Exocentric compounds in Akan

ABSTRACT

The purpose of this paper is to study exocentric compounds in Akan and to present a

Constructions Morphology account of their properties. I show that the complex issue of

exocentricity is better accounted for if we distinguish between formal and semantic

exocentricity. I present Bauer’s (2008, 2010) typology of exocentric compounds (bahuvrihi,

exocentric synthetic, transpositional exocentric, exocentric co-compounds and metaphorical

exocentric compounds) and test the Akan data against them, showing that all Akan compound

types (N-A, N-N, N-V, V-N, and V-V) have exocentric subtypes which fall into three of the

five types identified by Bauer. They are bahuvrihi compounds and exocentric synthetic

compounds of which I identify two subtypes each and transpositional exocentric compounds.

I present the properties of each identified type of Akan exocentric compound using

formalisms from Construction Morphology (Booij 2010b) in which compounds are word-

level constructions capable of having holistic properties. The unexpressed features of the

exocentric compounds are regarded as holistic constructional properties that do not emanate

from the constituents per se.

1. Introduction

One of the main criteria for classifying compounds is the presence of a constituent that shares

formal and/or semantic properties with the whole compound. Based on this criterion, two

principal types of compounds are identified, as exemplified in (1). A compound that has a

head element in it (1a) is called an endocentric compound whilst one that has no head element

in it (1b) is called an exocentric compound (Bloomfield 1933). Endocentric compounds are

defined as “compounds which are hyponyms of their head element” (Bauer 2010: 167).

1  
 
Exocentric compounds, on the other hand, are usually defined as the class that is left after

endocentric compounds have been removed (cf. Scalise and Guevara 2006: 192).

(1) a. orange juice from orange and juice

b. pickpocket from pick and pocket

Endocentric compounds are very common cross-linguistically and have received a lot of

research attention. Exocentric compounds, on the other hand, are not so common. As Dressler

(2006) remarks, exocentric compounds are marked relative to endocentric compounds; they

are less common cross-linguistically and in individual languages and less common in the

word classes in which they occur. They are also marked with regard to the classes of words

they form. Bauer (2008: 71) captures the observed general lack of productivity of exocentric

compounds succinctly when he remarks that “in very few languages is the formation of

exocentric compounds a productive method of word-formation”. However, it has been

claimed that in some languages, including Turkana (Dimmendaal 1983), Seediq (Holmer

1996) and Kayardild (Evans 1995), exocentric compounds appear to be the most frequent

compound type. Again, in Italian, the most productive type of compounding is exocentric (cf.

Bauer 2010: 174).

Given this situation, exocentric compounds receive no more than passing mention in the

discussion of compounds in individual languages. Indeed, they are severely under-described

even in languages where they are clearly attested. Scalise and Guevara (2006: 185), thus,

characterise exocentric compounds as “a rather unexplored aspect of compounding

phenomenon.”

Theoretical and typological studies of exocentric compounds are also rare. Exceptions

in this regard are Scalise and Guevara (2006) and Bauer (2008, 2010). This dearth of cross-

linguistic studies of exocentric compounding is clear from the fact that one of the aims of

2  
 
Bauer (2008: 51-52) was to “provide a (provisional) terminology for discussing them

[exocentric compounds].” Again, Bauer (2010: 168) regards his study as “a necessary

precursor” to a more detailed study.

Even these typological studies are severely restricted in the languages they cover.

Scalise and Guevara (2006), for example, base their study on data from seventeen languages

of which only two (Chinese and Korean) are spoken outside of Europe (cf. Scalise and

Guevara 2006: 204, n. 3). Bauer’s (2008, 2010) studies are relatively more extensive,

covering data from about fifty typologically diverse languages collected from descriptive

grammars.i

The reasons for the lack of attention to exocentric compounds are varied. For general

theoretical studies of compounding, one may suggest that exocentric compounds pose a

puzzle because they are not easy to account for within the most popular theoretical models

which assume that the properties of compounds percolate from their constituents, especially

the head constituent (cf., inter alia, Lieber 1980, 1983, 1992; Scalise 1984). For such models,

the absence of a head constituent may be described as a source of embarrassment, leading

proponents to posit zero morphs in order to account for missing properties of the compounds.

For descriptive works on compounding, however, one gets the impression, on reading

Bauer (2008, 2010), that researchers simply do not pay close enough attention to exocentric

compounds because they are not very common anyway or probably because they

inadvertently get classified with endocentric compounds. As Bauer (2008: 55) puts it;

potentially exocentric constructions are not necessarily described as such, even in good
language-descriptions or in descriptions which focus on word-formation. It is
accordingly difficult to be sure that other languages do not also have a wider range of
exocentrics than can be determined from reading the grammars of those languages.

3  
 
The present study, therefore, contributes to an area of contemporary linguistic morphology

that is in need of attention, by showing that Akan (Kwa, Niger Congo) has various subclasses

of compounds that may be properly classified as exocentric compounds.

We do not know very much about exocentric compounds in Akan and it seems to me

that there are two main reasons for this. The first reason, which accords with Bauer’s

observation above, is that researchers have simply not paid attention to exocentric compounds

in Akan. This is because, as expected, we discover instances of exocentric compounds in the

oldest grammar of Akan (Christaller 1875), although they are not so-named. Christaller

(1875: 25, 29) identified what he called spurious compounds which he contrasted with

genuine compounds. About the spurious compounds, he writes:

Spurious compounds are also made up of two or more words, but none of them expresses
the genus of the thing of which the whole compound denotes a species; such a
compound, as it were, occupies the place of the qualifying word, and the general notion
(person or animal, or thing) is left to be supplied (Christaller 1875: 25).

The second reason is that research on Akan compounds have been generally focused on issues

in the phonology of compounds, mostly tonal perturbation in compounds (Abakah 2005,

2006; Obeng 2009; Marfo 2004; Dolphyne 1988; Abakah 2004; Anderson 2013; Anyidoho

1990) and vowel harmony (Dolphyne 1988), leaving out the grammatical and semantic

properties.

In the rest of the paper, I briefly introduce Construction Morphology and the formalism

to be employed in the representation of the properties of exocentric compounds (section 2).

For a proper understanding of exocentricity, the notion of headedness in compounding is key.

I discuss this in section 3, focusing on what will be essential for the issues at hand, since

headedness has been thoroughly discussed in the literature (cf., inter alia, Williams 1981;

Selkirk 1982; Scalise 1984; Zwicky 1985; Di Sciullo and Williams 1987; Bauer 1990; Scalise

4  
 
1992; Štekauer 2000; Arcodia 2012). In section 4, I take a closer look at the literature on

exocentricity, noting how it has been defined. Here, I argue, based on the literature (Scalise

and Guevara 2006; Plag 2003; Bauer 2008, 2010), that we have to distinguish between formal

exocentricity and semantic exocentricity.

Bauer (2008, 2010) provides criteria for exocentricity based on which he identified five

types of exocentric compounds. In section 5, I present Bauer’s criteria as well as the

properties of the five types of exocentric compounds he identifies (bahuvrihi, exocentric

synthetic, transpositional exocentric, exocentric co-compounds and metaphorical exocentric

compounds). I then apply the criteria to Akan compounds, showing which of Bauer’s types

are attested in Akan and what other types, if any, may be added to Bauer’s Typology (section

6). This is where I employ Construction Morphology formalism in modelling the properties of

Akan exocentric compounds. Section 7 is the conclusion.

Data for this study were collected from a variety of written texts, including a children’s

reader on fishing, an Akan translation of Plato’s Apology of Socrates and a wordlist from

Christaller’s (1933) dictionary. I also elicited data from native speakers at the University of

Ghana and also provided some examples, as a native speaker – names of various

items/entities, plants, characters in folktales, etc., that, upon careful considerations, sounded

exocentric.

2. Construction Morphology

Construction Morphology (CM) “aims at a better understanding of the relations between

morphology, syntax and the lexicon and of the semantic properties of complex words” (Booij

2010a: 543). The main tenets of CM are a theory of the notion construction, a theory of word

structure and a theory of the lexicon.

5  
 
 In CM, the notion construction, as developed in Construction Grammar (Goldberg

1995, 2006; Michaelis and Lambrecht 1996), is adopt to develop “a framework in which both

the differences and the commonalities of word-level constructs and phrase-level constructs

can be accounted for” (Booij 2010b: 1). Thus, in CM, words are morphological constructions

– word-level form-meaning pairs that may have holistic properties – and they are formed by

schemas that generalize over sets of existing complex words and also serve as recipe for

forming new ones. The schema in (2), for example, generalizes over right-headed compounds

and also serves as a pattern for forming other right-headed compounds.

(2) < [[a]Xi [b]Yj]Yk ↔ [SEMj with relation R to SEMi]k >

The upper-case variables X and Y stand for the major lexical categories (N, V & A). The lower-case
variable a and b stand for arbitrary strings of sound segments, whilst i, j and k are indexes for the
matching properties of the constituents of the compound and the compound as a whole

The schema says that given any two constituents ([X]i and [Y]j) the compound so formed will

be a types of “Y” ([Y]k). The variables a and b represent arbitrary phonological strings.

Schemas and the constructions that are formed by them co-exist in the lexicon, a

structured repository of connected complexes comparable to a map, where two types of

relations hold – “instantiation” and “part of” relations. Instantiation holds between a schema

and a word that is formed by the schema and part of obtains between a complex word and its

constituents. For example, right-headed compounds may be of any syntactic category.

Therefore, in a language with right-headed nominal compounds like Akan, there would be a

specific instantiating schema with the variables a and b, substituted by the syntactic category,

in this case noun (N), as shown in (3).

(3) < [[a]Xi [b]Yj]Nk ↔ [SEMj with relation R to SEMi]k >

< [[N]i [N]j]Nk ↔ [SEMj meant for SEMi]k >

6  
 
An even more specific schema with the variables substituted by specific lexical items

instantiates the general template for Akan N-N compounds (4), showing what a word formed

by the dominating schema looks like. Here, we see the two kinds of relation; the compound

àsɔ́rédáń instantiates the Akan N-N compound schema while its constituents, àsɔ́ré and dáń,

have a part of relations with the compound. The more specific schema inherits its non-unique

properties from the dominating schema.

(4) < [[a]Xi [b]Yj]Nk ↔ [SEMj with relation R to SEMi]j >

< [[N]i [N]j]Nk ↔ [SEMj meant for SEMi]k >

[[àsɔ́ré]Ni [dáń]Nj]Nk ‘church building’

[àsɔ́ré] ‘church’ [dáń] ‘building’

The lexicon in CM is structured hierarchically because complex words bear multiple

relations. For example, (4) does not stand alone as an abstract word-formation schema. It is

connected to other words in the lexicon that contain either àsɔ́ré, such as Mɛ̀tɔ́dèsè àsɔ́ré

‘Methodist church’ or dáń such as dáń-wá ‘cottage’. This creates a network of related words,

which is a conceptualisation of the lexical knowledge of the speaker of the language.

3. Headedness in compounding

In syntax, the notion head characterizes the dominant member in an asymmetrical relationship

within a construction (cf. Croft 2001: 241). The notion was explicitly applied to all

morphological constructs in the early 1980s by Williams (1981) and Selkirk (1982) and

defended strongly (cf. Štekauer 2000; Di Sciullo and Williams 1987). However, not all

linguists share the view that the notion extends naturally to all morphological constructions,

especially affix-derived words (cf. Bauer 1990; Zwicky 1985). For compounding, though, the

7  
 
notion head has existed longer and has had a central role in its study (cf. Scalise and Fábregas

2010).

As discussed in section 1, the classification of compounds may be based on the

presence of a head constituent. It may also be based on the word class of the head (yielding

nominal compound, verb(al) compound, adjectival compound, etc.) or the position of the head

in the compound, (yielding left-headed/right-headed compounds). However, there appears to

be a near-universal preference for right-headedness. This motivated Williams (1981: 248) to

propose what was initially presented as an absolute universal, the so-called right-hand head

rule which he formulated as: “[i]n morphology, we define the head of a morphologically

complex word to be the righthand member of that word”.

Later research showed that left-headedness also abounds, especially in the Romance

languages which have predominantly left-headed compounds (cf. Scalise 1984). Therefore, it

was suggested that the head position could be a parametric feature, so that the morphology of

a language is either left-headed or right headed, depending on the language (Lieber 1992;

Scalise 1992; Selkirk 1982). However, this idea of parametric variation in headedness has

been shown not to be sustainable because in many languages, including Mandarin Chinese

and Vietnamese, there are both left-headed and right-headed compounds (Ceccagno and

Scalise 2006; Ceccagno and Basciano 2009; Packard 2000). A language like Nizaa (Benue-

Congo, Cameroon) even has an almost equal number of right-headed and left-headed N-N

compounds (Pepper 2010). Appah (2013a) has shown that Akan endocentric N-N compounds

are mostly right-headed, but there are left-headed and dual-headed N-N compounds as well.

N-A compounds, however, are all left-headed.

Sometimes, depending on which constituent shares what properties with the compound,

a distinction is drawn between a formal head and a semantic head, which may not necessarily

be coextensive (cf. Abakah 2004; Katamba and Stonham 2006; Scalise and Guevara 2006;

8  
 
Selkirk 1982). This is especially the case in constructivist approaches (Lieber 1980, 1989;

Selkirk 1982; Lieber 1992), where the head determines all properties of the compound

through the mechanism of feature percolation (see Blevins (2006) for the distinction between

abstractive and constructive models).

The semantic head is the constituent that shares its lexical conceptual information with

the whole compound, so that the compound will be a hyponym of the semantic head (Scalise

and Guevara 2006: 190). That is, the class of elements denoted by the compound will be a

subset of the class of elements denoted by the head of the compound. For example, table mat,

is a type of mat, the head. The formal head is the constituent that shares its formal features

(lexical category and subcategorization frame) with the whole compound, so that the whole

compound has the same distributional properties as its formal head (Scalise and Guevara

2006: 190). That is, mat can occur where table mat is expected to occur without a drastic

change in the meaning of the construction, but the same cannot be said about the non-head

constituent table occurring alone where table mat is expected. Booij (2012: 79) sums this up

nicely in the observation that “headedness of a compound is not only relevant for its formal

properties, but also for its semantic interpretation.”

Other scholars distinguish between a semantic head, a syntactic head and a

morphological head. Dressler (2006), for example, argues that the compound pickpocket has

no semantic head because the referent is not named in the compound itself. Pick is the

syntactic head because it selects pocket as its internal argument. The morphological head is

pocket because when the compound is pluralized, the plural marker attaches to pocket (as in,

pickpocket-s not pick-s-pocket). Dressler (2006: 32-33) again argues that even languages that

have either predominantly or exclusively left-headed compounds may have the tendency to

mark inflectional categories on the right constituent. However, the problem, as discussed in

9  
 
the literature (Bauer 2010, 2008; Booij 2012), is that the position of the inflectional element

may be a default in the language and not necessarily identify a morphological head.

4. Characterizing exocentricity and exocentric compounds

The property of exocentricity is regarded as an anomaly in language design because its

presence implies that we are unable to account for all the properties of some complex unit.

However, this aberrant quality is claimed to be a defining property of compoundhood because

it is assumed that there are no exocentric derived words (Scalise and Guevara 2006; Scalise;

Bisetto and Guevara 2005). This is captured in Bisetto and Scalise’s (2007) formulation of the

so-called exocentricity principle, in (5).

(5) Exocentricity principle

In morphology, exocentricity is an exclusive property of compounds
(Bisetto and Scalise 2007: 363)

This principle does not capture the whole truth, however, as Bauer (2008: 54-55) has argued

that studies have shown that the effect of exocentricity may be realized through both

affixation (Haspelmath 1993; Ralli 2013) and conversion (Schönefeld 2005). Booij (2012: 82)

also holds the view that the so-called exocentricity is just a special use which “is not unique

for compounds”. He argues:

When a teacher wants to address a pupil whose name she does not now, she might say:
That red sweater should shut up, meaning that the pupil with the red sweater should
keep his mouth closed. Hence, this use of compounds appears to be a conventionalized
use of words, in which the word for a part denotes the whole (pars-pro-toto)
(Booij 2012: 82).

Because of the wide variety in the formal and semantic properties of exocentric compounds, it

has to be noted that Booij’s observation holds true for only a subset of exocentric compounds.

10  
 
This will become clear from the discussion of Bauer’s (2010, 2008) typology of exocentric

compounds in section 5.

To come up with a characterisation that captures the wide variety of exocentric

compounds, both formal and semantic properties of the compounds have to be taken into

account. Thus, Scalise and Guevara (2006: 192-193) base their definition of exocentricity on

the distinction between formal head and semantic head, arguing that “neither the notion of

formal head, nor the notion of semantic head, considered separately, suffice to define a

compound as endocentric or exocentric.” They observe that an endocentric compound will

have at least one formal head and one semantic head which must be coextensive. If a

compound realizes any other possibility, so that, in a compound with one formal head and one

semantic head, the two do not coincide, the compound will be exocentric. Hence, Bauer

(2010: 174) observes that exocentric compounds have to be defined negatively, as the

remnant that is left after a well-defined group (endocentric compounds) has been removed.

Scalise and Guevara’s approach suggests that we have to distinguish between formal

exocentricity and semantic exocentricity, based on the distinction between formal head and

semantic head respectively (cf. Scalise and Guevara 2006: 194-202). I believe that, if we paid

close attention to the literature, we would have to accept this position. For example, Booij

argues that:

what have been called exocentric compounds or bahuvrihi compounds are a specific
semantic category of endocentric compounds based on metonymy: a part of an entity is
used to refer to the whole entity. This is what is at stake with the classical English
example of a Bahuvrihi compound redskin. (Booij 2002: 143)ii

There are cases where the constituents of the so-called exocentric compound belong to a

completely different word class from the word class of the resultant compound. Here, we can

11  
 
say that even if the putative exocentric compound can be regarded as semantically

endocentric, because it can be interpreted metonymically or metaphorically, we still have to

accept that the compound is formally exocentric because its syntactic category cannot be

shown to emanate from its constituents. Such constructs legitimize the distinction between

semantic and formal exocentricity.

Bauer (2010, 2008) characterises exocentric compounds as those which fail the

hyponymy test. It would appear, then, that Bauer makes exocentricity a purely semantic

matter. However, Bauer’s formulation of the nature of exocentricity shows clearly that the

morphosyntactic notion of word class has to be taken into consideration. Bauer (2010: 167)

observes that exocentric compounds may fail the hyponymy test in a number of ways:

(a) they may fail to display a head element;

(b) they can function as a member of a word class which is not the word class of their
head element;
(c) they can have a head element of the correct class, but apparently the wrong
denotation.

Thus, a proper interpretation of Bauer’s approach to exocentricity would include acceptance

of the distinction between formal and semantic exocentricity. Various versions of this position

exist in the relevant literature, including Plag (2003: 145-246) who mentions semantically

exocentric compounds, suggesting the existence of formally exocentric compounds, as well as

Ralli and Andreou who distinguish between formal and semantic exocentricity in their work

on Greek compounds (cf., inter alia, Ralli and Andreou 2012; Ralli 2013; Andreou and Ralli

2015).

Based on the above discussion, I will characterise exocentricity as either the situation

where no constituent of a complex unit can be designated as the head or the situation where

an element of meaning or form is external to the construction.

12  
 
 I observe, in this paper, that the distinction between formal exocentricity and semantic

exocentricity is crucial for defining the classes of exocentric compounds in Akan.

Deconstructing exocentricity this way makes it easy for us to see that certain compounds that

are interpretable by means of metonymy and/or metaphor may nevertheless be exocentric

because some formal property is not deducible from the constituents. This is the case for

Akan [V-V]N compounds, as will be illustrated in section 6.3.

5. Typology of exocentric compounds

As noted in the introduction, two main strands of research have attempted cross-linguistic

characterisation of exocentricity. However, only Bauer (2008, 2010) provides a typology of

exocentric compounds and the appropriate terminology for discussing them. Scalise and

Guevara (2006) consider whether the three macro-types of compounds – attributive,

subordinate and coordinate (Scalise and Bisetto 2009; Bisetto and Scalise 2005) have

exocentric versions. They also deal with the syntactic category of the input and of the output

as well as their structural types and combinations. However, for our purpose, their approach

would not be helpful. Therefore, in this section, I will present Bauer’s typology of exocentric

compounds, setting the stage for section 6, where I present exocentric compounds in Akan.

The types posited by Bauer (2008, 2010) are bahuvrihi, exocentric synthetic, transpositional

exocentric, exocentric co-compounds and metaphorical exocentric compounds. I discuss them

in turn.

5.1. Bahuvrihi compounds

In Bauer’s typology of exocentric compounds, bahuvrihi is a principal type. Indeed this

Sanskrit expression is commonly used as a cover term for exocentric compounds (cf., inter

alia, Bloomfield 1933; Katamba and Stonham 2006; Marchand 1969). Bahuvrihi is the type of

13  
 
compound in which the whole is not a hyponym of either constituent. Rather, it “expresses

some salient facet of the denotatum” (Bauer 2008: 56). The compound bahu vrihi ‘much rich’

itself exemplifies this type of compound because it may refer to ‘one who/which has much

rice’ although none of the constituents names a person. This is the possessive interpretation of

bahuvrihi compounds.

Regarding the English example red-eye which does not denote anything which

possesses red eye but means “cheap whisky/overnight flight”, Bauer (2010: 167) observes

that the compounds refers to that which causes red eye – the cheap whisky or overnight flight.

Thus, from Bauer (2008, 2010), we may distinguish at least two types of bahuvrihi

compounds – the possessive type and the causer type.

In terms of form, Bauer observes that the bahuvrihi compound par excellence is made

up of an adjective and a noun which is the possessed element and often “denotes a person or

an object which has a NOUN and which is ADJECTIVE” (Bauer 2008: 56). A noun, a verb or a

quantifier/numeral may replace the modifying adjective in a bahuvrihi compound. For

example, in Table 1, we find various instances of bahuvrihi compounds in which a head noun

(marked with an asterisk) is followed by a modifier (adjective, quantifier, verb and noun).

Table 1. Examples of bahuvrihi compounds (Bauer 2010: 170)

Language Form Gloss Translation Pattern Source
Koasati nakeó-baski ear + long mule N*+A Kimball 1985: 396
Matchez ?a:t-tau:?i foot + many centipede N*+Q Mithun 1999: 468
Apache c?is-tèł body + to be wide terrapin N*+V Hoijer 1946: 76
Turkana e-wur-ù-mosi smell + rhinoceros tree sp. N*+N Dimmendaal 1983:293
Finnish partasuu beard+mouth bearded N+N*(A) Sulkala & Karjalainen 1992: 361

Bauer (2008: 58-59), observes that the exocentric analysis of bahuvrihi compounds, suggests

that there is an unexpressed head which carries at least some features of the denotatum of the

whole. Thus, for the English example, red cap which is [-human], we have to assume that the

14  
 
unexpressed feature in the head position is [+human] because the compound refers to a

human entity.iii

An alternative analysis denies that such compounds are exocentric. Rather, they are

regular endocentric compounds which are interpreted according to a figure of speech –

synecdoche. The problem I see with this analysis is that it may not be able to account for the

full range of bahuvrihi compounds. For example, as indicated in (n.ii), it is not clear how such

Chinese compounds as dōngxi ‘thing (lit. east + west)’ and jiānghú ‘vagabond (lit. river +

lake)’ may be analysed in this way. The same can be said about the Tzutujil compound

kolwachixik ‘wake up quickly (lit. basket + face)’, as hinted at in Bauer (2008: 60-61).

Finally, Bauer (2008: 59-61), observes that sometimes the bahuvrihi compound may

function adjectivally or verbally. He refers to them respectively as adjectival bahuvrihi and

verbal bahuvrihi. Crucially, in the adjectival bahuvrihis, the head is not an adjective and the

verbal bahuvrihi does not contain a verb at all. Thus, the terminology is purely inspired by

how the relevant compounds are used and so going back to the sense in which a compound

may fail the hyponymy test, we see that these compounds are exocentric because “they […]

function as a member of a word class which is not the word class of their head element”

(Bauer 2010: 167).

Determining the word class of the Akan bahuvrihi compound will not be an issue in the

discussion because compounding in Akan is a noun-forming process and the compounds are,

without exception, deployed as nouns (Appah 2013b, 2015).

5.2. Exocentric Synthetic compounds

Next in Bauer’s typology is the exocentric synthetic compound. The term synthetic

(sometimes, verbal nexus) is ordinarily reserved for compounds with deverbal nominal heads

(marked in English, for example, by -er or -ing) or deverbal adjectives (marked by -ed or -

15  
 
ing) which inherit the internal arguments of the underlying verbs. An example is shopkeeper

in which keeper (the head) contains the verb keep and the agentive suffix -er and the modifier

is shop, the object of the underlying verb keep.

This type of synthetic compound is endocentric. For example, Katamba and Stonham

(2006: 321) begin their discussion of verbal compounds with the words: “[o]ne class of

endocentric compounds, which is referred to technically as verbal compounds, stands out

from the rest in that it exhibits quite consistent semantic readings”.

In the exocentric synthetic compound, there is no morpheme like English -er, which

corresponds to the external argument of the verb. Rather, the head verb and its internal

argument form a noun that denotes the entity that performs the role of the external argument.

In the French compound gratte-ciel ‘skyscraper’ (lit. scratch-sky), only the verb and its

internal argument are present but the compound as a whole refers to the external argument –

“that which scratches the sky”.

Bauer’s examples, which are structurally quite varied, are in Table 2. The first two

examples contain head V*s and N internal arguments. The third example contains a head V*

and an adjectival subject complement. In the fourth example, the head V* has a verbal direct

object. The last example also has a V* head and an N object but the compound is an abstract

noun and not an agentive noun, making it similar to English activity-denoting compounds like

whale watching (Bauer 2010:170).

Table 2. Examples of exocentric synthetic compound (Bauer 2010: 171)

Language Form Gloss Translation Pattern Source
Babungo mè-vǝ̀lú’ swallow + eggs snake sp. V*+N Schaub 1985: 249
Japanese tsume-kiri nail + cut nail clipper N+V* Hinds 1986: 366
Damana munzisa-kuaga viscous + appear gum A+V* Trillos Amaya 1999: 71
Italian lascia-passare permit + to pass a pass V*+V Maiden and Robustelli
2000: 33
Korean sal-in kill + person manslaughter V*+N Sohn 1994: 414

16  
 
Interestingly, the constituents designated as the head of the nominal compounds are verbs.

This appears inconsistent with the result expected from applying the IS A test (Allen 1978).

That is, since the heads are verbs, it would be expected that the compounds would be verbs.

This, however, makes sense in the context of the distinction between syntactic and semantic

heads; the compounds have formal heads but lack semantic heads because, for example,

whereas the Japanese compound tsume-kiri [nail+cut] refers to the activity designated by the

verb phrase, the compound denotes the instrument for performing the activity ‘nail clipper’.

This shows that the key consideration in characterizing a compound as exocentric, in this

case, is semantic, so that, even compounds that are formally/syntactically headed can be

exocentric because they fail the hyponymy test (Bauer 2010). This lends support to the

argument for a distinction between formal exocentricity and semantic exocentricity.

The exocentric synthetic compound mostly refers to the agent of the denotatum of the

compound. However, there are instances where the compound has an activity/action rather

than agentive reading. It may also be an abstract noun (Bauer 2010: 170). Thus, an important

feature of the exocentric synthetic compound is that, it is only an aspect of the meaning of the

compound that is not expressed in the compound – the agent/instrument or the action. So,

ultimately, the meaning of the compound “can be deduced from the meaning of the verb and

its argument and the unexpressed agent or action” (Bauer 2010: 171).

5.3. Transpositional exocentric compounds

We have seen from the two types of exocentric compounds described above that usually, only

an aspect of the meaning of the compound that is not present in the compound. In the

exocentric synthetic compound, for example, it is the agentive or instrumental meaning that is

not coded in the compound. In a transpositional exocentric compound it is only the word class

of the output compound that is not overt (Bauer 2010: 171). Thus, the meaning of the

17  
 
transpositional exocentric compound is clear, only that it “functions as a member of an

unexpected word-class” (Bauer 2008: 64). This makes it an extreme case of formal

exocentricity, where there is no relation whatsoever between the word-class of the

constituents and that of the output, making the exocentric synthetic compound look like

conversion or transposition (Bauer 2008). See Table 3.

Table 3. Some examples of transpositional compounds (Bauer 2010: 172)

Language Form Gloss Translation Source
Khmer khɔh trǝw wrong + right morality Ourn & Haiman 2000: 484
Turkana ŋi-kari-mɔjɔŋ thin + old The Karimojong tribe Dimmendaal 1983: 294
Damana tua kuaga to see + to live visible Trillos Amaya 1999: 72
Vietnamese ºbà con grandmother + be related Thompson 1987: 127
child
Swahili ujauzito come + heavy pregnancy Maina 1987: 2
Mandarin zhuăn-yăn turn + eye ‘instantly’ Li & Thompson 1981: 79

As Table 3 shows, typical examples would be ones in which two adjectives are interpreted as a

noun, two nouns interpreted as a verb, a sequence of verb and adjective interpreted as a noun

and a sequence of verb and noun is interpreted as an adverb (see, Bauer 2010: 172).

5.4. Exocentric co-compounds

Bauer (2008, 2010), identified a class called exocentric co-compounds. This is the class of

coordinate compounds in which two constituents enjoy parity in terms of their importance in

the compound. See Table 4.

Table 4. Some examples of exocentric co-compounds (Bauer 2010:173)

Language Form Gloss Translation Classification Source
Chantyal Nhe thara Milk buttermilk Dairy products Co-hyponymic Noonan 2003:328
German Schleswig- additive, names
Holstein
Lezgian kar-k’walax job work job, business co-synonymic Haspelmath 1993:108
Sanskrit keśaśmasurú hair beard additive, common Burrow 1955:217
N
Old Uyghur ulug-i kičig-i big-its little-its size scalar Wälchli 2005:138
English blue-green compromise

18  
 
Bauer (2010: 172) argues that there are virtually no endocentric co-compounds. The few

attested ones are the appositional compounds like singer-songwriter which refers to a person

who is both a singer and a songwriter. The problem with this classification is that such

compounds are not endocentric in the regular sense in that the referent is not made up of a

‘singer’ and a ‘songwriter’. Rather, each constituent characterises a separate property of the

referent, and enjoys parity in sharing head-like characteristics (Katamba and Stonham 2006;

Fabb 1998).

Whilst Haspelmath (2002) treats them as exocentric compounds, Plag (2003: 147)

suggest that they are right-headed, based on the fact that the plural inflectional marker

attaches to the right constituents – song-writers/*songs-writer(s). But, as has been previously

observed, this position is not totally sustainable “since right-edge inflection is a default in

English” (Bauer 2010: 172) as evidenced by plural marking in compounds like trade-offs,

pickpockets, model Ts, mother-in-laws, etc. in which the constituent that bears the inflection is

not the grammatical head.

Finally, some scholars regard such constructs as phrases and not compounds (cf. Adams

2001: 82). However, as Bauer (2010: 172) observes, this view does not tell us anything

beyond informing us that the constituents are coordinated, thus, shifting the problem of the

characterisation of the compounds from the domain of morphology to the domain of syntax.

5.5. Metaphorical exocentric compounds

The metaphorical exocentric compounds is characterized as the most controversial type.

According to Bauer (2010) this type arises when a compound fails the hyponymy test because

the head element or the compound as a whole has a metaphorical interpretation. Examples

include dust bowl ‘an area with no vegetation’, catlick ‘quick wash’. It may also arise when

the head has metonymic interpretation (e.g., phone neck ‘pain in the neck caused by using a

19  
 
phone’, or the whole compound is metonymic (e.g., bear skin ‘hat won by certain soldiers’).

For an extensive study of such compounds, see Benczes (2006) and for the appropriateness of

the label of exocentricity for these compounds, see Bauer (2010: 174).

6. Exocentric compounds attested in Akan

In this section, I discuss the classes of Akan exocentric compounds, following the

characterisation of the types of exocentric compounds in section 5. For the present paper, I

adopt Dolphyne’s (1988: 117) characterization of Akan compounds as words made up of

“two or more stems”, including reduplicated forms. I should point out that, although Akan

compounds may be contain more than one stem, they are invariably binary-branching and the

usefulness of treating some reduplicated forms as compounds will become clear later when I

discuss bahuvrihi compounds.

In the discussion, I distinguish between the relation obtaining between the constituents

within the compound and the overall interpretation of the compound which invariably

includes some unexpressed semantic feature. I do not discuss exocentric co-compounds

because they are rare in Akan (Appah 2015), much in agreement with the fact that co-

compounds are generally rare in Africa (Wälchli 2005).

In section 6.1, I discuss bahuvrihi compounds, noting that we may distinguish between

possessive and non-possessive types. The second type of bahuvrihi compounds identified by

Bauer (2010), the “causer type” is not attested in Akan. This is the type in which the

unexpressed semantic feature is the causer of the denotatum of the compound, as exemplified

by red-eye ‘cheap whisky/overnight flight’ in which “cheap whisky”/“overnight flight”

causes the “red-eye”.

In section 6.2, I deal with exocentric synthetic compounds. I distinguish two subtypes –

participant and non-participant exocentric synthetic compounds. The former are nouns which

20  
 
refer to one of the participants in the event designated by either the verb or the denotatum of

the compound (because sometimes the event is expressed by the whole compound). The latter

refers to the action, state, manner or the fact of the event designated by the predicate.

In section 6.3, I discuss transpositional exocentric compounds, showing that it is

exemplified by Akan V-V compounds, since they are invariably nominal.

In the formal representation of the compounds, I suggest that all the compounds

ultimately instantiate the schema in (6) which generalizes over all compounds in which a

crucial semantic feature of the whole is not present in the constituents. There are extreme

cases where the meaning of the compound may not be directly related to the meaning of the

individual constituents. That is, the compound pairs a form and a meaning directly, because

the meaning is that of the construction and may not necessarily relate to those of the

constituents. The parenthesized portion of the semantic pole expresses this observation that

the meaning of the whole may be optionally related to the meanings of the constituents.

(6) < [[a]Xi [b]Yj]Nk ↔ [SEM ([SEMi | SEMj])]k >

Where the meaning of the compound can be shown to be related to the meaning of either

constituent or to their combined meaning, but the meanings of the constituents do not exhaust

the meaning of the whole, the extra-compositional feature is represented as a semantic

operator (the unindexed SEM) over the meaning of the compound, or the meaning of the

relevant constituent. This is captured by the disjunction (|).

6.1. Bahuvrihi compounds in Akan

As noted in section 5.1, Bahuvrihi is the type of compound in which the whole is not a

hyponym of either constituent. Rather, the compound refers to a salient property of the

denotatum, which is not named in the compound (Bauer 2010). The bahuvrihi compounds in

21  
 
my database may be classified in various ways. Following Bauer (2008, 2010), I distinguish a

possessive type (Table 5) and a non-possessive type (Table 6) which I discuss them in turn.

6.1.1. Possessive bahuvrihi compounds in Akan

The possessive bahuvrihi compound refers to the possessor of its denotatum. For example, the

Akan compound ìtsìr̀ kɛ̀sé ‘big head’, kɔ̀ǹtséńtséń ‘long neck’ and ànàǹtá ‘crooked legs’ refers

to the entity that possesses a big head, long neck and crooked legs respectively.

Table 5: Possessive bahuvrihi compounds in Akan

Compound Constituents Element gloss Meaning Pattern
ìtsìrk̀ ɛ̀sé ìtsíŕ + kɛ̀sé head + big person with a big head N-A
ìtsìrk̀ èlèǹkélè ìtsíŕ + kèlèǹkélè head + big person with a big head N-A
kɔ̀ǹtséńtséń kɔ́ń + tséńtséń neck + long person with a long neck N-A
hwèǹkɛ̀séɛ́ hwéné + kɛ̀séɛ́ nose +big person with a big nose N-A
àkòmàhyéẃ àkòmá + hyèẁ heart + hot a quick tempered person N-A
yèfòǹdòhwè~dóhwèè yèfóń + dòhwè~dóhwèè belly + bloated person with bloated belly N-A
náńkɔ̀ǹhwéáá náń-­‐kɔ́ń  +  hwéáá ankle + thin person with thin legs N-A
ànàǹtá à-náń + (ǹ)tá PL-leg + twin person with crooked legs N-N
àsò kɛ̀tɛ́ àsó + kɛ̀tɛ́ ear + mat person with big ears N-N
tsìrp̀ á tsíŕ + pá head + bald bald head person N-V
àsòásóá àsóá + àsóá ear + ear rabbit RED-N

In terms of the categories of the constituents and the relation obtaining between them within

the compound, we see that the possessive bahuvrihi compounds mostly have N-A structure,

where the nouns refer to body parts such as ìtsíŕ ‘head’, àsóá ‘ear’, àkòmá ‘heart’, and kɔ́ń

‘neck’. The adjectives tend to be either dimension adjectives, like kɛ̀sé ‘big’, and tséńtséń

‘long’ or physical property adjectives like hyèẁ ‘hot’ (Dixon 2004; Osam 1999).

The relation between the constituents within the [N-A]N compound may be spelt out as

“having/is”. That is, [Ni R Aj]Nk translates into “Ni having property Aj” or “Aj predicated of

Ni”. See (7).

(7) < [[N]i [A]j]Nk ↔ [SEMi having/is PROPERTYj]k >

|
< [[kɔ́ń]i [tséńtséń]j]Nk ↔ [kɔ́ń i which is tséńtséń j]k >

22  
 
There are a number of N-N structures as well. Here too, the right-hand constituent may

describe a property of the left-hand constituent, as in ànàǹtá ‘crooked legs’, in which náń

‘leg’ is said to have the property of being crooked. See (8).

(8) < [[N]i [N]j]Nk ↔ [SEMi having/is PROPERTYj]k >

|
< [[ànáń]i [ǹtá]j]Nk ↔ [ànáń i which is ǹtáj]k >

The relation between the constituents of the N-N possessive bahuvrihi compound may also

be: “Left-hand constituent is like right-hand constituents.” This is seen in àsò kɛ̀tɛ́, in which

the lefthand constituent àsó ‘ear’ is likened to kɛ̀tɛ́ ‘mat’. Thus the relation between the

constituents here may be spelt out as “is like”. That is, [Ni R Nj]Nk translates into “Ni is like

Nj”. Thus, àsò kɛ̀tɛ́ is àsó ‘ear’ which is like kɛ̀tɛ́ ‘mat’. See (9).

(9) < [[àsò]i [kɛ̀tɛ́]j]Nk ↔ [àsói which is like kɛ̀tɛ́j]k >

The only example with an N-V structure (tsìr̀ pá ‘bald person’) has a stative verb, which is

translatable into English by an adjective. In other words, this is a type of compound that may

normally be realised as an N-A/A-N compound in some languages. See (10).

(10) < [[tsìr̀ ]i [pá]j]Nk ↔ [headi which is baldj]k >

The pattern is, however, not unique to Akan because Bauer (2008: 58) lists many examples of

verb-noun bahuvrihi compounds from five languages in which the property concepts are

expressed by the verb because, in those languages, adjectives either constitute a very small

class or are non-existent (Bauer 2008: 56, 58).

23  
 
 Finally, reduplicating àsóá ‘ear’ forms the word àsòásóá ‘rabbit’.   This   pattern of

forming a word for a rabbit is somewhat comparable to how the word for mule is formed in

Koasati (see Table 1), differing only in that Akan reduplicates the word for ear to get rabbit,

whilst Koasati combines the noun, ear, and an adjective. However, this pattern of forming a

bahuvrihi compound through reduplication is not yet reported in the literature, probably

because there is no consensus on whether reduplication is a compounding process. I suspect,

though, that this may not be unique to Akan, since reduplication is employed for various

functions in the languages of the world.

With regard to the over all interpretation of the compound, the speaker must look

beyond the internal semantic relations to find the extra-compositional semantic feature, what

Bauer (2008: 58-59) calls this an unexpressed element in the head position which carries

some features of the denotatum. For all the compounds in Table 5, except the last one, we

have to assume that the unexpressed semantic feature is “possessor”. Formally, Bauer (2008:

58-59) represents this as a zero morph that carries the feature [+human]. An alternative

analysis posits no zero morph. Rather, the unexpressed semantic feature is treated as an

operator over the denotatum of the compounds (Appah 2013b, 2015), as shown in (11).

(11) < [[tsìr̀ ]i [pá]j]Nk ↔ [POSS [SEM]i which is [SEM]j]k >

I believe this alternative rightly construes the problem at hand as a semantic and not a formal

one. Thus, there is no need to posit a zero morph which will be an absolute necessity in a

source-oriented view of derivation (Zager 1981; Appah 2015).

6.1.2. Non-possessive bahuvrihi compounds in Akan

The non-possessive bahuvrihi compound is the type that does not refer to the possessor of the

denotatum of the compound. It is mostly difficult to tell the meaning of the compound by

24  
 
looking at the meaning(s) of the constituents because there is usually no link between the

semantics of the constituents and that of the referent. See examples in Table 6.

Table 6. Non-possessive bahuvrihi compounds in Akan

Compound   Constituents   Gloss   Translation   Pattern  
àbírékyíré ábɔ́dwèsɛ́ àbírékyíré + àbɔ̀dwèsɛ́ goat + beard elephant grass N-N
àbòròkyíŕábá àbòròkyíŕ + àbá oversees + seed fruit used as a bait in fishing N-N
àkòmàbɔ́né àkòmá + bɔ̀né heart + bad intemperance N-A
ànòkórɔ́ ànó + kórɔ́ mouth + one unity N-Num
àsòàhyéẃ àsóá + hyèẁ ear + hot difficulty N-A
dɛ̀ǹkyɛ̀m̀ bóɔ́ dɛ́ńkyɛ́ḿ + bóɔ́ crocodile + stone diamond N-N
ènyìkáḿ ènyí + àkáḿ eye + mark earmark (lit. eyemark) N-N
hwèǹtéáá hwéné + téáá nose + slim rosemary, a spice N-A
kòdwó bèsíá kòdwó + bèsíá Monday-born effeminate man N-N
male + woman
m̀ móá náń m̀ -móá + náń PL-animal + leg a lost course N-N
ǹtrɔ̀ ̀ bànáḿ ǹtrɔ̀ ́ bá + náḿ garden egg + fish a weakling N-N
òdwáńkɔ́kɔ̀ɔ́ òdwáń + kɔ̀kɔ̀ɔ́ sheep + red a dog (for the Kwahu tribe) N-A

In terms of the categories of the constituent of the compounds, we see that there is a head

noun that may occur on either side and a modifier that may be a noun, an adjective or a

numeral. Thus, with regard to the internal syntax, these compounds are all headed because

there is a discernable head-dependent relation within the compound. This is consistent with

the view that syntactically all compounds are headed because they are generated by the same

rules that generate endocentric compounds (cf. Katamba and Stonham 2006; Plag 2003). This

should not be interpreted to mean that there are no exocentric compound (Booij 2002; Pepper

2010), because an exocentric compound is one which fails the hyponymy test because it either

has no head or some crucial semantic feature needed for interpretation is missing.

The membership of this class is semantically diverse. One group refers to an entity by

metaphorically extending a salient property of the denotatum to the entity in question, so that

the compound has to be interpreted metaphorically. For example, in ǹtr̀ ɔ̀bànáḿ ‘weakling’

the referent’s lack of physical strength is compared to the physical properties of náḿ ‘fish’,

which is further modified by the even more fragile physical properties of cooked ǹtr̀ ɔ́bá

‘garden egg’. Similarly, in àbírékyíré ábɔ́dwèsɛ́ ‘elephant grass’ (lit., goat beard), sometimes

25  
 
referred to as ɔ́bòǹsáḿ ábɔ́dwèsɛ́ (lit. Satan’s beard), the fact of how difficult it is to uproot

the plant due to the strength of the root is what earns the plant the nickname goat’s/Satan’s

beard. Therefore, in characterizing the semantics of these compounds, we have to say that the

denotatum possesses a property that may be ascribed to the referent of the bahuvrihi

compound, albeit indirectly.

Finally, there are other Akan non-possessive bahuvrihi compounds whose meanings are

simply impossible to tell by looking at those of the constituents. They include m̀ móá náń ‘a

lost course’ (lit. animal legs), òdwáńkɔ́kɔ̀ɔ́ ‘a dog’ (lit. red sheep) and àsòàhyéẃ ‘difficulty’

(lit. hot ear). This makes such compounds behave like the English compound hotdogs

(sausage), which is neither a dog nor must be eaten hot.

The non-possessive bahuvrihi compounds may be represented formally as shown in

(12). The meaning is a constructional property, not related to the meanings of the constituents.

(12) < [[N]i [N]j]Nk ↔ [SEM]k >

[[àbòròkyíŕ]Ni [ábá]Nj]k ‘a fruit used as a bait for fishing’

To finish the discussion on bahuvrihi compounds, I should like to mention that Bauer’s (2008:

56) observation that bahuvrihi compounds typically denote people (especially as nicknames),

animals and plants, is largely confirmed by the Akan data. For example, all the bahuvrihi

compounds in Table 5 refer to humans except one which refers to an animal. A number of

bahuvrihi compounds in Akan also refer to plants, including àbírékyíré ábɔ́dwèsɛ́ ‘elephant

grass’ (Table 6). There are, however, a number of these compounds that are abstract, including

àkòmàbɔ́né ‘intemperance’, m̀ móánáń ‘a lost course’ and ànòkórɔ́ ‘unity’.

26  
 
6.2. Exocentric Synthetic compounds in Akan

There are two constructs in Akan that fit the description of exocentric synthetic compounds as

described in section 5.2. The first construct is uniquely exocentric with a characteristic tonal

melody. The second is typically analysed as endocentric in the Akan compounding literature

(Anderson 2013; Anyidoho 1990). However, Appah (2013b), working with Grimshaw’s

(1990) definition of a synthetic compounds as one in which one constituent takes an argument

which must be satisfied in the compound, argued that those compounds can be analysed as

exocentric synthetic compounds. That position is taken as a point of departure in this paper. In

discussing this, I distinguish between the action/event designated by the predicate and the

participants involved in the action. The compounds in this class may refer to either the action

or the participants (cf. Payne 1997: 225-230). Thus, I isolate two main classes of exocentric

synthetic compounds – participant exocentric synthetic compounds and non-participant

(action nominal) exocentric synthetic compounds. I discuss them in turn.

6.2.1. Participant exocentric synthetic compounds in Akan

The first class of exocentric synthetic compounds is the participant nominals. There are three

types – those that refer to the agent or the patient of the denotatum of the compound and those

that refer to the location of the action/event designated by the predicate. The agentive and

patient types have [V-N]N structure, where the noun is the internal argument of the verb. This

makes the construction look like straightforward lexicalisation of verb phrases. However, the

compounds refer to one of the participants in the action designated by the verb or to some

property that is typically associated with a participant.

In the agentive type, the compound refers to the agent of the action/event designated by

the compound, as exemplified by the compounds in Table 7. For instance, the literal meanings

of the compounds in the first and second rows are ‘fetch water’ and ‘attend funeral’

27  
 
respectively. However, the two refer to the person who fetches water and the person who

attends funerals respectively (Dolphyne 1988). Thus, there is a meaning component

(“agent/person who …”) that is external to the compound.

iv
Table 7. Agentive exocentric synthetic compounds in Akan
Compound Constituents Gloss Translation Pattern
kɔ́ǹsúó kɔ́ + ǹsúó fetch + water one who fetches water V-N
kɔ́àyíé kɔ́ + àyíé attend + funeral one who attends funerals habitually V-N
kɔ́ànyíná kɔ́ + ànyíná fetch + firewood one who fetches firewood V-N
bɔ́ɔ̀tiré bɔ́ + etíré plait + hire one who plaits hair/hair dresser V-N
bɔ́ǹsúó bɔ́ + ǹsúó hit + water professional car wash(er) / one who V-N
washes cars
díàwúó dí + àwù cause + death murderer (this is a surname) V-N
kúm̀ kɔ́ḿ kúḿ + kɔ́ḿ kill + hunger early-maturing maize variety V-N
kyέǹkwáń kyέ + ǹkwáń share soup one who shares soup V-N
hwéǹkwáń hwé + ǹkwáń sweep/collect + soup one who likes to eat foods that V-N
include soup
díàbórɔ́ dí + àbórɔ́ engage in + wickedness a malevolent person V-N
díàsɛ̀m̀ pá dí + àsɛ̀m̀ pá do + good things a benevolent person V-N
sò~sɔ̀ kànéá sɔ́ + kànéá RED~turn on + light fire fly V-N

This additional meaning component is expressed as an operator over the denotatum of the

compound, as schematized in (13), which states that the compound refers to the agent of the

denotatum of the compound – the person who fetches water.

(13) < [[V]i [N]j]Nk ↔ [AGENT [SEM]k]k >

|
[[kɔ́]Vi [ǹsúó ]Nj]Nk ‘person who fetches water’

The second type of participant exocentric synthetic compound is the patient type. It refers to

the entity that undergoes the effect of the denotatum of the compound. For example, the first

compound in Table 8 refers to the entity that suffers a misfortune rather than the one that

causes the misfortune. The other refers to an animal that is located somewhere, rather than the

one “locating” something somewhere. Again, the meaning, patient, is external to the

compound.

28  
 
Table 8. Patient exocentric synthetic compounds in Akan
Compound Constituents Gloss Translation Pattern
díàmìá dí + àmìá suffer + state of being one who has suffered a V-N
helmed in misfortune.
dáàmòná dá+àmòná sleep+hole an animal that dwells in holes V-N

To sum up, we observe that generally in these compounds, rather than mention the entity that

the compound refers to directly, a characteristic property of the entity is named – a

characteristic activity or profession or a typical place of abode in the case of the second

example in Table 8, where, although the compound refers to an animal, the animal is not

expressly named in the compound. These compounds, therefore, have to be interpreted

metonymically in that a characteristic property of the entity is used to represent the whole

entity.

The compounds can be assumed to instantiate the schema in (14) which states that the

compound refers to an entity typically associated with the denotatum of the compound, where

“being associated with” may be interpreted to mean “undergoing the effect of being helmed

in” as in dí àmìá.

(14) < [[V]i [N]j]Nk ↔ [Entity typically associated with SEMk]k

|
< [[dí]Vi [àmìá]Nj ]Nk ↔ [ one who suffers a misfortune ] >

The second example in Table 8 is slightly different; it refers to an entity that is associated with

some event/process named by the first constituent which occurs at a location named by the

second constituent. This may be represented as (15).

(15) < [[dá]Ni [àmòná]Vj]Nk ↔ [ANIMAL which habitually does [SEM]i at [SEM]j]k >

29  
 
As noted above, the class of agentive/patient exocentric synthetic compounds has a

characteristic tonal melody [High-Low*-High*(-High)]. v This is the case for all the

compounds except sò~sɔ̀ kànéávi ‘fire fly’. Thus, the difference between the compounds in

this class and the corresponding verb phrase, as can be seen in (16), is the present of this tonal

melody on the compounds and the absence of the same from the verb phrase. Note that this is

not to suggest that the compound has a phrasal provenance.

(16) Related VP Compound

(a) kɔ̀ ǹsú kɔ́-ǹsúó
go water go-water
‘fetch water’ ‘a person who fetches water’
(b) kɔ̀ àyíé kɔ́-àyíé
go funeral attend-funeral
‘attend a funeral’ ‘one who attends funerals habitually’
(c) dà àmòná dá-àmòná
sleep whole sleep hole
‘sleep in holes’ ‘an animal that sleeps in holes’
(d) kùm̀ kɔ́ḿ kúm̀ -kɔ́ḿ
kill hunger kill-hunger
‘kill hunger’ ‘hunger killer (a species of maize)’

This characteristic tonal melody has to be interpreted as a holistic constructional property, one

that uniquely identifies this class of compounds. In other words, the construction itself

imposes this tonal melody, notwithstanding the tones of the individual constituents in

isolation.

The third class of participant exocentric synthetic compounds has an N-N structure (see

Table 9). In this compound the left-hand constituent mostly names a concrete object such as

àfí(é) ‘house/home’, m̀ -mòfrá ‘children’, dùá ‘tree’, etc., and, in one instance, an activity –

àkɔ́ḿ ‘ritual dance (of the traditional priest)’. Note that àkɔ́ḿ is an action nominal derived

from the verb kɔ́ḿ ‘to perform the ritual dance’ through the prefixation of a-. The right-hand

30  
 
constituent, on the other hand, is always a locative noun like àsé ‘underside/bottom’ and dó

‘top’.

Table 9. Locative exocentric synthetic compounds in Akan

Compound Constituents Gloss Translation Pattern
àfíásé àfí+àsé house + under prison N-N
dùáásé dùá+àsé tree + under name of a town N-N
m̀ mòfráásé m̀ -mòfrá+àsé PL-child + under childhood (time) N-N
gáès-àsévii gáès-àsé guys+under place where guys meet N-N
àkɔ́ḿásé àkɔ́ḿ+àsé ritual.dance + under location of a ritual dance N-N
ḿbóádúá dò ḿbóádúá+dò fishing net tree + top place for keeping fishing N-N
nets/location of mboadua

The noun that is formed usually refers to a location in time or space. I found one example

where the compound refers to a location in time - m̀ mòfráásé ‘childhood (lit. children under)’.

Here the denotatum is not an actual location – the underside of children. There is no such

place. The only conceivable physical space will be the soles of the children’s feet. However,

that is not even a possible interpretation as far as this compound is concerned. The left-hand

constituent names children, but the compound refers to childhood – the period when some

humans could be referred to as children. It is in this sense that this compound is exocentric.

Where the compound names a location in space (physical location), it could refer to the

space for performing the activity named by the left-hand constituent, as in the case of

àkɔ́ḿásé which refers to the space where àkɔ́ḿ, the ritual dance, is happening. Here, the right-

hand constituent refers literally to the underside or bottom of the ritual dance. Needless to say

that the ritual dance itself cannot have a bottom, although the person performing the dance

can conceivably be said to have an underside, which will be the sole of the person’s feet.

Thus, àkɔ́ḿásé has to be regarded as exocentric because the compound does not refer to just

where the feet of the dancer treads but the whole setting, including the people watching the

dance. The properties of the compound may be represented as shown in (17).

31  
 
(17) < [[X]Ni [Y]Nj ]Nk ↔ [ LOC [SEM]i ]k >
|
< [[àkɔ́ḿ]Ni [ásé]Nj ]Nk ↔ [ LOCATION for àkɔ́ḿi ]k >

Finally, in other cases, the right-hand constituent actually names a part of the left-hand

constituent which is a physical location. However, as their meanings show, what the

compound actually refers to is outside of the compound. This is the case in àfíásé ‘prison (lit.

the bottom of a house)’, dùá- ásé ‘name of a town (lit. bottom of a tree)’ and ḿbóáduá dò

‘place for keeping fishing nets (lit. fishing-net treetop)’.

6.2.2. Non-participant (action nominal) exocentric synthetic compounds in Akan

As indicated in section 6.2, aside from the subclass of participant exocentric synthetic

compounds, there is a class of compounds which previous scholars analysed as endocentric

but which Appah (2013b) argued that they could be analysed as exocentric synthetic

compounds. Recounting the whole argument in Appah (2013b) is beyond the scope of this

paper. However, for completeness, I include some salient parts of the argument.

This is a very common type of compound in which a verb and its internal argument

occur, usually with the linear order inverted. The compounds are usually of varied semantic

characterisation, mostly nouns that refer to the action, manner of carrying out the action, the

fact of the action occurring, the result of the action, etc. This is exemplified by èdzìbàǹdzí ‘act

of eating (lit. food eat)’. Further examples are in Table 10.  

In these compounds, the internal argument of the verb assumes a generic non-referential

function and the meaning of the whole compound can somehow be worked out from the

meaning of the verb and that of its argument. However, it will not be accurate to claim that

they are endocentric compounds because they fail the hyponymy test. Whereas the verb is the

selecting element in the construction, it cannot be said to be the formal head of the compound

32  
 
because it does not have the category of the whole. In like manner, the left-hand noun

member cannot be said to be the semantic head because it does not have the full complement

of semantic properties of the whole compound).viii

Table 10. Action nominal exocentric synthetic compounds in Akan

Compound Constituents Gloss Translation Pattern
m̀ -bɔ̀frà̀ kúḿ m̀ -bɔ̀fŕá + kúḿ PL-child + to kill infanticide N-V
àdzè-sɛ́!é àdzé + sɛ́é thing + to destroy wastefulness N-V
èdzìbàǹdzí èdzìbáń + dzí food + to eat (act of) eating N-V
àdzè-tɔ́ àdzé + tɔ́ thing + to buy (act of) buying N-V
àdzè-tɔ́ń àdzé + tɔ́ń thing + to sell (act of) selling N-V
èdwúmá-yɛ́ èdwúmá + yɛ́ work + to do (act of) working N-V
èdzìbàǹnòá èdzìbáń + nòà food + to cook (act of) cooking N-V
bàkà-nú bàká + nú lagoon + to stir fishing in a lagoon N-V
àdzè-sùá àdzé + sùà thing  +  to  learn education, learning N-V
àsè-kyèrɛ́ àsè + kyèrɛ́ meaning + to show interpretation/explanation N-V

Previous accounts of these Akan compounds follow some accounts of English compounds

like shoemaker in assuming that the right-hand constituent in the Akan compound is

nominalized prior to becoming a part of the compound and that the nominal syntactic

category and the other properties percolate from the nominalized right-hand constituent.

Appah (2013b) argued that, given the fact that in most cases there is no formal making of the

putative prior nominalization of the right-hand constituents, we need not be forced into

claiming that such compounds are endocentric and that, following Bauer (2008, 2010), we

may still analyse them as synthetic compounds, without committing to defending the prior

nominalization of the right-hand constituent.

The definition of synthetic compound in Appah (2013b), following Grimshaw (1990), is

not necessarily a compound with a deverbal head constituent. Rather, a synthetic compound,

in contradistinction to a root compound, is one in which one of the constituents takes an

argument which must be satisfied in the compound. As she puts it, “[t]he essential difference

between the root and synthetic compounds, then, is in the argument-taking properties of their

33  
 
heads. The characteristic differences between the two kinds of compounds follows from this

difference” (Grimshaw 1990: 70).

Given this definition and the implication that the potential head of the N-V compound

has the wrong syntactic category, Appah (2013b) argued that we have to assume that the

syntactic category does not come from either constituent. Rather, it is a constructional

property. Appah (2015) has subsequently defended this position, arguing that the syntactic

category of all Akan compounds are inherited from a constructional meta-schema for

compounding which is pre-specified to bear a syntactic category – N. Thus, the fact that the

compound in question does not have a nominal head is not a problem because the meta-

schema supplies the nominal syntactic category.

The action nominal exocentric synthetic compound also has a characteristic tonal

melody, where the tones on all the syllables in the first constituent of the compound are low.

All the tones in the second constituent are also low except the last, if the second constituent

has more than one syllable. Monosyllabic second constituents bear high tones. This tonal

melody occurs irrespective of the tonal melodies of the individual constituents in Isolation.

Thus, the tonal melody may also be properly construed as a constructional property of the

action nominal exocentric synthetic compound; the construction imposes this tonal melody.

6.3. Transpositional exocentric compounds in Akan

As noted in section 5.3, the transpositional exocentric compound is the type in which only the

word class of the compound is not overt, meaning that the compound functions as a member

of a word class that is different from either constituent. Examples of Akan transpositional

exocentric compounds are those in which two verbs compounded yield a noun (Table 11).

34  
 
Table 11. Transpositional exocentric compound in Akan
Compound Form Gloss Translation Pattern
tútá tú+tá uproot + fart a weed V-V
tséká tsé+ká hear+say hearsay V-V
prɛ̀hwɛ́ prɛ̀+hwɛ́ tap+see gossiping V-V
gyédzí gyé+dzí receive + eat faith/belief V-V
kámá ká+má say+give intercession V-V
kàsákyérɛ́ kàsà+kyèrɛ̀ talk+teach/show counselling V-V
dzímá dzí+má eat+give intercession V-V
sɔ́hwɛ́ sɔ́+hwɛ́ try+see tribulation/temptation V-V
sòmákɔ́ sòmà+kɔ́ send+go faithfulness in going on errands V-V
pùtsí-kyérɛ́ pùtsì+kyèrɛ̀ tap+show act of gossiping by touching the V-V
addressee and pointing at the target

The two verbs in each such transpositional exocentric compound may be found occurring

together as verbs in a serial verb construction but the compound that is formed may never be

used as a verb because it is a noun.

These compounds provide the strongest evidence for the view that the syntactic

category of the Akan compound is a holistic constructional property that does not depend on

the syntactic category of the constituents. As noted in section 6.2.2, this position is defended

in Appah (2015) with the argument that given the fact that the Akan compound is invariably

nominal, we are led to assume that there is a meta-schema for compounding in Akan which is

pre-specified to bear a nominal syntactic category which all Akan compounds inherit.ix

In terms of their semantics, we see that although the constituents mostly designate

actions, the compounds tend to be abstract. For example, the compounding of gye ‘to receive’

and dzi ‘to eat’ yields a compound that means ‘faith/belief’. These compounds, therefore,

have to be interpreted metaphorically. For instance, as shown in (18), the semantics of gyedzi,

is rendered as a concept that is metaphorically related to the combined meaning of the

constituents because it has to be understood that, for one to believe, one has to metaphorically

take some message and eat it (cf. Balmer and Grant 1929).

(18) < [[V]i [V]j]Nk ↔ [CONCEPT metaphorically related to [SEM]i and [SEM]j]k >

[[gyé]i [dí]j]Nk ‘faith/belief’

35  
 
Finally, as noted in section, 4, for this type of compound, we need to distinguish between

semantic exocentricity and formal exocentricity because, in reality we can tell the meaning of

the compounds. It is only the syntactic category which we cannot predict unless we posit a

zero nominalizing morpheme, as done in previous analysis (cf. Anderson 2013; Obeng 2009).

6.4. Summary of the typology of exocentric compounds in Akan

From sections 6.1 to 6.3, I have presented the types of exocentric compounds in Akan based

on Bauer’s typology. I have shown that three of the types are attested in Akan. I have also

shown that two of these types can have subtypes. The typology of Akan exocentric

compounds may be summed up on a taxonomic tree like (19).

(19) Typology of Akan exocentric compound

bahuvrihi exocentric synthetic transpositional exocentric

possessive non-possessive action (nominal) participant

agentive patient locative

It is clear from the discussion that each one of these types has a characteristic feature that

clearly identifies the internal coherence of the membership of the class.

7. Conclusion

In this paper, I have discussed exocentric compounds in Akan. I reviewed the literature on

exocentricity, showing that the notion is not easy to characterize. I argued that we have to

distinguish between formal and semantic exocentricity. I reviewed Bauer’s (2010, 2008)

criteria for exocentricity and the types identified. I then discussed 3 main types of exocentric

36  
 
compounds with subtypes in Akan. They are the bahuvrihi compounds, exocentric synthetic

compounds and the transpositional exocentric compounds. I posited two subtypes of

bahuvrihi compounds – the possessive type and the non-possessive type. I also posited two

broad types of exocentric synthetic compounds – the non-participant exocentric synthetic

compound and the participant exocentric synthetic compound. The latter also has three

subtypes – the agentive, patient and locative types of participant exocentric synthetic

compounds. The transpositional exocentric compounds do not have any subtypes.

Notes
                                                                                                                         
i
Thus, Scalise and Guevara somehow overstate their potential achievement in the study of exocentricity when
they state that they aim “to survey the distribution of exocentric compounding in the world’s languages” (2006:
185, emphasis added).
ii
This position is also argued for in cognitive approaches to compounding which indicate that so-called
exocentric compounds can be interpreted either metonymically and metaphorically (cf. 2005b, 2005a, 2006b,
2006a, 2010). However, it seems rather strong because we find compounds that are not straightforwardly
interpretable by means of metaphor or metonymy. I can immediately think of the Chinese compounds in (1) in
which there is quite a semantic distance between the compounds and their constituents. Thus, the view that
exocentric compounds can be interpreted metonymically and/or metaphorically can be true for only some classes
of exocentric compounds and mostly for those found in the familiar languages of Europe. As Scalise and
Guevara (2006: 199) observe, “[i]n the European Languages, exocentricity is generally restricted to metonymic
conceptual relations between the constituents and the whole (and even more, a relation of inalienable possession,
as in the so-called possessive compounds), very rarely reaching the level of semantic <<freedom>> observed in
Chinese.”
(1) dōngxi ‘lit. east + west = thing’
jiānghú ‘lit. river + lake = vagabond’ (Scalise and Guevara 2006: 199)
iii
A reviewer points out that there is support for the view that an unexpressed head carries the properties of the
denotatum, so that as Ralli (2013: 100, 113) argues based on data from Modern Greek compounds, the so-called
exocentric compounds are not entirely headless because they contain derivational suffixes which are added to the
combination of the two lexemes. I would argue, however, that whereas for a language like Modern Greek and
even Italian (Bisetto 1999 and Zuffi 1981, reviewer’s comment), there may be formal markers that provide
evidence for the claim that an unexpressed head carries the features of the denotatum, for Akan, there is no such
evidence and proponents of similar accounts have had to accept this fact. Anderson (2013) proposed such an
account for Akan verb-internal compounds, but he says in conclusion that exocentric compounds don’t conform
to the expected pattern.
iv  I should point out that, in principle, every Akan noun has a prefix. Thus, we may find these nouns sometimes
bearing vowel prefixes.
v
The star is meant to account for the fact that some compounds have more than one contiguous low tone before
the final high or more than one high tone at the end of the compound. In the Asante dialect, a noun has an
additional final mid vowel, which is said on high tone, if the stem terminates in a high vowel. Thus, whilst the
Fante and Akuapem dialects have nsu ‘water’, Asante has nsu-o ‘water’. The parenthesized high tone accounts
for the optional final vowel.
vi
Note that this is not the regular name of the insect. I included it because that is how people of my age referred
to the insect, when we were little and it has the same structure and semantics as the compounds discussed here.
vii  This is the name of a place at the edge of Wesley Methodist Church in Sekondi - Ghana, where, when I was
growing, the (strong) guys in the neighborhood met at night to talk.
viii
As noted in section 3, ordinarily, various formal tests, would be applied to determine which constituent is the
head of the compound, including determining which constituents carries any inflectional marking (cf. Dressler

37  
 
                                                                                                                                                                                                                                                                                                                                                                                            
2006). In Akan the only relevant inflectional marking for nouns is number. However, being abstract nouns, the
compounds that fall into this class do not inflect for number. Thus we cannot determine the head of the
compound by looking at the pattern of number marking.
ix
A reviewer has suggested that the postulation of a meta-schema invalidates the distinction between exocentric
and endocentric compounds and that it could have serious repercussions on how to treat compounds in general in
a way that goes beyond the simple descriptive issues. I agree that the idea of a meta-schema may be quite a
profound claim with potential implications for the analysis of compounding that I cannot appreciate presently.
However, I believe that the postulation of the meta-schema does not invalidate the endocentric-exocentric
distinction in compounding. The meta-schema is significant in accounting for the nominal syntactic category of
the Akan compound and it has been defended in Appah (2013b, 2015). The important point worth making here
is that the fact that we recognise the paring of form and meaning in specific constructions does not prevent us
from recognising higher-level generalisations. The meta-schema embodies one such higher-level generalisation
in the analysis of compounding in Akan. So to go to the reviewers point about the neutralization of the
distinction between endocentric and exocentric compounds, we can say that one possible interpretation of the
existence of the meta-schema is that all Akan compounds are formally exocentric, given that the form class is
inherited from the meta-schema. Rehearsing the argument in the present paper will increase the size of the paper.

References  

Abakah, Emmanuel Nicholas. 2004. The segmental and tone melodies of Akan. Trondheim:
Norwegian University of Science and Technology Dissertation.
Abakah, Emmanuel Nicholas. 2005. Tone rules in Akan. Journal of West African languages, XXXII
(1-2): 109–134.
Abakah, Emmanuel Nicholas. 2006. The Tonology of Compounds in Akan. Languages and
Linguistics, 17: 1-33.
Adams, Valerie. 2001. Complex Words in English. Harlow: Longman.
Anderson, Jonathan C. 2013. Verb-internal compound formation in Akan. Journal of West African
languages, XL (1): 89-104.
Andreou, Marios and Angela Ralli. 2015. Form and Meaning of Bahuvrihi Compounds: Evidence
from Modern Greek and Its Dialects. In L. Bauer, L. Körtvélyessy and P. Štekauer (eds.),
Semantics of Complex Words. 163-185. Dordrecht: Springer.
Anyidoho, Akosua. 1990. On Tone in Akan Compound Nouns. Paper presented at the the 19th West
African Languages Congress, University of Ghana, Legon.
Appah, Clement Kwamina Insaidoo. 2013a. The case against A-N compounding in Akan. Journal of
West African languages, XL (1): 73-87.
Appah, Clement Kwamina Insaidoo. 2013b. Construction Morphology: Issues in Akan Complex
Nominal Morphology. Lancaster: Lancaster University dissertation.
Appah, Clement Kwamina Insaidoo. 2015. On the syntactic category of the Akan compound: a
product-oriented perspective. Acta Linguistica Hungarica, 62 (4): xxx-xxx.
Arcodia, Giorgio Francesco. 2012. Constructions and headedness in derivation and compounding.
Morphology, 22 (3): 365-397.
Balmer, W. T. and F. C. F. Grant. 1929. A grammar of the Fante-Akan language. London: Atlantis
Press.
Bauer, Laurie. 1990. Be-heading the word. Journal of Linguistics, 26 (1): 1-31.
Bauer, Laurie. 2008. Exocentric compounds. Morphology, 18: 51-74.
Bauer, Laurie. 2010. The typology of exocentric compounds. In S. Scalise and I. Vogel (eds.), Cross-
disciplinary issues in compounding. 167-175. Amsterdam/Philadelphia: John Benjamins
Publishing Company.
Benczes, Réka. 2005a. Creative noun-noun compounds. Annual Review of Cognitive Linguistics, 3 (1):
250-268.
Benczes, Réka. 2005b. Metaphor- and metonymy-based compounds in English: a cognitive linguistic
approach. Acta Linguistica Hungarica, 52 (2): 173-198.

38  
 
                                                                                                                                                                                                                                                                                                                                                                                            
Benczes, Réka. 2006a. Analysing Metonymical Noun-Noun Compounds: the Case of Freedom fries.
In R. Benczes and S. Csábi (eds.), The Metaphors of Sixty: Papers Presented on the Occasion
of the 60th Birthday of Zoltán Kövecses. 46-54. Budapest: Eötvös Loránd University.
Benczes, Réka. 2006b. Creative Compounding in English: the semantics of metaphorical and
metonymical noun-noun combinations. Amsterdam: John Benjamins Publishing Company.
Benczes, Réka. 2010. Setting limits on creativity in the production and use of metaphorical and
metonymical compounds. In A. Onysko and S. Michel (eds.), Cognitive perspective on word
formation. 219-242. Berlin/New York: Walter de Gruyter.
Bisetto, Antonietta and Sergio Scalise. 2005. The classification of compounds. Lingue e linguaggio,
(2): 319-310.
Bisetto, Antonietta and Sergio Scalise. 2007. Selection is a head property. Acta Linguistica
Hungarica, 54 (4): 361-380.
Blevins, James Peter. 2006. Word-based morphology. Journal of Linguistics, 42 (3): 531-573.
Bloomfield, Leonard. 1933. Language. New York: Holt, Rinehart and Winston.
Booij, Geert E. 2002. The Morphology of Dutch. Oxford: Oxford University Press.
Booij, Geert E. 2010a. Construction morphology. Language and Linguistics Compass, 4 (7): 543-555.
Booij, Geert E. 2010b. Construction morphology. Oxford: Oxford University Press.
Booij, Geert E. 2012. The Grammar of words: an introduction to linguistic morphology. Oxford:
Oxford University Press.
Ceccagno, Antonella and Bianca Basciano. 2009. Sino-Tibetan: Mandarin Chinese. In The Oxford
handbook of compounding (eds) R. Lieber and P. Štekauer. Oxford: Oxford University Press.
Ceccagno, Antonella and Sergio Scalise. 2006. Classification, structure and headedness of Chinese
compounds. Lingue e linguaggio, 2: 233-260.
Christaller, Johannes Gottlieb. 1875. A Grammar of the Asante and Fante Language called Tshi
(Chwee, Twi) based on the Akuapem dialect with reference to the other (Akan and Fante)
dialects. Basel: Basel Evangelical Missionary Society.
Christaller, Johannes Gottlieb. 1933. Dictionary of the Asante and Fante Language called Tshi
(Chwee, Twi): with a grammatical introduction and appendices on the geography of the Gold
Coast and other subjects. Basel: Basel Evangelical Missionary Society.
Croft, William. 2001. Radical construction grammar: Syntactic theory in typological perspective.
Oxford: Oxford University Press.
Di Sciullo, Anna-Maria and Edwin S. Williams. 1987. On the definition of word. Cambridge, MA:
MIT Press.
Dixon, R. M. W. 2004. Adjective classes in typological perspective. In R.M.W. Dixon and A.Y.
Aikhenvald (eds.), Adjective classes: A cross-linguistic typology. 1–49. Oxford: Oxford
University Press.
Dolphyne, Florence Abena. 1988. The Akan (Twi-Fante) language: Its sound systems and tonal
structure. Accra: Ghana Universities Press.
Dressler, Wolfgang U. 2006. Compound types. In G. Libben and G. Jarema (eds.), The representation
and processing of compound words. 23-44. Oxford: Oxford University Press.
Fabb, Nigel. 1998. Compounding. In A. Spencer and A.M. Zwicky (eds.), The handbook of
morphology. 66-83. Oxford: Basil Blackwell.
Goldberg, Adele E. 1995. Constructions: a construction grammar approach to argument structure.
Chicago: University of Chicago Press.
Goldberg, Adele E. 2006. Constructions at work: the nature of generalization in language. Oxford:
Oxford University Press.
Grimshaw, Jane. 1990. Argument structure. Cambridge, MA: MIT press.
Haspelmath, Martin. 2002. Understanding morphology. London: Arnold.
Haspelmath, Martin. 1993. A grammar of Lezgian. Berlin and New York: Mouton de Gruyter.
Katamba, Francis X. and John T. Stonham. 2006. Morphology. Basingstoke: Palgrave Macmillan.
Lieber, Rochelle. 1980. The organization of the lexicon. Cambridge, MA: Massachusetts Institute of
Technology dissertation.
Lieber, Rochelle. 1983. Argument linking and compounds in English. Linguistic Inquiry, 14 (2): 251-
285.

39  
 
                                                                                                                                                                                                                                                                                                                                                                                            
Lieber, Rochelle. 1989. On percolation. Yearbook of Morphology, 2: 95-138.
Lieber, Rochelle. 1992. Deconstructing morphology: Word formation in syntactic theory. Chicago,
Illinois: University of Chicago Press.
Marchand, Hans. 1969. The Categories and Types of Present-Day English Word-Formation: A
Synchronic-Diachronic Approach. Munich: C. H. Beck Verlagsbuchhandlung.
Marfo, Charles Ofosu. 2004. On tone and segmental processes in Akan phrasal words: A prosodic
account. Linguistik online, 18: 93–110.
Michaelis, Laura A. and Knud Lambrecht. 1996. Toward a Construction-Based Theory of Language
Function: The Case of Nominal Extraposition. Language, 72 (2): 215-247.
Obeng, Samuel Gyasi. 2009. Akan deverbal nouns. In S.G. Obeng (ed.), Topics in Descriptive and
African Linguistics: Essays in Honor of Distinguished Professor Paul Newman. 96-110.
Muenchen: LINCOM Europa.
Osam, E. Kweku. 1999. Adjectives in Akan. Afrika und Übersee, 82 (2): 189-211.
Packard. 2000. The morphology of Chinese. Cambridge: Cambridge University Press.
Payne, Thomas E. 1997. Describing morphosyntax: A guide for field linguists. Cambridge: Cambridge
University Press.
Pepper, Steve. 2010. Nominal compounding in Nizaa - a cognitive perspective. London: SOAS,
University of London Dissertation.
Plag, I. 2003. Word-formation in English. Cambridge: Cambridge University Press.
Ralli, Angela. 2013. Compounding in Modern Greek. Dordrecht: Springer.
Ralli, Angela and Marios Andreou. 2012. Revisiting exocentricity in compounding: Evidence from
Greek and Cypriot. In F. Kiefer, M. Ladányi and P. Siptar (eds.), Current issues in
morphological theory:(Ir) regularity, analogy and frequency. Selected papers from the 14th
international morphology meeting, Budapest, 13–16 May 2010. 65-81.
Amsterdam/Philedelphia: John Benjamins.
Scalise, Sergio. 1984. Generative morphology. Dordrecht: Foris Publications.
Scalise, Sergio. 1992. Compounding in Italian. Rivista di linguistica, 4 (1): 175-199.
Scalise, Sergio and Antonietta Bisetto. 2009. The classification of compounds. In R. Lieber and P.
Štekauer (eds.), Oxford Handbook of Compounding. 34–53. London: Oxford University Press.
Scalise, Sergio, Antonietta Bisetto and Emiliano Raúl Guevara. 2005. Selection in Compounding and
Derivation. In W.U. Dressler, F. Rainer, D. Kastovsky and O. Pfeiffer (eds.), Morphology and
Its Demarcations. 133-150. Amsterdam/Philadelphia: John Benjamins Publishing Company.
Scalise, Sergio and Antonio Fábregas. 2010. The head in compounding. In S. Scalise and I. Vogel
(eds.), Cross-disciplinary issues in compounding. 109-125. Amsterdam/Philadelphia: John
Benjamins Publishing Company.
Scalise, Sergio and Emiliano Raúl Guevara. 2006. Exocentric compounding in a typological
framework. Lingue e linguaggio, 2: 185-206.
Schönefeld, Doris. 2005. Zero-derivation - functional chnage - metonymy. In L. Bauer and S. Valera
(eds.), Approaches to Conversion/Zero-Derivation. 131 - 157. Münster: Waxmann.
Selkirk, Elisabeth O. 1982. The syntax of words. Cambridge, MA: MIT Press.
Štekauer, Pavol. 2000. Beheading the word? Please, stop the execution. Folia Linguistica, 34 (3-4):
333-356.
Wälchli, Bernhard. 2005. Co-Compounds and Natural Coordination. Oxford: Oxford University
Press.
Williams, Edwin S. 1981. On the notions" Lexically related" and" Head of a word". Linguistic Inquiry,
12 (2): 245-274.
Zager, David. 1981. On orientation or, constraining false analogy. Linguistics, 19 (11-12): 1107-1131.
Zwicky, Arnold M. 1985. Heads. Journal of Linguistics, 21 (1): 1-29.

40