Heavy-Head-Light: Non-Existent Combination of

*Heavy-Head-Light: Non-Existent Combination of Head-Complement Orders
In this paper, we will point out that no language has heavy complements in pre-head position and
light complements in post-head position (*Heavy-Head-Light). We will argue that this word order
combination does not exist in any language because movement of heavy complements into the specifier
position of a head violates a constraint prohibiting phrasal categories in compound-like structure.
Previous studies have pointed out that some word order combinations are non-existent. First,
Greenberg (1963) states that if the pronominal object follows the verb, so does the nominal object (V-Prn
 V-DP). That is, there is no language in which a verb is followed by its pronominal object and preceded
by its DP object (*DP-V-Prn). In addition to consistent VO languages (V-Prn/DP (1)) and consistent OV
languages (Prn/DP-V (2)), there are mixed languages such as Bantu and Romance which show Prn-V-DP
order (3). However, there is no ‘mirror-Bantu/Romance’ language (*DP-V-Prn (4)).
Similarly, there are mixed orders DP-V-CP (5). However, as Newmeyer (2005) shows, no language
has nominal objects obligatorily in post-verbal position and sentential objects in pre-verbal position
(*CP-V-DP (6)). CP is heavier than DP in that CP contains more branching nodes than DP, which is
heavier than pronoun. Thus, we can generalize *DP-V-Prn (4) and *CP-V-DP (6) as *Heavy-V-Light.
*Heavy-V-Light can further be generalized into *Heavy-Head-Light, because we can find no
examples parallel to (4) and (6) in the cases of P and N. First, in addition to consistent orders P-Prn/DP
(head-initial) and Prn/DP-P (head-final), there is a mixed type Prn-P-DP (7). However, as far as we know,
there is no ‘mirror-Germanic’ language with DP-P-Prn order (*DP-P-Prn (8)).
Second, let us assume adverbial subordinators (complementizers) to be a kind of preposition taking a
sentential complement (e.g. for in English). Examination of the data in Haspelmath et al. (2005) shows
that out of 541 languages, there are 46 languages with the DP-P-IP order (9) and only two languages
(Buduma and Gününa Küne) that have prepositions and clause-final subordinators/subordinating suffixes
(*IP-P-DP (10)). Thus, we can express *DP-P-Prn (8) and *IP-P-DP (10) as *Heavy-P-Light.
Third, in English, ’s-genitives are generally preferred for short phrases (including pronouns, e.g. his
book), while of-phrases are preferred for longer phrases (e.g. books of this period) (Biber et al. 1999).
This fact shows that English has a Light-N-Heavy order (11). However, as far as we know, there is no
‘mirror-English’ language with Heavy-N-Light order (12). This becomes clearer when we consider
Genitive-Noun and Relative Clause-Noun orders. Examination of the data in Haspelmath et al. (2005)
shows that out of 640 languages, there are 134 languages with Gen-N-Rel order (DP-N-CP (13)) and only
one language (Tigré) with Rel-N-Gen order (*CP-N-DP (14)). Thus, we conclude that
*Heavy-Head-Light holds with V, P and N in almost all languages. We also re-examine data from the
exceptional languages above and show that they are not real counter-examples to *Heavy-Head-Light.
Next, we consider why this pattern is not allowed. This might be due to its marked information
structure, which goes against the prevalence of light-heavy order in languages. However, performance
theories such as Hawkins’s (1994) Early Immediate Constituent cannot explain why Light-Head-Heavy
orders are prevalent in languages in spite of the fact that they are less efficient in parsing than
Light/Heavy-Head and Head-Light/Heavy.
We will provide a formal explanation in terms of a morpho-syntactic constraint banning phrasal
compounds. Following Kayne (1994) and Julien (2002), we assume that movement of complement into
the specifier position of a head derives head-final order. We argue that this movement changes a
right-branching phrase [XP Spec [X [YP …]] into a (phonologically) left-branching structure [[YP …] X],
which is more like a compound word than a phrase. The compound nature of left-branching structure is
found in the agglutinative morphology of head-final languages and phonological phenomena such as
Japanese Voicing and Korean n-Insertion, which apply in left-branching structure (15a), but not in
right-branching structure (15b). Then, elements in the pre-head position are phonologically ‘incorporated’
into the head to make a ‘phrasal compound’ (cf. over the fence gossip, a sleep all day look).
However, phrasal compounds are not allowed freely because they violate a constraint prohibiting
phrasal categories in compounds (cf. No Phrase Constraint (Botha 1984)). This constraint can be violated
by ‘incorporating’ a certain size of constituent into a head, that is, by moving a complement into the
specifier position. Head-initial languages do not allow even pronouns, the lightest complement, to move
into the specifier of a head; Light-Head-Heavy languages allow only light complements to move into
Spec; Head-final languages allow even CPs, the heaviest complement, to move into Spec. However, there
cannot be any language that moves only heavy constituents but not light constituents. Thus, we can
explain why *Heavy-Head-Light holds in languages without exception.
We hope that this study shows not only an interesting typological universal but also a new approach
to the interface between phonology, morphology and syntax.
(1)
V-Prn/DP [English]
call her/Mary
(2) Prn/DP-V [Japanese]
kanojo-o/Mari-o yobu
her-Acc/Mari-Acc call ‘call her/Mari’
(3) Prn-V-DP [Swahili]
a. nili-ki-tafuta
(Prn-V)
I-Past-it-look ‘I looked for it.’
b. nilitafuta kisu
(V-DP)
I-Past-look knife ‘I looked for a knife.’
(4) * DP-V-Prn [non-existent] (#: a hypothetical language)
a. # Mary(Acc) see
(DP-V)
b. # see her
(V-Prn)
(5) DP-V-CP [Persian]
a. Bache-hâ panjare râ shekast-and
(DP-V)
child-Pl window Obj break-Past-3Pl ‘The children broke the window.’
b. An zan
mi
danat ke
an mard sangipartab kard (V-CP)
that woman Cont knows Comp that man rock
threw
‘The woman knows that the man threw a rock.’
(6) * CP-V-DP [non-existent]
a. # The children broke the window.
(V-DP)
b. # The woman that the man threw a rock knows.
(CP-V)
‘The woman knows that the man threw a rock.’
(7) Prn-P-DP [Germanic]
a. therein ( in there)/in that place (Prn-P/P-DP) [English]
b. damit ( mit da-) ‘with it’/mit Honig ‘with honey’ (Prn-P/P-DP) [German]
c. daarmee ( met daar-) ‘with that’/met luchtpost ‘by airmail’ (Prn-P/P-DP) [Dutch]
(8) * DP-P-Prn [non-existent]
# that place in/in there (DP-P/P-Prn)
(9) DP-P-IP
[Finnish, 46 languages]
a. syntymäpäiväni johdosta (DP-P)
birthday
for
‘for birthday’
b. Äiti
laulaa, jotta lapsi nukkuisi. (P-IP)
mother sings so-that child sleep-con-3Sg
‘Mother is singing so that the child will go to sleep.’
(10) * IP-P-DP [non-existent? Cf. Buduma and Gününa Küne]
(11) Prn-N-PP [English]
his book/books of this period (Prn-N/N-PP)
(12) * PP-N-Prn [non-existent]
(13) DP-N-CP [Swedish, 134 languages]
a. en flichas hatt ‘a girl’s hat’ (DP-N)
b. en vän, som bor i Stockholm (N-CP)
a friend who lives in Stockholm ‘a friend who lives in Stockholm’
(14) * CP-N-DP [non-existent? Cf. Tigré]
(15) a. [[nise tanuki] shiru] → nise danuki jiru
mock badger soup
‘mock-badger soup’
b. [nise [ tanuki shiru]] → nise tanuki jiru
mock badger soup
‘mock badger-soup’
References
Biber, D. et al. 1999. Longman grammar of spoken and written English. Longman.
Botha, R. 1984. Morphological mechanisms. Pergamon Press.
Greenberg, J. 1963. Some universals of grammar with particular reference to the order of meaningful
elements. In J. Greeenberg, ed. Universals of language. MIT Press, 73-113.
Haspelmath, M., et al. 2005. The world atlas of language structures. Oxford Univ. Press.
Hawkins, J. 1994. A performance theory of order and constituency. Cambridge Univ. Press.
Julien, M. 2002. Syntactic heads and word formation. Oxford University Press.
Kayne, R. 1994. The antisymmetry of syntax. MIT Press.
Newmeyer, F. J. 2005. Possible and probable languages. Oxford University Press.