as
argued for in Anagnostopoulou (2003). Crucially, however, Voice is only licit in
participles describing irreversible states resulting from prior events, and is not allowed
in participles describing transitory and reversible states resulting from prior events.
The former are what Kratzer (2000) calls “resultant state participles”, which do not
license the adverb ‘akoma’ (still). The latter are what Kratzer (2000) calls “target state
participles” which license the adverb ‘akoma’ (still). In Anagnostopoulou (2003), I
pointed out that by-phrases, instruments and agent-oriented adverbs are sharply
ungrammatical in the presence of ‘akoma’ still, a fact suggesting that Voice cannot be
included in target state participles (Voice can only be licensed in resultant state
participles, i.e. when ‘akoma’ still in (10) is missing, by-phases and instruments are
well-formed) :
lastixa ine
akoma fuskomena
The tires are
still inflated
‘The tires are still inflated by Mary’
(*apo tin
lastixa ine
akoma fuskomena
The tires are
‘The tires are still inflated with the pump’
(*me tin
with the
But note that –menos uniformly occurs on participles with event implications in Greek,
regardless of whether they describe irreversible/resultant states, hence containing
Voice, or transitory and reversible states/target states, hence disallowing Voice. This
is so because v and not Voice is the head demarcating the inner vs. outer domain for
Greek participial morphology, like English (see (8) above): -menos is the outer
morpheme attaching above eventive v in Greek participles, while –tos is an inner
4. Meaning domains, v and Voice
Having presented evidence that eventive v, and not Voice, qualifies as the head
delimiting the domain for inner vs. outer morphology in both English and Greek
adjectival participles, let us now turn to the question of idiosyncratic vs.
compositional meaning in Greek participles. The issue is discussed in detail in
Anagnostopoulou & Samioti (2013, to appear). In this section, I briefly summarize
their main points.
Greek –tos participles often have a special meaning not related to the
corresponding verbs. By contrast, –menos forms always have a compositional
meaning identical to that of the corresponding verbs:
a. spas-tos
b. spas-menos
The verbal meaning of “break” is retained in the -menos participle in (11b), while
(11a) has a distinct interpretation: it means “folding” and applies to objects consisting
of parts that can “break into” smaller pieces:
break-ti-fem umbrella /
‘folding umbrella’
‘folding table’
‘broken umbrella”
‘broken table’
The two meanings are closely related but sufficiently distinct to be expressed by
different roots in a language like English, √BREAK vs. √FOLD. We can account for
the facts in (11)-(13) in terms of Marantz’s (2013) theory of contextual allosemy.
Marantz argues that contextual allosemy is the semantic analogue of contextual
allomorphy and amounts to local polysemy resolution within a specific context (cf.
(15) in Harley’s paper):
Contextual allosemy = the choice of the meaning of a morpheme – the
appropriate alloseme- within a local environment.
What the –menos facts show, as opposed to the –tos facts, is that once a little v head
introducing an event variable is merged with a root like √SPAZ, the “verbal” literal
meaning of the root is fixed, and further attachment of the adjectivizing head
associated with -menos cannot alter this meaning:
v = Spelled Out in the context of a = verbal meaning
Root = Spelled Out in the context of v=
choice of an alloseme
in the context of v (√SPAZ interpreted as “break”)
Therefore, -menos participles always inherit the meanings of the corresponding verbs.
Following Marantz (2013), and contra what Harley proposes, we can assume that this
systematic correspondence between the verbal meaning of the root in verbs and the
corresponding adjectival –menos participles with event implications is due to the fact
that v is a phase head. Once v is merged with a complement, it triggers spell-out, and
the higher head “a” in (15) can no longer serve as context for the choice of an
alloseme located below that phase head. By contrast, v is absent in the –tos participle,
and –tos can serve as context for choice of the meaning of √SPAZ as “fold”, not the
same meaning as that of the corresponding verb:
Root =
choice of an alloseme
the context of -tos (potentially different alloseme than in
the context of v (√SPAZ interpreted as “fold”).
Importantly, “spas-tos” and “spas-menos” can form closely related (though not
identical) idioms:
‘broken Greek (not very good Greek)’
foni voice
‘broken voice (voice that does not sound clear)’
It so happens that “spas-tos” and “spas-menos” both have an idiomatic interpretation
and a “literal” interpretation. The crucial difference between the two types of
participles, though, concerns the way in which the root is interpreted in the “literal”
and not in the idiomatic interpretation. In the –menos participle, the root has exactly
the same literal meaning as the root of the corresponding verb, while the root in the –
tos participle receives a distinct interpretation. Anagnostopoulou & Samioti (2013; to
appear) extensively argue that the boundary for idiomatic interpretations of the type
illustrated in (17) with adjectival participles is indeed the Voice head introducing an
agent argument, as was originally proposed by Marantz (1997) for word idioms and
phrasal idioms, in agreement with what Harley proposes here for word idioms and
what Harley & Stone (2013) propose for phrasal idioms.
Anagnostopoulou & Samioti (2013, to appear) furthermore point out that there
exist semantically empty categorizers (in the cases we discuss, verbalizers which are
not exponents of a little v head introducing an event variable), which are included in
stative –tos participles and do not count as boundaries for contextual allosemy of the
root in the context of -tos.
More specifically, Greek productively employs verbalizing suffixes which
have been analyzed as root verbalizers (Alexiadou 2001, 2009; Giannakidou &
Merchant 1999):
Root- verbalizing elements
Greek: -iz, - on-, -en/an, -ev,- -az, -a :
These elements are present on the verbs in (19):
white-iz-1sg ‘whiten’
ice-on-1sg ‘freeze’
As expected by the “outer-analysis” of –menos participles, verbalizers are also
productively present in –menos participles:
But many Greek –tos participles also include such verbalizing elements. This happens
when the root is classified as a predicate of entity or as a predicate of property and not as
a predicate of event, in terms of the basic ontology of roots discussed in Harley’s paper
(p. 15). Anagnostopoulou & Samioti (to appear) argue that these verbalizers are needed
to turn an “entity” or “property” root into an “event” root, in order for it to be able to
further combine with –tos:
‘steaming hot’
axn-os ‘steam’
vid-a ‘screw’
Despite the presence of verbalizers, the –tos participles in (21) do not have event
implications, they do not license manner modification or repetitive readings (see
Anagnostopoulou & Samioti 2013, to appear for detailed discussion). The participles
of the type illustrated in (21) show that the abstract little v head described in the
decomposition literature as the the semi-functional head introducing an event variable
must, under certain well-defined conditions, be dissociated from morphological
verbalizers. And despite the presence of an overt verbalizer in –tos participles like (21)
above and (22) below, they differ from the corresponding –menos participles w.r.t.
allosemy in exactly the same way as was described in (11)-(13) above for participles
based on “event” roots like √SPAZ. In both (22a) and (22b) below, the root √KOKIN
means “red”. In the context of –tos, however, as in (22a), “red” acquires a very
specific meaning, it means “cooked with a red sauce” (see (23a)), and the verbalizer –
iz is ignored. On the other hand, “kokin-iz-menos” in (22b) can apply to anything that
has become red as a result of a “becoming red event”, as shown in (23b):
made red
made red
kreas/ kotopoulo/*magoulo
meat/chicken/ *cheek
‘meat/ chicken with a red sauce’
kokin-iz-meno derma/ magulo/ mati/ xroma
red-v-meno skin/ cheek/ eye/ colour
‘skin/cheek/eye/colour that has turned red as a result of an event’
Following Marantz (2013), Anagnostopoulou & Samioti (2013) propose that the
verbalizing heads in –tos participles are semantically empty heads, the zero semantic
counterparts of phonologically empty v heads in cases of allomorphy (Embick 2010).
Such semantically empty heads are ignored for allosemy and, therefore, -tos can serve
as a context determining allosemic choice of the root √KOKIN in (22a) and (23a),
which is quite distinct from the meaning of the root in (22b) and (23b), the canonical
verbal meaning.
In conclusion, the above considerations lead to a more refined picture of how
to understand the hypothesis in (1) in connection to a) idiosyncratic meaning and b)
the kind of little v head serving as a phase head. First, it is necessary to treat
differently examples like (2a) and (2c), where an idiomatic meaning is possible along
with the compositional meaning, from cases like (2b) and (2d) where only the
idiosyncratic meaning is available. The boundary for idiomatic interpretation in cases
like (2a) and (2c) indeed seems to be Voice, as Harley proposes, unlike the
interpretation determined at the little v cycle, which “fixes” what can be called the
“verbal” meaning of the root. Crucially, though, it is the little v head introducing an
event variable that serves as a phase head in participles. In cases where a categorizing
little v head does not introduce an event variable, qualifying as semantically empty
(Marantz 2013), then a) inner and not outer morphemes attach outside this head and b)
the semantically empty head is ignored for contextual determination of the meaning of
the root. From the present perspective then, in examples like (2b) and (2d) which only
have an idiosyncratic interpretation and do not inherit the compositional meaning of
the verbal constituent they contain, the derivational suffixes –ify and –ate must be
seen as semantically empty verbalizers.
Alexiadou, Artemis. 2001. Functional structure in nominals: nominalisation and
ergativity. Amsterdam: John Benjamins.
Alexiadou, Artemis. 2009. On the role of syntactic locality in morphological
processes: the case of Greek nominals. In A. Giannakidou & M. Rathert (eds.),
QP, Nominalizations and the role of DP. Oxford: Oxford University Press, 253280.
Alexiadou, Artemis & Elena Anagnostopoulou. 2008. Structuring participles. In C.
Chang and H. Haynie (eds.) Proceedings of the 26th West Coast Conference on
Formal Linguistics. Somerville, MA: Cascadilla Press, 33-41.
Alexiadou, Artemis, Elena Anagnostopoulou & Florian Schäfer. To appear. External
Arguments in Transitivity Alternations. A Layering Approach. Oxford: Oxford
University Press.
Anagnostopoulou, Elena. 2003. Participles and Voice. In A. Alexiadou, M. Rathert &
A. von Stechow (eds.) Perfect Explorations. Berlin: Mouton de Gruyter, 1-36.
Anagnostopoulou, Elena & Yota Samioti. 2013. Allosemy, Idioms, and their
Domains : Evidence from Adjectival Participles. In R. Folli, C. Sevdali & R.
Truswell (eds.) Syntax and its Limits. Oxford : Oxford University Press, 218-250.
Anagnostopoulou, Elena & Yota Samioti. To appear. Domains within Words and their
meanings : a case study. In A. Alexiadou,, H. Borer & F. Schäfer (Eds.) The
syntax of Roots and the roots of Syntax. Oxford: Oxford University Press.
Arad, Maya. 2003. Locality constraints on the interpretation of roots: The case of
Hebrew denominal verbs. Natural Language & Linguistic Theory 21: 737–778.
Borer, Hagit. 2008. Notes for Late Insertion. Paper given at WCCFL 27 UCLA.
Borer, Hagit. 2009. Very Late Insertion. Paper presented at the Root Bound workshop,
February 20-21, USC.
Bruening, Benjamin. To appear. Word Formation is Syntactic: Adjectival Passives in
English. In Natural Language and Linguistic Theory.
Dubinsky, S. & Simango, R. 1996. Passive and stative in Chichewa: Evidence for
Modular Distinctions in Grammar. Language 72: 749-81.
Embick, David. 2003. Locality, Listedness and Morphological Identity. Studia
Linguistica 57: 143-169.
Embick, David. 2004. On the structure of resultative participles in English, Linguistic
Inquiry 35: 355-392.
Embick, David. 2010. Localism vs. Globalism in Morphology and Phonology.
Cambridge MA: MIT Press.
Giannakidou, Anastasia & Jason Merchant. 1999. Why Giannis can’t scrub his plate
clean: On the absence of resultative secondary predication in Greek. In A. Mozer
(ed.) Greek Linguistics’ 97: Proceedings of the 3rd International Conference on
Greek Linguistics. Athens: Ellinika Grammata, 93-103.
Harley, Heidi & Megan Schildmier Stone. 2013. The ‘No Agent Idioms’ Hypothesis.
In R. Folli, C. Sevdali & R. Truswell (eds.) Syntax and its Limits. Oxford : Oxford
University Press, 251-275.
Kratzer, Angelika. 1994. The Event Argument and the Semantics of Voice. Ms.,
University of Massachusetts at Amherst.
Kratzer, Angelika. 1996. Severing the external argument from its verb. In J. Rooryck
& L. Zaring. Dordrecht: Kluwer, 109-137.
Kratzer, Angelika. 2000. Building statives. In L. Conathan, J. Good, D. Kavitskaya, A.
Wulf, and A. Yu (eds.) Proceedings of the 26th Annual Meeting of the Berkeley
Linguistics Society, 385-399.
Marantz, Alec. 1997. No escape from Syntax: Dont' try morphological analysis in the
privacy of your own lexicon. In: Dimitriadis, A, Siegel, L (eds) University of
Pennsylvania Working Papers in Linguistics, Philadelphia: University of
Pennsylvania, 201-225.
Marantz, Alec. 2001. Words and Things. Ms., MIT & NYU.
Marantz, Alec. 2013. Locality domains for contextual allomorphy across the
interfaces. In O. Matushansky & A. Marantz (eds.) Distributed Morphology
Today: Morphemes for Morris Halle. Cambridge MA: MIT Press, 95-115.
McIntyre, Andrew. 2013. Adjectival passives and adjectival participles in English. In
A. Alexiadou & F. Schäfer (eds.) Non-Canonical Passives. Amsterdam: John
Benjamins, 21-42.
von Stechow, Arnim. 1998. German participles in Distributed Morphology. Ms.
University of Tübingen.