Publishing Partner: Cambridge University Press CUP Extra Wiley-Blackwell Publisher Login
amazon logo
More Info

New from Oxford University Press!


Latin: A Linguistic Introduction

By Renato Oniga and Norma Shifano

Applies the principles of contemporary linguistics to the study of Latin and provides clear explanations of grammatical rules alongside diagrams to illustrate complex structures.

New from Cambridge University Press!


The Ancient Language, and the Dialect of Cornwall, with an Enlarged Glossary of Cornish Provincial Words

By Frederick W.P. Jago

Containing around 3,700 dialect words from both Cornish and English,, this glossary was published in 1882 by Frederick W. P. Jago (1817–92) in an effort to describe and preserve the dialect as it too declined and it is an invaluable record of a disappearing dialect and way of life.

New from Brill!


Linguistic Bibliography for the Year 2013

The Linguistic Bibliography is by far the most comprehensive bibliographic reference work in the field. This volume contains up-to-date and extensive indexes of names, languages, and subjects.

Email this page
E-mail this page

Review of  Complementation

Reviewer: Bruce Connell
Book Title: Complementation
Book Author: R. M. W. Dixon Chia-jung Pan
Publisher: Oxford University Press
Linguistic Field(s): Morphology
Book Announcement: 21.1579

Discuss this Review
Help on Posting
EDITORS: R.M.W. Dixon and Alexandra Y. Aikhenvald
TITLE: Complementation
SUBTITLE: A cross-linguistic typology
SERIES: Explorations in Linguistic Typology 3
PUBLISHER: Oxford University Press
YEAR: 2008

Mahamane L. Abdoulaye, Abdou Moumouni University, Niamey, Niger


This book is the third in the series ''Explorations in Linguistic Typology'' and
so belongs to a suite of state-of-the-art volumes on cross-linguistic and
typological studies produced or edited over the years by the two renowned
linguists (now at James Cook University, Australia; Aikhenvald and Dixon are the
general editors of the series and editors of the so far four published volumes).
The present volume addresses the issue of complementation based on the
examination of a number of languages, in particular the eleven languages treated
in the case-study chapters. These chapters are all guided by the initial
theoretical chapter written by Dixon, which also serves as an introduction to
the co-edited volume. Beside the main chapters, the book also contains a table
of contents, a preface, bio-sketches of the contributors, a list of
abbreviations, and three indices for authors, languages, and subjects.


In the first contribution (Chapter 1, ''Complement clauses and complementation
strategies in typological perspective''), Dixon claims that there are
typologically three groups of languages regarding complementation: (1) languages
using only true complement clauses (henceforth, CCs), (2) languages with a mix
of CCs and complementation strategies (henceforth, CSs), and (3) languages
devoid of CCs and having only CSs. One of the main ideas in the book indeed is
the distinction between CCs and the functionally similar notion of CS. Both are
used to complement a main verb (such as 'hear, see, think') with a proposition.
In a CC construction, the propositional complement has the attire (more or less)
of a typical nominal argument (say a nominal Object). By contrast, in a CS
construction, the propositional complement is attached to the main verb through
various (sometimes peripheral) linking structures based on available
constructions such as relative clauses, serial verb constructions, purposive
clauses, direct speech constructions, nominalization, sequential (or chaining)
constructions, apposition, etc.

According to Dixon, previous studies (notably Noonan 1985 and its web-based
revised version) did not make the distinction between CCs and CSs mainly because
they did not consider it necessary to define the criteria for what a CC is. As a
consequence, alternatives to the CCs, i.e., the CSs, were not given due
attention (see Dixon 1995; however, if one goes by the anecdote on p. 263, then
the hint on the CSs may have come as early as in the 1970s). Dixon claims (p.
15f) that a CC can be recognized by the following four criteria: First, the
propositional complement has the internal constituent structure of a clause
(core arguments inside the CC should have the same marking as in regular
clauses). Secondly, the propositional complement functions as a core argument of
the main verb, most often as direct object (O), but also as A(gent) of a
transitive verb (cf. p. 19 ''{John's having carried the log home} shows {that he
is a strong fellow}''), as S(ubject) of an intransitive verb (cf. p. 17 ''{that
John was an academic} didn't matter''), as an extended core argument of a
transitive or intransitive verb (cf. p. 18 ''the doctor promised Mary {that he
would cure her}''), or as a copula subject or a copula complement (cf. p. 18
''{that John did it} is true'' and ''the truth is {that John did it}''). Third, the
CC always describes a proposition (a fact, an activity, or a potential event).
Finally, in any language that has them, CCs function as core arguments of
prototypical CC-taking verbs such as 'see', 'hear', 'know', 'believe', 'like',
'tell', and 'want'.

CSs hence differ from CCs in that they are not core arguments of the main verb.
For example, in a CS construction using a relative clause model, the proposition
proper is under a head noun that is itself argument of the main verb. In CSs
using a serial verb construction model, both the notional ''main'' verb and the
notional ''lower'' verb are in fact in the main clause in a symmetrical
construction (cf. Tariana ''nu-ira-de nu-nu dina'' lit. 'I order I come him' for
'I ordered him to come', p. 197). Dixon notes that CS structures may give rise
to real CC structures (cf. the English ''that''-CC: ''I saw {that you came}'', which
is thought to historically derive from an apposition structure ''I saw that. You
came'', where the pronoun ''that'' refers to the second clause before later
becoming a complementizer).

The formal types of CCs and CSs naturally depend on the morphosyntax of a
particular language. However, Dixon proposes a cross-linguistic semantic
classification distinguishing three relevant types of complementation clauses.
In the Fact type, the complementation clause refers to a fact and tends to be
most similar to a main clause (cf. p. 17 ''{that John was an academic} didn't
matter''). In the Activity type, the complementation clause describes an activity
going on and tends to be similar to NPs (cf. p. 25 ''I remember {seeing him last
Friday}''). Finally, in the Potential type, the complementation clause refers to
a hypothetical event and tends to have properties placing it between clauses and
NPs (cf. p. 28 ''I would like {to go}''). These types can all appear distinct in a
given language or they can conflate to two or even to one type. Many languages,
however, have a plethora of CCs and CSs, so that a semantic type can be
expressed by more than one formal construction. For English, Dixon discusses
four main forms of CCs: ''that''-clauses (p. 4), infinitive clauses (p. 6),
gerundive nominalization (i.e., the ''-ing'' form without ''of'', p. 15), and
interrogative clauses (p. 26).

Given this complex situation on the side of the complementation clauses (a given
language may have many formal types of CCs, many formal types of CSs, or a
combination of CCs and CSs), the issue of the selection of a particular
complementation type in a given context is a central theme throughout the book.
A key factor determining the form of the propositional complement is the
identity of the main verb and, for this reason, Dixon establishes a
classification of the propositional complement-taking verbs. He claims that all
these verbs belong to a restricted set that comprises three groups. He first
distinguishes between Primary verbs and Secondary verbs. The Primary verbs
branch into Primary-A verbs (that do not allow a propositional complement and
are hence out of discussion, cf. 'break') and Primary-B verbs, which just allow
a propositional complement. The Primary-B verbs encompass four sub-groupings:
Attention verbs ('see', 'smell', 'find', etc.), Thinking verbs ('consider',
'imagine', 'suppose', 'know', etc.), Liking verbs ('love', 'regret', 'enjoy',
etc.), and Speaking verbs ('report', 'promise', 'order', etc.) By contrast,
Secondary verbs do not just allow but require a propositional complement. They
branch into three sub-groups. The Secondary-A verbs require the same subject
between main and complement clause and include phasal verbs ('begin',
'continue', 'finish'), ''try''-verbs, negators ('not') and modals ('can',
'should'). The Secondary-B verbs allow a same or different subject in main &
complement clause (with omission of same subject in complement clause). They
include verbs such as 'want', 'wish', 'hope', 'plan', 'intend', and 'pretend'.
Finally, the Secondary-C verbs normally require a different subject in main and
complement clause (or they can allow a same subject, in which case it cannot be
omitted in the complement clause). Secondary-C verbs include 'make', 'force',
'cause', 'let', and 'help'. It should be noted that some verb types (or
subtypes) may be lacking in a language if their function is taken over by
affixes, operators, or periphrastic constructions (such as when in Dyirbal,
which lacks Secondary-C verbs, one cannot literally say ''he made me laugh'' and
one has to always specify the inducing action, for example, ''he told me a joke
and I laughed'', in which case, of course, there is no syntactic complementation,
cf. p. 264).

Despite some general correlations between types of verbs and the three semantic
types of CCs, the selection of a semantic type of CC (to say nothing of a
particular complementation construction in a language) depends on the particular
verb at hand (i.e., different verbs of a grouping even at the lowest level –
say, ''thinking'' verbs - may select different semantic types of CCs; in fact,
many individual verbs take more than one semantic type of CC). For example in
English, we find all three semantic types of CCs occurring in each sub-group of
the Primary-B verbs, except the Attention verbs, which only take Fact and
Activity CCs (cf. p. 27). Among the Secondary-A verbs, the phasal verbs take the
Activity type CC and the ''try''-verbs the Potential type CC. The Secondary-B
verbs take the Potential and the Fact type of CCs, while the Secondary-C verbs
are restricted to the Potential type. Dixon claims (p. 31) that these
correlations should, on the whole, hold for other languages as well.

The eleven case-study chapters overall adopt the framework laid out by Dixon. In
Chapter 2, ''Complement Clause Types in Pennsylvania German'', Kate Burridge shows
that Waterloo County Pennsylvania German (Germanic, USA) has seven kinds of CC
constructions and no CS. Looking at the table on p. 65, one sees that seventeen
Primary and Secondary verbs (out of twenty one verbs listed) take two or even
three semantic types of CCs and it is rather difficult to draw a pattern. It
should be noted though that when a verb takes more than one CC construction, the
meaning may be different. For example CCs introduced by the complementizer FER
'for' describe events that are less real in contrast to CCs introduced by ZU
'to' (cf. p. 67 ''ich hab thirty kieh {zu melken} 'I have thirty cows to milk
(next)' vs. ''ich hab thirty kieh {fer melken} '[then/after that] I (would) have
thirty cows to milk' [translations adapted by the reviewer]). Burridge also
shows that the FER CCs are gaining ground at the expense of the ZU CCs, and this
is perceptible over just a few decades.

In Chapter 3, ''Complement Clause Types in Israeli'', Ghil'ad Zuckermann starts
with a particularly handy overview of Israeli grammar (syntactic order, noun,
verb, and clause categories) and claims that this language has six kinds of CC
constructions (with criteria for argument status) and no CS (the author uses the
term ''Israeli'' as an exact equivalent for ''Modern Hebrew'' and thinks that the
language is mixed Semitic and Indo-European, cf. p. 73). The most frequent CC,
introduced by the complementizer SHE (derived from an Old Hebrew relativizer),
can by itself express Fact, Activity, and Potential semantics and combines with
almost all the listed Primary-B verbs (sixty-one verbs out of sixty-four listed,
cf. p. 87ff) and almost all the given Secondary verbs (twenty-five verbs out of
thirty-two listed). The majority of the Primary-B verbs take at least four kinds
of CCs.

In Chapter 4, ''Complement Clause Type and Complementation Strategy in Jarawara'',
R.M.W. Dixon shows that Jarawara (Arawa, Brazil) has just one CC and a second
complementation construction the status of which is not clear. This language has
an AOVX (or OAVX) word order and the CC appears in the normal position, as O or
S, before the verb. The CC expresses Activity and Potential semantics and
combines mostly with phasal Secondary-A verbs and with intransitive verbs (verbs
of motion and stance, stative descriptive verbs, quantity verbs) in
constructions that in most other languages do not use a complementation
structure (for example, a speaker would literally say ''your talking is good'',
with a complementation construction, for ''you talk well'', cf. p. 110). The CC
may also appear in O function with Attention, Thinking, and Liking verbs and
also with the Secondary verbs 'want' and 'try'.

In Chapter 5, ''Complement Clause Types and Complementation Strategy in White
Hmong'', Nerida Jarkey says that White Hmong (China, affiliation undecided) has
three CCs expressing the three semantic types of CCs, and one CS. The three CCs
appear as argument of most Primary-B verbs and the Secondary-B and C verbs. By
contrast, the CS, which is based on a serial verb construction, only appears
with Secondary verbs, in particular with the Secondary-A verbs. As usual,
particular verbs may appear with more than one type of CC, with different
meanings. For example, the verb ''xav'' meaning 'think, want' is disambiguated in
context: It means 'think' with the complementizer TIAS 'that' and 'want' with
the complementizer KOM 'to' (cf. p. 131ff).

In Chapter 6, ''Complement Clause Types and Complementation Strategy in Dolakha
Newar'', Carol Genetti proposes four kinds of CCs and one CS for Dolakha Newar
(Tibeto-Burman, Nepal). One reason to take the CCs as arguments of the main verb
is the fact that the language has strict transitivity, i.e., apart from very few
labile verbs, a verb is either transitive or intransitive (this type of argument
also applies to White Hmong in the previous chapter). The CCs appear like
regular NPs in preverbal position in an AOV word order. Primary-B verbs tend to
occur with a simple nominalized CC form expressing Activity semantics and a
nominalized clause introduced by a complementizer KHA that expresses Fact and
Potential semantics. The two other CC constructions occur with fewer verbs (cf.
p. 147). The CS, based on an auxiliary construction, is used mostly with
Secondary verbs (cf. p. 156).

In Chapter 7, ''Complement Clause Types and Complementation Strategies in
Akkadian'', Guy Deutscher examines Old Babylonian Akkadian, a Semitic language
(Ancient Mesopotamia, 2000-1500 BC) with an AOV (or OAV) word order. Deutscher
claims that the language has two CCs and seven CSs. One of the CCs is introduced
by the complementizer KIMA (derived from a conjunction meaning 'as, like,
instead of, when, because') which used to be positioned before the main verb but
migrated to post-verbal position by 1700 BC (cf. p. 162). This CC expresses Fact
semantics. The second CC is an infinitive clause expressing Activity and
Potential semantics. The CSs include two varieties of a purposive construction,
a syntactic nominalization, an apposition structure, an ''as you know''
construction, a relative clause and a direct speech construction. The Primary-B
verbs take both CCs and all seven CSs. Secondary-A and C verbs tend to take the
infinitive CC, while the Secondary-B verbs take the purposive CS. Another
construction (not included among the seven CCs) is used by Secondary phasal
concepts and the verb 'try' (for example, the sentence for ''give him the field
in its entirety'' is literally ''finish to him and give him the field''; cf. p.
176). The verb 'finish' also appears in a construction with the complement verb
in oblique position and its subject raised to the main clause (for example, the
sentence for ''crushing the sesame is finished'' would literally be ''the sesame is
finished in crushing''; cf. p. 175). According to Deutscher, Akkadian is overall
poor on propositional complement-taking verbs (cf. p. 173).

In Chapter 8, ''Complement Clause Types and Complementation Strategies in
Tariana'' Alexandra Y. Aikhenvald examines Tariana (Arawak, Brazil), which has
four CCs and four CSs. All the CCs are real core arguments of the main verb and
can be replaced by regular NPs. At least one of the CCs (the interrogative
pronoun-marked CC) is influenced by Tucano or Portuguese (cf. p. 192). It is
also claimed (p. 202) that an apposition construction can replace any type of
complementation (in the other papers, apposition is counted as a complementation
strategy). The general tendency is that the Primary-B verbs combine with the
CCs, while the Secondary verbs prefer the CS (in particular the CS based on the
serial verb construction, cf. p. 196f).

In Chapter 9, ''Complement Clause Type and Complementation Strategies in Goemai'',
Birgit Hellwig defines one CC and six CSs for Goemai (Chadic, Nigeria). The CC
is introduced by a complementizer GOEPE (and variants, derived from a locative
marker meaning ''at place'') and which expresses only Fact semantics. According to
Hellwig, the CC used to be peripheral and was later integrated into the O
position in a recent grammaticalization process (cf. p. 221). The CSs are based
on serial verb construction, nominalization, reported speech, purposive,
sequential, and consequential constructions. All Primary-B verbs tend to take
the CC, although they also take all the CSs (depending on the sub-group) with
different meanings (for example, the CS may express a personal evaluation while
the CC expresses a fact; cf. p. 217). The Secondary verbs combine with four CSs

In Chapter 10, ''Complement Clause Type and Complementation Strategies in
Matses'', David Fleck reports on Matses (Panoan, Peru and Brazil) and claims that
the language has just one CC used only with one verb, the verb 'want'. This CC
is marked by a complementizer TE and is not an O argument but an extended
argument that appears pre-verbally in a ''S-Extended O-V'' word order. Indeed, in
this language the verb 'want' is intransitive (the 'wanter' argument is marked
absolutive). The CC expresses Potential semantics. The rest of the propositional
complement-taking verbs combine with a CS based on adverbialized clauses formed
with twenty adverbializing suffixes the basic meanings of which include 'while,
after (temporal, inferential, and experiential), when, until, before, in order'.
Polysemous verbs such as ''tantia'' 'hear, listen, know how, know that,
understand, think, believe, remember' disambiguate their semantics through the
choice of the adverbializing suffix (cf. p. 243). The Polysemous verbs and the
fact that morphology and operators have taken over many verb meanings are the
reasons why Matses has few propositional complement-taking verbs (cf. p. 243).

Chapter 11, ''Complement Clause Type and Complementation Strategy in Kambera'' by
Marian Klamer describes Kambera (Austronesian, Indonesia), which has one CC and
one CS. That the CC is a core argument is shown by the fact that it is
cross-referenced on the main verb as S or O and appears in the same position as
regular NP arguments. The CC is used with Primary-B Attention and Thinking
verbs. The rest of the verb types combine with a CS based on a ''controlled
clause'' that is not an argument of the main verb. Nonetheless, this CS must
share at least one argument with the main verb. As in the other languages, some
verbs appear with the CC or the CS to modulate their semantics (for example, the
verbs ''pingu'', ''njadi'', ''monung'', which with the CC mean, respectively, 'know
about', 'be possible, appropriate', and 'trust', but with the CS they mean,
respectively, 'be able, can', 'be able', and 'hope'; cf. p. 250f). Besides the
CC and the CS, the author also mentions a quotative and a coordination
construction as possible CSs.

In Chapter 12, ''Complementation Strategies in Dyirbal'', R.M.W. Dixon discusses
Dyirbal (Austronesian, Australia), the only language in the volume without a CC
and which relies on three CSs (in this respect, Dyirbal is like many other
Australian languages; cf. p. 263n1). The three CSs are based on a purposive, a
relative clause, and a serial verb construction. All three CSs are used by the
Primary-B verbs (relative clause with Attention, Thinking, Speaking, and a few
Liking verbs; purposive with some Speaking verbs and the Liking verbs). The
Secondary-A verbs use the purposive CS, while only the Secondary-B verbs use the
serial verb construction (Dyirbal has no Secondary-C verbs). Despite these
correlations, some verbs with related meanings may take different CSs. For
example (cf. p. 275), the verb ''gigal'' 'order, let do' takes the purposive CS
while its converse ''jabil'' 'stop from doing, prevent' takes the relative clause CS.

In conclusion, the case-study chapters evidence a variety of patterns with
regard to the availability of the verb types, the CCs, the CSs, and their
interrelations. One can probably paraphrase Dixon (cf. p. 278) and say that the
strategy used by any particular verb in a language relates to the meaning of the
verb and the meaning of the types of complementation structures available in the
language. It should also be noted that complementation structures are subject to
grammaticalization processes, which by definition are progressive, and
equivalent constructions may receive different status (as CCs or CSs) in
different languages (cf. the sometimes varying treatment of apposition,
nominalization, and direct speech constructions in the chapters). Nonetheless,
it is also clear that Dixon's framework provides an excellent canvas for
exploring complementation in languages. More than this review can reflect, the
chapters are also a treasure of language-specific constructions or features and
recurring patterns. One of these patterns is ''raising'', which is evidenced in
many languages studied in the volume, and there seems to be some beginning of a
new insight into the nature of the phenomenon (according to Dixon, cf. p. 23,
the argument raised to object simultaneously belongs to the main and the
subordinate clause).


Besides the theoretical framework proposed, this volume has benefited from the
firsthand experience of the contributors, each with his or her subject language.
They all display a good familiarity with speaker communities and the
socio-linguistic situation of the language (except, naturally, for Akkadian).
Sometimes, this knowledge is the basis for proposed explanations, such as when
Burridge says that the humility of speakers towards God may have fueled the
spread of the FER-ZU complementizer (which presents events as less real) at the
expense of the ZU complementizer (which presents events in a more confident
way). It is also shown how endangered languages or languages otherwise subject
to pressure or influence may borrow or calque complementation constructions from
other languages (cf. the case of Tariana and Pennsylvania German).

There is however a potential point of criticism that may relate to the choice of
the case-study languages. The issue is that for the ten languages (including
English) reported to have CCs, the authors are unequivocal that these CCs are
core arguments of the main verbs, on par with regular NPs. This may be too
strong an assertion. There is evidence in Hausa (Chadic, Niger, Nigeria) and
Zarma Chiine (Songhay, Niger) that CCs can have an unstable status as core
arguments or at least score lower on an objecthood scale than a regular NP. In
Hausa, both regular object NPs and object CCs induce a ''plus direct object'' form
of the verb or a possessive linker on a noun (cf. ''sun neemi yaara'' 'they
looked.for[+DO] the.children'; ''sun neemi {su fita}'' 'they tried[+DO] {they go
out}'; ''naa yi maamaaki-n Abdu'' 'I was surprised about Abdu', lit. 'I did
the.surprise-of Abdu'; ''naa yi maamaaki-n {da yaara suka fita}'' 'I was surprised
that the children went out', lit. 'I did the.surprise-of {that the.children they
go.out}'; for details on the Hausa verbal system see Newman 2000 or Jaggar
2001). On these grounds, and taking into account Dixon's criteria for a CC (cf.
p. 15), one can say that Hausa has a CC. Nonetheless, contrary to the regular
NP, the CC may fail to induce the ''plus direct object'' form of the verb or the
possessive linker (for example, one may have as alternative the sentence ''sun
neemaa {su fita}'' lit. 'they tried[-DO] {they go.out}'). In Zarma Chiine, some
regular object NPs that are not patient or theme tend to appear post-verbally in
an AVOX word order (cf. ''Abdu ga di zankay'' lit. 'Abdu will see the.children').
Patient or theme objects can appear pre-verbally in an AOVX word order or
post-verbally (cf. ''Abdu ga tira dey'' lit. 'Abdu will buy' and ''Abdu ga
dey tira'' lit. 'Abdu will buy'). Abdoulaye (2008) presents a range of
evidence showing that the pre-verbal object is more affected than the
post-verbal object. It happens that object CCs in Zarma Chiine can only appear
in post-verbal position (cf. the sentence ''Faati ga naŋ {zankey ma koy habu}''
lit. 'Fati will let {the.children TAM go market} or its raising version ''Faati
ga zankey naŋ {I ma koy habu}'' lit. 'Fati will the.children let {they TAM go
market}). There are probably many other languages where CCs (by the relevant
criteria) nonetheless have a lesser direct object status than regular NPs. For
example, in Dryer (1980) it is shown that cross-linguistically, CCs tend to be
shifted towards the end of the sentence, while regular NP objects stay in
position. A discussion of this aspect of the CCs (cf. the CC's migrating to
final position in Akkadian) could have some implications in the appreciation of
complementation structures in languages.


Abdoulaye, Mahamane L. (2008) L'alternance ditransitive en zarma. Hyper Article
en Ligne (HAL). Available at :
(accessed on October 10, 2009)
Dixon, R.M.W. (1995) Complement clauses and complementation strategies. In: F.R.
Palmer, Ed., Grammar and meaning: Essays in honor of Sir John Lyons, pp.
175-220. Cambridge: Cambridge University Press.
Dryer, Matthew S. (1980) The positional tendencies of sentential noun phrases in
universal grammar. Canadian Journal of Linguistics 25: 123-195. Available at: (accessed on April
1, 2010)
Jaggar, Philip J. (2001) Hausa. Amsterdam/Philadelphia: John Benjamins.
Noonan, Michael (1985) Complementation. In: Timothy Shopen, Ed., Language
typology and syntactic description, Vol. 2, pp. 42 141. Cambridge: Cambridge
University Press.
Newman, Paul (2000) The Hausa language: An encyclopedic reference grammar. New
Haven/London: Yale University Press.

Mahamane L. Abdoulaye teaches linguistics at the Abdou Moumouni University, Niamey. His main research focuses on Hausa and Zarma Chiine morphology, syntax, and semantics.

Format: Paperback
ISBN: 0199556660
ISBN-13: 9780199556663
Pages: 304
Prices: U.S. $ 45.00