Nominal derivation from Noun Phrases in Akan

1

NOMINAL DERIVATION FROM NOUN PHRASES IN AKAN

Clement Kwamina Insaidoo Appah Department of Linguistics, UG

Abstract In Akan, nouns can be formed from both lexical and non-lexical categories. This paper deals with the formation of nouns from noun phrases. It looks at the derivational processes involved as well as the parts of the construction that remain as part of the derived noun. The paper argues that the derived noun, as noted by Mensah (2003), is an economic way of conveying the full meaning of the structure from which the noun was derived. To this extent, only those elements of the phrase that are central to the expression of the meaning of the construction will remain as part of the derived noun. 0. Introduction In this paper, I attempt to show how the morphological unification of the elements of the phrase proceeds. I show that though, in principle, different NPs can be nominalized, not all of them undergo the same derivational processes. I also show which elements of the phrase end up as segments in the derived nominal and what determines their selection. The paper is divided into four main sections. Section one discusses the processes involved in the derivation of nouns. Section two looks at the derivation of nouns from various NP types. Section three looks at some of the phonological processes that characterise the derivations discussed in section one and section four is the conclusion of the paper1. 1. The Derivational Processes2 In this paper, we show that there are five main processes involved in the derivation of nouns in Akan. They are subject dropping, object fronting, TAMP dropping, compounding and affixation 3 . We describe these processes in the following subsections. It is worth mentioning here that none of these processes is obligatory for any of the nominalizable groups. Even when a process appears to apply to all constructions or structures of a particular type (for instance NPs, as the case is in this paper), the details of the application will vary from one nominalizable structure to another. For this reason, we will indicate which process(es) are relevant to every NP type and the details that are peculiar to each one will be noted accordingly. 1 The following are the abbreviations used in this paper. 1 = First Person, 2 = Second Person, 3 = Third Person, A = Adjective, Aff = Affix, DEF = Definite determiner, FUT = Future, HAB = Habitual, N = Noun, NEG = Negative, NP = Noun Phrase, OBJ = Object, PAST = Past tense marker, Quan = Quantifier, REL = Relativizer. 2 These processes are not limited to the derivation of nouns from noun phrase. Rather, they apply to all instances of nominal derivation that involve lexical and/or non-lexical categories. 3 Whilst the last three processes are purely morphological, the first two are argument structure processes that have morphological implications in that they affect the structure of the noun that will be derived from the construction.

2

1.1. Subject Dropping This is the process by which subjects; both lexical and pronominal, are dropped in the process of deriving nouns from non-lexical categories. (1) shows that both the lexical and pronominal subjects are dropped in the derivation of the noun edwumayE ‘work’. 1a. Kwesi bE-yE edwuma no - edwuma-yE ‘work’ K. FUT-do work DEF ‘Kwesi will do the work’ b. Ama re-n-yE edwuma no - edwuma-yE ‘work’ A. PROG-NEG-do work DEF ‘Kwesi is doing the work’ c. ç-bE-yE edwuma no - edwuma-yE ‘work’ 3SGSUBJ-FUT-do work DEF ‘He will do the work’ d. ç-yE-E edwuma no - edwuma-yE ‘work’ 3SGSUBJ-do-COMPL work DEF ‘She did the work’ The above data suggest that, in some cases, it is only the verbal predicate and its internal arguments that enter the derivational process since no element in the subject phrases remain as segments in the derived noun. This raises the question as to why the need for the subject dropping process as it is being described here. To this, we will answer thus; we regard some nouns as having been derived from fully specified lexical entries. Otherwise, we will not be able to account for why some elements such as the quantifier modifying the head in the subject phrase of an embedded clause remains as part of the noun derived, though the head of the phrase does not. This is seen in (17a-c). The process of subject dropping may be obscured by the presence of what looks like the pronoun on the derived noun especially when the subject in the clause from which the noun is derived is the third person singular subject pronoun [ç-] ‘s/he’. The example below will make the point clearer. 2a. Papa no ko de foro boç - ç-koforoboç man DEF fight.HAB take climb stone ‘the man climbs hills whilst fighting’ b. ç-ko de foro boç - ç-koforoboç 3SGSUBJ-fight.HAB takes climb stone ‘S/he climbs hills whilst fighting’ ‘the mount-climbing warrior’ c. wo-ko de foro boç - ç-koforoboç 2SGSUBJ-fight.HAB takes climb stone ‘You climb hills whilst fighting’ ‘the mount-climbing warrior’

3

In the example (b) above, it looks like it is the subject pronoun [ç-] that has remained as part of the derived noun, but that is not the case. The [ç-] is just a nominal prefix. As the example (a) and (c) above show, a nominal prefix like that one will be attached to any noun that is of the same class as the one above even if the syntactic structure from which it was derived did not contain a subject pronoun like that one – the third person singular subject pronoun [ç-]. Furthermore, the example in (c) shows that even if the pronominal subject marker is any person apart from the third, the derived noun will still have [ç-] as the nominal prefix. This is also seen in (1) where notwithstanding the lexical subjects – Kwesi and Ama, the derived noun still has [ç-] as a prefix. So, the [ç-] in the subject position in the sentence in (1 and 2) should not be confused with the [ç-] of the derived noun. This is because, whereas the former is a pronominal subject marker, the latter is a simple nominal affix though they may possibly have the same etymon. In fact, one should not expect a clause with a third person singular pronominal subject – [ç-] to derive a noun with [ç-] as its prefix. For instance, the noun ç-som ‘service’ is derived from the simple verb stem som ‘to serve’ through the affixation of [ç-]. The conclusion we arrive at is that, in deriving nouns from non-lexical categories like phrases and clauses, all subjects4 are dropped. This will usually precede all other processes involved in deriving the nominal. This conclusion makes sense in that usually, in deriving the noun, it is the verbal predicate and its object(s) that are relevant5. Looked at from another angle, one could say that subject dropping is the process that results in the reduction of a construction that enters the derivation from an SVO to a VO. 1.2 TAMP Dropping TAMP is the acronym for Tense/Aspect, Mood and Polarity6. In the derivation of nouns from constructions that have tense aspect, mood and polarity (TAMP) makers, all the markers are dropped. This will essentially apply to verb phrases and clauses since in Akan these markers are realised as affixes on the verb (Dolphyne 1988; Saah 2002. This can be seen from the data in (1) above and (3) below where none of the tense/aspect markers remains to become a segment in the respective derived noun. TAMP dropping may be obscured (otherwise, considered irrelevant) when the verb is marked for habitual since the habitual has no segmental representation in Akan.

4 Including subjects of embedded clauses like relative clauses that serve as modifiers to the heads of noun phrases. 5 The exception to this is found with verbs of body sensation and other construction types like predicate adjectival constructions. In these, the verbs usually combine with the external NP in forming nouns (Appah 2003). 6 Our initial work has shown that dropping markers of negation is possible with the derivation of some abstract nouns such as those in example (1b), but with others like amfamfir ‘the unforgiving one’ which is derived from the clause below, the negation markers remain as segments in the derived noun. ç-m-fa asEm m-fir 3SG-NEG-take matter NEG-credit ‘S/he does not forgive’

4

1.3 Object Fronting The third process in the derivation of nouns from non-lexical categories is object fronting. In a very simple clause, where there are no intervening elements, this process could mean the inversion of the verb and the object. In a much complex clause, however, it will involve a real fronting of the object of the verb notwithstanding the number of intervening elements. Object fronting in its simplest form, (that which involves a simple inversion of the verb and object) features prominently in nominal derivation in Akan, even in deriving nouns from predicate adjectival constructions. The following are two examples. In (2a), the verb yE ‘do’ and the object adwuma ‘work’ are inverted. The same is true of the dzi ‘eat’ and its object edziban ‘food’ in (2b). In both examples, object fronting translates into a simple case of verb-object inversion because there are no elements intervening between the verbs and their respective objects. However, as noted above, there are situations where irrespective of the number of elements coming between the verb and its object, the object will be fronted. Also worth noting, is the fact that adverbials and grammatical categories like the definite article no ‘the’ are not selected in the derivational process. For example, ntEmtEm ‘fast’ in (3a) does not become a segment in the derived noun. 3a. Papa no yE edwuma ntEmtEm - adwuma-yE man DEF do.HAB work fast ‘(manner) of working’ ‘The man works very fast’ b. o-be-dzi edziban no - edziban-dzi’ 3sg-FUT-eat food DEF ‘(the act of) eating’ ‘s/he will eat the food’ The series of trees in (3a-c) below shows how the derivation of the noun edzibandzi ‘eating’ will be like. (3a) is the tree diagram for the sentence from which the noun is derived. (3b) shows that the part of the construction that enters the derivation is the VP. (3c), which is a word structure tree, shows the fronted object in its new position within the derived noun. (3c) also shows that tense/aspect markers do not become part of the derived noun as the TAMP dropping process states. 4a. S b. VP NP VP V NP V NP be-dzi edziban o- be-dzi edziban c. N NStem VStem edziban dzi → edzibandzi ‘eating’

5

Object fronting will make the resulting noun look like the English compound ‘housekeeping’. These compounds, notes Payne (1997:93), consist of a noun plus a verb where the noun represents the object rather than the subject of the verb. This is what he terms object incorporation. He notes that “just about all transitive verbs that describe customary activities can incorporate a direct object…” (Payne 1997:221). 1.4 Compounding This is a concatenative process that results in the morphological unification of the elements of the clause or phrase that are being brought together to form a noun. It involves the fusing of all or some7 of the elements of the construction that is entering the derivational process into one morphological entity. In the example in (2a), which is repeated here as (5a), the parts of the construction that remain as segments in the derived noun are the two verbs ko ‘fight’, foro ‘climb’ and the object of the verb foro, boç ‘stone’. These are fused to form the nominal stem as shown in 4b. 5 a. Papa no ko de foro boç - ç-koforoboç man DEF fight.HAB take climb stone ‘the man climbs hills whilst fighting’ b. ko + foro + boç → ko-foro-boç fight climb stone ‘fight-climb-stone’. Compounding follows the three processes – subject dropping, TAMP dropping and object fronting, and may be the end of the derivation if the derived noun does not take a nominal affix. Otherwise, it is followed by affixation. 1.5. Affixation This process involves the addition of a nominal(izing) affix (prefix or suffix or both) to a stem to derive a noun. The stem could be a simple lexical stem like adjective, verb or another noun. If the derivation involves a non-lexical category, the stem to which affixes are attached could be the structure that is got after compounding. For example the output of the compounding process in (5b) -ko-foro-boç has to undergo affixation to have the appropriate affix [ç-] attached to the stem to derive the noun ç-koforoboç. This usually follows subject dropping, TAMP dropping and object fronting if only these other processes are relevant to the derivation in question. The choice of an affix is determined by the semantics of the entity to be named; whether it is human or not, and/or whether it wields power (social, financial, spiritual, etc) or not. The choice of the affix is crucial in the classification8 of the derived noun. The phonetic shape of a vowel nominal affix is determined by the vowels of the stem to which it is attached. This is the effect of vowel harmony.

7 We say some of the elements of the clause because, as it will become clear in the succeeding discussion and from earlier examples, not all the elements in a particular construction will end up as segments in the derived noun. It is just the salient elements that end up in the final product. These usually include the verb and that which is ‘verbed’. The factors that determine which elements of the construction are salient will be touched on later. 8 See Osam (1994), for a discussion of the defunct noun class system in Akan and Appah (2003) for discussion of how the classification of nouns influences the choice of an affix.

6

2.0 Deriving the Noun This section deals with how the processes described in the last section work. Here, we show which of the process(es) is/are involved in the derivation of any particular noun and what the output of the process(es) is/are. Bresnan (2001) notes that, just as the functional structure of a sentence is related to its phrase structure in a systematic way, so is the functional structure of a word related to its morphemic structure. She, therefore, proposes a representation of the internal structure of words with a “word structure tree” analogous to a phrase structure tree derived by using word structure rules similar to phrase structure rules. Following her lead, therefore, we attempt to capture the internal structure of the derived noun by means of a word structure tree formed from sub-lexical rules. As the word structure trees will show, the affixes get attached at different levels of the derivation for cases where there are both prefixes and suffixes. The ordering in such cases is, for the most part, arbitrary. However, since suffixes tend to be inflectional in nature, and it is a cross-linguistic fact that, in cases where both derivational and inflectional affixes occur, the inflectional affixes tend to be at the periphery of the word, it seems most plausible to me to let prefixation precede suffixation. One other reason for the choice of this progression is that in most cases, prefixation results in a noun before the suffix is added. The same cannot be said for suffixation. 2.1. Deriving nouns from noun phrases (NPs) The noun phrase in Akan, like many other languages, comes in varying degrees of complexities. They range from the simple (single-word (common/proper noun, pronoun)) NP to the NP with varying weights of modification, Appah (2003). We will concern ourselves with three types of NPs. They are the NP with a pre-head nominal modifier, the NP with an adjectival head modifier and the NP with an embedded relative clause serving as a modifier to the head. 2.1.1 Deriving nouns from NPs with a pre-head nominal modifier In Akan, the head of an NP can be modified by another noun in pre-head position. These may be possessive constructions9. This NP type can have a noun derived from the head noun and the modifying nominal. The elements of the phrases are not inverted during the derivation; their linear order is maintained. The processes involved in the derivation are compounding and affixation. That is, the elements of the phrase are concatenated and the appropriate affix is selected and attached to the product of the compounding process. The sub-lexical rules in (5a) below and the word structure tree following it, in (5b), show the derivation and the internal structure of the derived noun respectively. 6a N → (NAff) NStem b. N

NStem → NStem1 NStem2 (NAff) NStem NStem1 NStem2 … … …

9 Possessive constructions involving parts of the body behave differently from these constructions.

7

The following show how the derivation of the noun ahemfie ‘palace’ from the phrase çhene fie ‘chief’s house’ will be like. (7a) is the noun phrase and (7b) is the phrase marker for it. When the elements of the phrase are compounded it yields hen-fie ‘chief-house’ as shown in (7c). This new compound stem then undergoes affixation to have the appropriate affix selected and attached to it. After affixation we get [a-] attached, giving us a-hemfie ‘palace’ as shown in (7b). 7a. ç-hen fie b. NP SG-chief house ‘A palace’ N N’ N çhene fie c. N d. N NStem NStem NPre NStem hen fie Nstem1 NStem2 a- hem -fie We are of the view that it is only the stem of the word çhene ‘chief’ (that is hene) that enters the derivation and that the nominal affix gets attached when the derived stem undergoes affixation. This, as discussed in Appah (2003), is indicative of the shift in the class affiliation of the noun; a shift from class 1 to class 6 in Osam’s (1993b) and (1994) classification. The following are more examples of this type of NP and the nouns that are derived from them. As it can be seen from the example in (8c), there is a reversal in the linear order of the elements of this phrase. This is not very common with the derivation of nouns from this type of NPs, though. (8) Phrase Derived Noun a. çman penyin çmampenyin ‘nation elder’ ‘Head of state’ b. çman çba çmamba nation child ‘citizen’ c. dua mpa mpadua10 wood bed ‘wooden bed’

10 Though this word is glossed ‘wooden bed’, it is not limited in its reference to beds made of wood. In Fante, it is used as a general term for a bed notwithstanding the material it is made of – wood or metal. However, whenever it is used, what comes to mind first is a bed made of wood so to indicate that it is a metal bed and not a wooden one, the word dadze ‘metal’ is used, as in dadze mpadua ‘a metal bed’ (literally, a metal wooden bed).

8

d. nyansa asEm nyansasEm wisdom matter ‘a wise saying’ 2.1.2. Deriving nouns from an NP with adjectival head-modifier In Akan, when an adjective is used attributively, it follows the noun head. This order of occurrence may or may not be altered during the derivation of a noun from the phrase. Furthermore, most of the adjectives involved in this type of NPs come in reduplicated forms such as, papa ‘good’, denden ‘hard’, etc, but the reduplication is reversed during the derivation11. The elements in the phrase are then compounded to derive the noun. Some of the stems derived after compounding will also undergo affixation. We give an example below to show how the derivation proceeds. The sub-lexical rules and the internal structure of the nouns they derive are also shown. The sub-lexical rules and the word structure trees in (9a-b) subsume all the nouns derived from this type of NP. 9a. N → NStem1 (NAff) b. N

NStem1 → (NAff) NStem2

N2 → NStem3 AStem NStem1 (NAff) (NAff) NStem2 NStem3 AStem …. … … …. Below is how the derivation of the noun okumpa ‘good husband’ from the noun phrase kun papa ‘good husband’ will be like. (9a and b) show the phrase and the phrase markers respectively. It has already been mentioned that the reduplication in the adjectives is reversed for this type of derivation. Therefore, when the reduplication on the adjective is reversed, we get kun pa. These two elements of the phrase are then compounded and we get kunpa. Following the compounding, the derived structure undergoes affixation to yield o-kunpa ‘good husband’ as the word structure tree in (10) shows. The derivation of the noun ç-yerepa ‘good wife’ from the phrase yere papa ‘good wife’ follows the same pattern as the derivation okunpa ‘good husband’. The phrase and the noun derived from it are given in (11). (10) a. kun papa b. NP husband good’ ‘Good husband’ N AP

kun papa

11 It looks like the reversal of reduplication in these adjectives is to ensure that only mono-morphemic stems (in other words, the stems as they were before the reduplication), enter the derivational process as a way of simplifying the derived noun. This is rather typical of the Fante dialect.

9

c N NAff NStem NStem AStem o- kun pa 11. Noun Phrase Derived noun yere papa → ç-yerepa wife good ‘good wife’ Apart from the examples above in which the reduplication in the modifying adjective is undone, yet the linear order of the elements of the phrase is maintained, there are other examples in which the linear order of the elements in the phrase is reversed. In these examples, the reduplication in the adjective stem is retained. (12) contains an example. It involves the derivation of the noun a-fEfEdze from the NP adze fEfEw ‘beautiful things’. The derivation involves the inversion of the elements of the phrase, compounding and the affixation of [a-]. This is shown in (12 a, b, and c) respectively. 12 a. a-dze fEfEw b. NP SG-thing beautiful ‘a beautiful thing’ N AP adze fEfEw c. N Npref NStem Astem NStem1 a- fEfE dze → afEfEdze ‘beautiful things’ There is a third type that shares the features characteristic of the first two. These involve monomorphemic adjectival stems. With these, there is no reversal in the linear order of the elements of the phrase. The example below involves the derivation of the noun duasin ‘stump’ from NP dua sin ‘half tree’. When these two stems (dua ‘tree’ and sin ‘half/fraction’), are compounded it yields duasin ‘stump’ (Literally, half tree). This example takes no affix and the linear order of the two stems is as it would in a phrase in which the adjective stands in an attributive position relative to the noun. This is shown in (14 a-b). 13 a. dua sin b. N ‘tree’ (N) ‘half’ (A) NStem1 AStem Dua sin

10

The questions that need to be answered with regards to the reversal of the linear order of the elements of the construction are; what conditions the reversal? Is it something about the noun or the adjective? What is common to the nouns and/or to the adjectives that are found in these constructions? Furthermore, about the undoing of the reduplication in the adjectives we want to see what conditions it aside from the need for ease in pronouncing the derived noun. To enable us to answer the above questions, we present more examples of the three types of nominalizable NPs and the nouns derived from them below. 14. Noun gloss Adjective gloss Compound gloss a i. asEm ‘matter’ papa ‘good’ asEmpa ‘Gospel’ ii. adwen ‘mind’ papa ‘good’ adwenpa ‘good intention’ iii. nkçmç ‘conversation’ dE EdE(w) ‘sweet’ nkçmçdE(w) ‘sweet talks’ b i. adze ‘thing’ dçkçdçkç ‘sweet’ adçkçdçkçdze ‘sweets’ ii. adze ‘thing’ gyingyan ‘vain’ agyingyandze ‘vanity’ iii. çbra ‘life’ ahuhuw ‘flirtatious’ ahuhuwbra ‘flirtatious life’ c i. akokç ‘chicken’ ber ‘female’ akokçber ‘hen’ ii. aboa ‘animal’ fun ‘corpse’ aboafun ‘carcass’ iii oguan ‘sheep’ nyin ‘male’ oguannyin ‘ram’ d i. Nipa ‘person’ gyingyan ‘worthless’ nipagyingyan ‘vain person’ ii. Nipa ‘person’ gyingyan ‘worthless’ *gyingyannipa ‘vain person’ A look at the adjectives that are found in these constructions and so enter this derivation shows that they are mainly adjectives that express value such as papa ‘good’, physical property like dew ‘sweet’ and human propensity like gyingyan ‘vain/worthless’, Osam (1999)12. These adjectives combine with nouns of different semantic categories. However, it is clear from the data above that in combination with nouns referring to non-human impalpable entities, the adjective and the noun will be inverted if a noun is to be derived from them. For example in combination with adze ‘thing’ the adjective gyingyan ‘vain/worthless’ will be inverted when they enter the derivational process. In combination with the noun nipa ‘person’, on the other hand, inverting them results in an ill-formed structure as the examples in (d) show. Regarding the undoing of the reduplication in the reduplicated stems, there appears not to be any regular pattern in the selection of the nouns that these reduplicated stems must combine with in order for the reduplication to be reversed. The only regular thing about them is that they remain in a post-head modifier position; otherwise, they remain reduplicated in the derived noun.

12 The distinction between these classes of adjectives is not very clear to me. What counts as human propensity relative to a human entity can only be regarded as expressing value in a non-human entity. For instance, the adjective gyingyan ‘worthless’ used to modify a human entity will be regarded as expressing human propensity, but relative to another noun like asEm ‘matter’ it will refer to the value of what is being said. It remains a fact however, that underlying both situations is a tacit value judgement.

11

Therefore, the conclusion we arrive at is that it is more of the semantics of the noun rather than that of the adjective that determines whether the elements in the construction will have their linear order altered during nominal derivation13. That is, to the extent that it is agreed that these adjectives, notwithstanding their semantic classes, combine freely with nouns, we may conclude that it is a semantic property of the noun that causes the reversal in the linear order of the elements of the phrase during nominal derivation. In addition, the undoing of the reduplication in reduplicated adjectival stems is dependent on their position relative to the noun they modify. If they get inverted, the reduplication is retained. Otherwise, it is undone. 2.1.3. Deriving nouns from NPs with embedded relation clauses This subsection deals with the derivation of nouns from NPs with embedded relative clauses. In the following discussion, it will become clear that the derivation of nouns from this type of NP is not straightforward. This is not just because the constructions involved are more complex than all the other NP types we have seen so far, but also because of the extent of the arbitrariness that characterizes the selection of some of the parts of the construction that become part of the derived noun. Discussing the nominalization of idiom-chunks in Akan, Mensah (2003) notes that to convey the fullest meaning expressed by an idiom, the obligatory parts of the idiom must combine to form the nominal. We share this view and add that the fact is true of all nominalizations involving non-lexical categories and that what he refers to as the obligatory parts of the idiom corresponds to what we refer to as the elements of the construction that are central to the expression of the meaning of the construction. Here, it makes sense to add that whether or not an element will be considered central to the expression of the meaning of the construction depends on the semantics of the head of the construction14. We look at three examples of this subtype of NP from which nouns are derived. 15. a AsEm a a-ba kç matter REL PERF-come go ‘A matter which has come and gone’ b. NP N CP C S VP V V AsEm a- a-ba kç

13 This makes sense to the extent that we expect the head of the phrase to generally, but not exclusively, control the occurrence of any process in the phrase. 14 This is not the whole truth, though, since, as it will become clear from the following discussion, sometimes, what will traditionally be regarded as the head of a group may not get selected but its dependents may.

12

From the structure above we get abakçsEm ‘history’. The following is how the derivation happens. First, the two verbs in the clause are compounded. Second, the resulting compound is compounded with the noun head of the NP, but the compound verb and the noun are first inverted. This is shown diagrammatically below. V – V compounding yields the structure in (c). The result of the V – V compounding is then compounded with the head noun asEm ‘matter’ after they have been inverted, reversing the linear order and giving the structure in (d). The resulting structure then undergoes affixation picking up the prefix [a-] as the structure in (e) shows. c. VStem VStem VStem ba kç → ba-kç d. NStem VStem NStem VStem VStem ba kç sEm → ba-kç-sEm e N NAff NStem VStem NStem V V a- ba kç sEm → abakçsEm ‘history’ The other structure which is like this but shows some difference in the way the derivation of the noun takes place is the one in (17a) below. From this structure we get the noun çdomankoma ‘the gracious one’. 16. a Adom a çno nko ma grace REL 3SGSUBJ only give.HAB ‘grace which he alone gives’

13

b NP N' N CP C' C S NP VP Pro Quan V adom a çno nko ma The derivation starts with the embedded clause. In accord with the subject dropping phenomenon, the emphatic pronominal subject in the embedded clause is dropped15. The rest of the relative clause together with the head of the NP is compounded yielding the structure in (c) – the noun adomankoma, with the noun adom ‘grace still having its prefix [a-]. The correct form of the derived noun should have [ç-] as its prefix so the stem has to undergo affixation picking up the [ç-] prefix. This yields the structure - ç-adomankoma as shown in (d). This is still not right. The reason is the presence of the old prefix [a-] of adom ‘grace’. It has to be deleted. c. NStem N REL Quan V adom a nko ma → adom-a-nko-ma Grace which alone give d. N NAff NStem N REL Quan V ç- adom a nko ma → ç-adom-a-nko-ma We can give two possible reasons for the deletion of the prefix of the noun adom. One, the two vowels, in combination, violates the permitted vowel sequences within the same word (Dolphyne 1988). Two, in such derivation, as in the ones in (6-7 and 13)

15Interestingly, the quantifier in the NP is not dropped. This could be put down to the fact that it is central to the expression of the meaning of the construction. That is, the uniqueness of the referent of the derived noun is better expressed with the quantifier remaining as a segment in the derived noun.

14

above, it is the nominal stem alone that enters the derivation so that there is no clash of affixes (if the noun happens to be at the boundary of the derived stem as the case is in this instance). This should not sound implausible since in the derivation of the noun abakçsEm ‘history’, the noun asEm ‘matter’ enters the derivational process without its affix. The prefix which marks class identity in the noun must be dropped because the derived noun belongs to a different class from that of the base noun, and the shift in class must be signalled by the new affix it assumes. The other reason (one above) could work for abakçsEm too since the [ç-] vowel still comes before the [a-] of asEm. However, that may not be sufficient explanation for the absence of the vowel since there are compounds (such as nkwasea-sEm ‘nonesense’), in which the word asEm comes after a word that has got a final [-a] vowel and yet the [a-] prefix of asEm still gets deleted, even though [aa] is a permitted vowel sequence in Akan (Dolphyne 1988). This leads us to the conclusion that it is the nominal stem alone that enters the derivational process so that, as the second reason above indicates, there will be no clash of affixes. Therefore, after deleting the [a-] prefix of adom ‘grace’ and prefixing [ç-] we get çdomankoma ‘the gracious one’. 2.2 Observation So far, in this section of the work, we have seen three types of Noun Phrases (NPs) from which nouns are derived. They are the NP with a pre-head nominal modifier, the NP with an attributive adjective head-modifier and the NP with an embedded relative clause. We have also seen various subtypes of these NPs. Deriving nouns from the first type is rather straightforward; apart from one example (8c), which involved the reversal of the order in which the elements of the phrase appeared, all the others involved a concatenation of the elements (Compounding) and/or affixation. The second type of NP has three subtypes from which nouns are derived. What is notable among the features is that, in combination with nouns referring to non-human impalpable entities, the adjective and the noun in the construction will have their order reversed. Moreover, reduplicated adjectives will have the reduplication reversed depending on their position relative to the nouns they modify. If they get inverted, the reduplication is retained. Otherwise, they are undone. The third type of NP from which nouns are derived has a considerable level of arbitrariness about the selection of the elements of the construction that remain as segments in the derived noun. For example in (17), the head of the subject phrase (the emphatic pronoun çno) does not get selected but the modifier nko ‘only/alone’ does. This situation could definitely be put down to the fact that these elements of the construction, rather than what will otherwise be considered important in the expression of the meaning of the clause (including what might be considered the traditional head of the phrase), help to convey the ideas better. For instance, the word nko ‘only/alone’ helps bring out the uniqueness of the referent of the derived noun in (17). That should account for why, though the noun it modifies did not get to become a segment in the derived noun, the quantifier did.

15

3.0 Some Basic Phonological Issues It is virtually impossible for one to discuss the morphology of Akan without coming to a point where something must be said about ‘the phonology of the morphology’ discussed. A number of phonological processes characterize the derivational processes and the derivations discussed so far. In this section we describe and exemplify some of the most basic of these processes. These include segment or syllable loss, vowel harmony, homorganic nasal assimilation, reduplication and tonal changes. These phonological processes occur as a way of making the utterance more effortless and smoother (Bota 2002:81). 3.1 Homorganic nasal assimilation This is the process by which a speaker adjusts the place of articulation of a nasal in anticipation of the place of articulation of a following consonant. That is, the nasal adopts the place of articulation of the consonant that follows it, (See Katamba 1989). Bota (2003) makes a distinction between two types of nasals in this regard. They are The N-Morpheme and the Already-Associated Nasal. The former is the morpheme whose place of articulation is not specified and so receives its place of articulation from the following consonant and the latter is the nasal which has a specified place of articulation but changes to agree with the following consonant with respect to the place of articulation. In this paper, we have seen two examples of the latter – the derivation of ahimfie ‘palace’ and okumpa ‘good husband’ in (7) and (10) respectively. In (7), the alveolar nasal [n] in çhene takes on the labiodental place of articulation of the initial sound of the word fie ‘residence’ – the labiodental fricative [f] so that it is realized as [› M]. This is exemplified in (17a). 17a. çhene fie → a-himfie → / a›hi ›M ›fi¤e¤ / king house ‘palace’ In (10) the alveolar nasal [n] of the noun kunu ‘husband’ assimilates to the bilabial place of articulation of the initial segment of the following stem – papa ‘good’. The nasal is therefore realized as [m]. This is exemplified in (17b) below. b. kunu papa → o-kumpa → /o›ku›m›pa¤ / husband good ‘good husband’ Even though these are the only two examples seen in this work this process can seen in any case involving the compounding of two or more stems where a nasal terminates the first and the next begins with a consonant. 3.2 Vowel harmony This is the co-occurrence restriction on the distribution of vowels in words of more than one syllable. They are principles that require that vowels found in a word of more than one syllable agree on the presence of a pertinent phonatic feature. In Akan the most common harmony is that based on tongue height (ATR)16. That is, whether

16 There are other types of vowel harmony in Akan like the rounding harmony of the Fante dialect which requires that in addition to agreeing on ATR, vowels in a word should agree on roundedness. This will, for instance make the future aspect marker (which is realised as [bE-] in Akuapem and

16

the vowel in question is produced with an Advanced Tongue Root or with a retracted (unadvanced) tongue root. Based on this ATR feature, Akan vowels are grouped into two and a word (stem + affixes) are normally expected to have vowels from one group. 18. i/I, e/E, Q/a, o/ç, u/U The vowels on the left of the slash in each pair are the set of vowels produced with advanced tongue root (+ATR vowels) and those on the right are the ones produced with unadvanced tongue root (-ATR vowels). Normally (but not necessarily), the vowels in a stem as well as their vowel affixes have vowels from the same group. In case the vowels do not agree in terms of their ATR status, there may be assimilation and usually, it is the +ATR vowels that assimilate the –ATR ones. In (7), the first stem has -ATR vowels and the second has +ATR vowels. When they are compounded the -ATR vowels are raised to +ATR. This is exemplified in (19). 19. çhene /ç ›hI¤nI¤/ + fie / ›fi¤e¤ / → / a›hi › M ›fi¤e¤ / 3.3 Loss of Segments or Syllables In putting together the elements of a phrase some stems may lose their final segments or syllables. In (7), (10) and (11) the first stems terminate in high vowels for the Asante dialect. These high vowels are deleted before the stems are compounded. This must happen before the homorganic nasal assimilation discussed above and exemplified in (17) can take place. This is shown in (20). 20a. ç-hene fie → a-him-fie → / a›hi › M ›fi¤e¤ / SG-king house ‘palace’ b. o-kunu papa → o-kum-pa → /o›ku›m›pa¤ / husband good ‘good husband’ We have argued, in this paper, that when words enter the derivational process they do so without their affix. That could also be an instantiation this phonological process. 3.4 Changes in the basic tones of stems When stems are compounded in the derivation of nouns, there is usually some amount of change in the basic tonal pattern of the stems. The following are examples. 21a. ç›-he¤ne¤ fi¤e¤ → a›hi›m›fi¤e¤ / a ›hi›M ›fi¤e¤ / king house ‘palace’ b. o›-ku¤nu¤ pa¤pa¤ → o›-ku ›m›-pa¤ → /o›ku›m›pa¤ / husband good ‘good husband’

Asante), be realised as [bç-] in Fante if the verb root to which it is affixed has the rounded half-open back vowel [ç], as in kç ‘go’. So that the sentence – S/he will go – which is rendered in Akuapem and Asante as ‘ç-bE-kç’ will be rendered in Fante as ç-bç-kç.

17

From the data above, we see that the first elements in the compounds are said on low tones notwithstanding their tonal patterns before entering the derivational process. For example hene ‘chief’ and kunu ‘husband’ have only high tones on their syllables but in the derived noun they have only low tones. According to Dolphyne (1988), this is because the first elements qualify the second. However, the generalization does not seem to be borne out by the data used in this work since in (20b) it is the second element that qualifies the first. 3.5 Reduplicated stems We indicated 2.1.2 that in deriving nouns from NPs with adjectival head modifiers those modifying adjectives that enter the derivational process reduplicated have the reduplication reversed. The following are some more examples. 22a. agya papa → agyapa → / æ›d¸a›pa¤ / father good ‘good father’ b. adwuma denden adwumaden → / a›d¸u›ma ›dI¤n¤ / work hard ‘hard work’ c. çyere papa çyerepa → / ç›j I›rI›pa¤ / wife good ‘good wife’ In this section of the paper, we have seen some of the basic phonological issues that characterise the derivation of nouns from NPs. 4.0 Conclusion This work has been about the derivation of nouns from noun phrases. We first looked at the derivational processes involved in section (1) which we indicated are not specific to the derivation of nouns from noun phrases. In section (2) we saw how the derivational processes discussed in (1) worked with various NP types. In section (3) we discussed and exemplified some of the basic phonological issues that characterise the derivational processes and the derivations that we saw. The challenge for now and the question for feather research is; how can we get well-defined criteria for the selection of the various parts of the construction that will remain as segments in the derived noun. For now, though, it appears, the only way to deal with the phenomenon is to tackle each construction on its own merit minding their idiosynctratic features and with time and feather research find out what features they share. So that generalizations can be made which can later be reduced to rules that can be fed into a computer to find out if given a phrase (NP) the computer can derive or parse the word. References Agyekum, Kofi. (2000) “Lexical expansion and elaboration in Akan: AfisEm and the

media.” A paper presented at the WALC. 15th-19th August 2000, University of Ghana, Legon.

18

Akanlig-Pare, G. (1999), Nominal derivation in Buli in Cahiers Voltaïques, Gur Papers 4: 99-109. Aning, S. K. (1998) Nominal Constructions in Akan. M.Phil Thesis, Dept. of Linguistics, TNU, Trondheim Appah, C. K. I. (1999), Ulinganishi wa uundaji wa maneno katika Kiswahili na Kiakan: uhusiano na tofauti. Unpublished project work, Dept. of Swahili, University of Dar es Salaam, Tanzania. Appah, C. K. I. (2002), Nominal derivation in Akan. Ms, Dept of Linguistics, NTNU, Trondheim. Appah, C. K. I. (2003), Nominal derivation in Akan; A descriptive analysis. MPhil Thesis, Institute of Language and Communication studies, NTNU, Trondheim. Bota, G. (2002), Some Aspects of Bono Phonology. M.Phil Thesis, Dept. of Linguistics, UG, Ghana. Bresnan, J. (2001), Lexical Functional Syntax. Oxford: Blackwell. Chomsky, N. (1970) Remarks on Nominalization. In Jacobs, R. and P. Rosenbaum (eds) Readings in English Transformational Grammar. Waltham, Mass 184- 221. Dolphyne, F. A. (1988). The Akan (Twi-Fante) Language: Its Sound System and Tonal Structure. Accra Ghana Universities Press. Haspelmath, Martin. (2002) Understanding Morphology. London: Arnold. Katamba, Francis. (1989). Introduction to phonology. Longman: London Katamba, Francis. (1993). Morphology. London: Macmillan Press Ltd. Mensah, Eugene. (2003), Akan Verbal Body-Part Idioms. MPhil Thesis, Institute of Language and Communication studies, NTNU, Trondheim. Nti, S. K. (1999) Akan Adjectives – A Morpho-Syntactic Study. M.Phil Thesis, Dept. of Linguistics, NTNU, Trondheim. Osam, E. K. (1993b) The loss of the noun class system in Akan. Acta Linguistica Hafniensia

Vol. 26, 81-106. Osam, E. K. (1994). Aspects of Akan Grammar – A Functional Perspective. Doctorial

dissertation. University of Oregon, Eugene. Osam, E. K. (In preparation), 1.3 Noun Formation. Payne, Thomas E. (1997) Describing Morphosyntax. Cambridge: CUP.

19

Saah, Kofi K. (2002). The nature and function of Akan verbal Affixes. In M. E.

Kropp Dakubu and E. K. Osam (Ed); Studies in the languages of the Volta Basin 1. Proceedings of the Annual Colloquiun of the Legon-Trondheim Linguistics Project. 180–196.

Spencer, Andrew. (1991), Morphological Theory. Oxford: Blackwell.