PALATALIZATION IN HAUSA MICHAEL A.O. OYEBOLA DEPARTMENT OF LINGUISTICS AND NIGERIAN LANGUAGES UNIVERSITY OF ILORIN. PALATALIZATION IN HAUSA ABSTRACT The paper applies the theory of lexical phonology to the analysis of the operation of the palatalization rule in Hausa. It presents a brief overview of the theoretical framework of lexical phonology. In the framework the lexicon is construed as being made up of different strata on which phonological rules are paired with morphological processes. The paper endeavours to account for the apparently deviant behaviour of Hausa alveolar and palatal consonants with respect to the palatalization process. This relates to the occurrence of alveolar consonants before front vowels and of palatal consonants followed by back vowels. Evidence for the cyclicitly of the palatalization rule is provided. By considering the formation of plurals and compounds in Hausa, the paper discusses the organization of the lexicon into different levels and provides justification for the setting up of the levels. INTRODUCTION We will discuss, in this analysis, certain aspects of word formation in Hausa. Specifically we will attempt to apply the theory of lexical phonology to the analysis of the operation of the palatalization rule in Hausa. The palatalization rule is a process intertwined with the process of the historical development of the language. We are interested in exploring the extent to which the theoretical framework provided by lexical phonology allows us to capture the facts exhibited by the distribution of palatal consonants in Hausa. We shall first give a brief overview of the theoretical framework of lexical phonology. We shall discuss the status of the palatalization rule in Hausa before proceeding to illustrate the application of the theory to the organization of the lexicon. We shall then attempt 2 to account for the problematic cases, i.e. those cases where palatalization seems to have applied contrary to what would have been expected, as well as where it fails to apply in spite of the fact that its structural description is met. 1. LEXICAL PHONOLOGY: - The theory of lexical phonology developed by Mohanan (1982) as well as Kiparsky (1982) provides a framework for accounting for the interaction between phonology and morphology within the lexicon. It construes the lexicon as being made up of different strata on which phonological rules are paired with morphological processes. Rather than having all the phonology apply before the morphology, it predicts that rule application can alternate between the phonology and the morphology at different strata of the lexicon. In other words, the theory allows us to apply some phonology before the morphology and then go back to phonology in a derivational process. The framework furthermore, allows us to account for the fact that some phonological rules may apply to certain morphemes without applying to others. If such rules are restricted to particular strata, the fact that they do not apply to some other morphemes will be accounted for. Lexical phonology distinguishes between lexical and post-lexical rules. Lexical rules are those which apply in the process of word formation within the lexicon while post-lexical rules apply after the completion of the word formation process. In Kiparsky’s framework all lexical strata are cyclic while for Mohanan 3 certain lexical strata are cyclic and some are non-cyclic. Cyclic rules obey the Strict Cycle Condition which blocks a rule from applying in an underived environment. A cyclic rule can only apply in an environment created, for example, by the application of a rule or by the juxtaposition of morphemes. Kiparsky argues that the strict cycle condition can be derived from the Elsewhere Condition. The Elsewhere Condition stipulates that we apply the more specific of two rules if the outputs of both rules are distinct. Kiparsky analyzes a lexical entry as constituting a lexical identity rule whose structural description is the same as its structural change. The Elsewhere Condition is then applied to determine the ordering relation between the lexical rule and any other rule with a corresponding structural description. By deriving the Strict Cycle Condition from the Elsewhere Condition the theory predicts that the conjunctive application of rules is blocked only when the outputs of the application of the rules are distinct. This insight leads to the differentiation between structure building and structure changing rules. The outputs of structure changing rules are distinct while those of structure building rules are non-distinct. Thus while the Elsewhere Condition blocks structure changing rules from applying in underived environments, structure building rules are allowed to apply. The application of structure building rules in underived environments is made possible by introducing the notion of underspecification. Underspecification involves leaving out predictable features of lexical entries and 4 writing lexical redundancy rules which fill in these features. This approach ensures that rules such as assimilation rules which have to apply in underived environments are not blocked. We summarize this overview of the theoretical framework by highlighting the distinction between lexical and post-lexical rules. Lexical rules may refer to word internal structure while post-lexical rules cannot refer to word internal structure. Lexical rules may not apply across words while post-lexical rules may apply across words. Lexical rules may be cyclic and if cyclic, they are subject to the strict cycle condition. Post-lexical rules, on the other hand, cannot be cyclic, hence they apply anywhere. Lexical rules are structure preserving while post lexical rules need not be structure preserving. Lexical rules may have exceptions while post-lexical rules cannot have exceptions. Finally lexical rules must precede all post-lexical rule applications. (This summary is due to Pulleyblank (1983) cited by Archangeli (1984:5)). Later works on lexical phonology include Durand, J. (1990) Katamba (1989) and Mohanan (1986). In a more recent work Giegerich (2005) argues that stratification should be conceptualized as base-driven rather than being affix-based. 2. PALATALIZATION IN HAUSA: Hausa alveolar consonants t, s, d, z and velars are palatalized before front vowels. While alveolar consonants change completely into their corresponding palatal consonants, palatalization occurs as a 5 secondary articulation on velars. Both ‘d’ and ‘z’ change to the voiced palatal affricate before front vowels. Gregersen (1967:172) argues that Hausa palatal consonants are historically derived from earlier alveolars before front vowels. However, while the palatal consonants synchronically constitute part of Hausa phonemic inventory, the palatalized velars occur allophonically before front vowels. The fact that palatal consonants and their corresponding alveolars occur constructively in several environments forces us to recognize palatal consonants as phonemes in Hausa. The contrastive distribution of palatals and alveolars is illustrated by the following set of examples: tasa ‘awaken’ casa ‘thrash’ tasha ‘station’. Gregersen suggests that the phonemic palatal consonants are, almost exclusively, restricted in their distribution to loan words. He proposes marking these words with a feature [+ foreign]. Although we agree with this observation we shall attempt to show that the apparently deviant behaviour of the alveolar and palatal consonants can be accounted for within the theoretical framework of lexical phonology. Examples of the deviant distribution given by Gregersen are: (a) the occurrence of alveolar consonants before front vowels: teebur ‘table’ asibiti ‘hospital’ dozin ‘dozen’ reediyoo ‘radio’. (b) palatal consonants not before front vowels: cooci ‘church’ jooji ‘judge’ washaa ‘washer’. 6 Since there are other words with deviant distribution of alveolar and palatal consonants whose historical status is as yet unclear, it would be better if the deviant behaviour could be made to follow from theoretical considerations. Having considered a few facts connected with the distribution of palatal consonants, we shall now discuss the status of the palatalization rule in the lexicon. To begin with, the occurrence of lexical exceptions referred to above suggests that the palatalization rule cannot be post-lexical, since post-lexical rules are exceptionless. Furthermore the rule applies lexically since there is no case where it occurs across words. It applies in all cases as a word internal or morphologically conditioned rule. The strongest evidence for its cyclicity comes from its interaction with other rules such as vowel lengthening and gemination. Vowel lengthening and gemination in Hausa are lexical rules since they make reference to word internal structure. For example Greenberg (1941, p. 321) draws attention to the morphophonemic alternation of long vowel in open syllable and short vowel in closed syllable in accordance with the normal syllabic structure of the language. In its interaction with the vowel lengthening rule, the palatalization rule can apply in any order. So for instance in the formation of the intensive form of the verb ‘to read’ we see the interaction of the palatalization rule with the vowel lengthening rule. The intensive form of the verb is formed through a suffix – e which is lengthened before an object pronoun. 7 1. 2. Palatalization Vowel lengthening /karant + e/ karance karancee 2 1 /karant + e/ karantee karancee The interaction of the palatalization rule with the gemination rule similarly provides evidence for the cyclicity of the rules. There are cases where the palatalization rule has to occur before gemination while there are other cases where gemination must occur before palatalization. These phenomena are reflected in the formation of the participle. The participle in Hausa is formed from the verbal base through a process of reduplication, gemination and suffixation. The masculine, feminine and plural participles have –e, -iya, - u suffixes respectively. For instance, the participles are formed from the verb dafa ‘to cook’ in the following way: dafa reduplication dafaf final c germination dafaff suffixation dafaffe, dafaffiya, dafaffu. In the following derivations palatalization and gemination apply in different orders. The derivation of ‘fansasshiya’ redeemed (fem) proceeds in the following way. Final c reduplication Final c germination Suffixation Palatalization fansa fansas fansass fansass + iya fansasshiya 8 (to redeem) If palatalization had occurred before gemination we would derive the wrong form fansashshiya. In the following derivation palatalization has to occur before gemination. The derivation of fahimtacce ‘intelligent’ (masc) proceeds in the following way: Verb stem Final c reduplication Suffixation Palatalization Germination fahimta fahimtat fahimtat + e fahimtace fahimtacce (to understand) If gemination had occurred before palatalization, we would need another rule of progressive assimilation to change *fahimtatce to fahimtacce, thus complicating the derivational process. The derivational processes, discussed above, show that palatalization, gemination and vowel lengthening rules are cyclic and since they are cyclic they cannot be post-lexical. 3. OPERATION OF THE PALATALIZATION RULE IN THE LEXICON: In this section we shall look at the organization of the lexicon within the theoretical framework of lexical phonology. In this regard, it will be shown that while the environment for the application of the palatalization rule is the same in all cases, it has to have different levels at which it can apply. The organization of the lexicon into different levels will be illustrated by considering the formation of plurals and compounds in Hausa. The lexicon is organized into three levels. The 9 first level consists of words which form their plural in an irregular manner. The affixes attached to these words represent + boundary affixes. Level two corresponds to that at which derivational morphology and compounding take place. Level three deals with regular word boundary inflection, and in particular the more regular plural formation. The following are examples of words which form their plural in an irregular manner. 1. 2. 3. 4. 5. 6. màcè sírdì yárò dámì Kàzá míjì ‘women’ ‘saddle’ ‘boy’ ‘bundle of corn’ ‘chicken’ ‘husband’ mátá plr. síràdá yárá dámmá kàjí mázá Examples 1, 5 and 6 illustrate the alternation between palatals and alveolars in the singular and plural. Example 2 is an exception. Since we are positing a palatalization rule, examples 1, 5 and 6 have the following underlying forms. /mat + e/ /kaz + a/ /maz + i/ ‘woman’ ‘chicken’ ‘husband’ plr. plr. plr. /matta/ /kaz + i/ /maz + a/ Palatalization applies in the appropriate environment i.e. before front vowels. In example 6 we have to posit an additional vowel harmony rule. There is evidence from Hausa plural formation for putting the class of words given above on a different level with respect to other plural formations. There appears to be a constraint in the formation of Hausa plurals which blocks the 10 occurrence of ‘a’as the penultimate vowel if the final vowel is ‘a’. That is how we can account for the change in the penultimate vowel in the following examples: /haraf + a/ /al’amar + a/ /hakarkar + a/ harufa al’amura hakurkura ‘sound’ plr. ‘affairs’ ‘ribs’ Similarly there is no other way of accounting for the epenthetic ‘u’ in farauta ‘finger nails’. This latter example is derived from underlying /fart + aCe/ (singular ‘farce’). The final vowel ‘e’ of the suffix changes to ‘a’ because of the general alternation between alveolar and palatals in singular and plural. The usual final consonant copying process would yield ‘farata’, thus the need for the epenthetic ‘u’, giving farauta. Compare fage ‘cleared space’ fagage. The rule referred to here can be formalized as follows ø u/ [a] plr……………… 3 As a matter of fact a general constraint which holds for most cases of plural formation can be formalized in the following way *[ v1 cv1] plr. If this rule is made to operate at the level of the regular plural formation, the fact that it does not apply to the irregular forms in examples 1-6 can be accounted for. Thus forms like mátá, síràdá and yárà are not blocked. We shall consider next the evidence for the need to distinguish level two from level three, i.e. a separate level for certain types of compounds as opposed to 11 that for the regular plural formation. Level 2 is distinct from level 1in the sense that while level 1 combines non-independent entities e.g. /mat + e/ level 2 combines words, or at least joins affixes to words. At level 2 the application of the palatalization rule, where it does apply, follows from different principles, in the sense that it must have applied prior to compounding. This is shown, for example, in the compound word ‘abinci’ ‘food’. Hausa derivational morphology utilizes prefixes and suffixes in forming substantives from verbs, nouns and adjectives. Examples of the suffixes are –ci and –ta. When nouns are formed in this way an ‘n’ is inserted between the noun and the other noun or suffix. The ‘n’ is the genitival ‘n’ for masculine nouns. The most common compound formation is the combination of àbù ‘thing’ with another word. The ‘u’ in abu is changed to ‘i’ and an ‘n’ is inserted between the words. For instance, àbù ‘things’ and ‘ci’ ‘eat’ become abinci ‘food’. Similarly we have abinsha ‘drink’ abinso ‘desire’. This process is also used in the combination of three words, for instance gidan abinci restaurant (from house, thing, eat) the compound formation rule can thus be formalized as follows. ø n/ ([ [ ]Noun masc ] ) ………………. 4 compound Among the words used as prefixes in the word formation process are mai- , plural masu -, Ma – is another prefix used. The differences in the behaviour of mai 12 – and masu- on one hand and ma – on the other provide evidence for making level two distinct from level three. There are three basic differences between the first two prefixes and the third one (ma-). The first two (mai – and masu-) are what, in a boundary type framework, would be referred to as word boundary affixes. They are derived from personal pronouns. Ma- exclusively serves as a prefix, indicating place as in makaranta ‘school’ or agent maciji snake (biter). While mai- and masu – are attached to nouns to serve as the possessive as in maigida ‘head of household’ ma – is attached to verbs. Perhaps the strongest piece of evidence for separating levels two and three comes from the different ways in which the words having these prefixes form their plural. While words with prefix ma- form their plural like the regular forms, the words with mai-prefix cannot form their plural as their corresponding plain nouns do. For example the plural of gida house is gidaje. However, the plural of maigida ‘head of household’ is not *maigidaje but masugida. Similarly, the plural of mairai ‘living being’ is not *mairayuka but masurai. On the other hand we have words with prefix ma-forming their plurals just like the following. aboki wakili friend representative abokai wakílaì 13 So far example we have maciji masallaci masauki snake mosque rest house macizai masallatai masaukai The only way by which we can account for the different ways of forming the plural in the compounds and the derived nouns is to make the level at which compounding takes place distinct from that of level three where the regular plural inflection takes place. The prefixes mai-and its plural masu – will be on level two while the prefix ma – will be on level three. By restricting the prefixes mai- and masu – to level two we correctly predict that while gidaje is a possible form, *maigidaje is not. If the regular plural inflection were to take place of level two, there would be no way of blocking the formation of *maigidaje. Level three which represents the level of regular plural inflection is the level at which the most productive ways of forming plurals take place. Among the most productive ways of forming plural is the form in which the final consonant of the stem is copied in the suffix – oCi. For example, we have taga – tagogi ‘windows’. Interestingly, most words of foreign origin form their plural in this way. The following are some examples: Likita minti Ofis ‘doctor’ ‘minute’ ‘office’ likitoci mintoci ofisoshi 14 At this stage, we can now give a graphic representation of the organization of the lexicon. We recall that by restricting words which form their plural in an irregular manner to level 1, we correctly predict that they will not undergo some of the processes which other words undergo. Compounds are formed at level two. Before going into the schematic representation of the lexicon, it is necessary to draw attention to a rule which must apply port-lexically. This rule ay e ………………… (5) applies post-lexically because it can apply across words and it appears to be without exception. ‘bai’ (phonetically [be]) in the expression bai zo ba ‘he has not come’ is derived from the combination of the negative marker ba ….. ba and the third person masculine pronoun: ba ya zo ba This derivation is borne out by such cases where there is no contraction. ba ta zo ba she has not come The fact that this rule is post-lexical accounts for the fact that a word like sosai ‘completely’ (given orthographically) is realized phonetically as [so:se:] and not *[so:ʃe:] i.e. not palatalized . Since rule 5 applies post-lexically, one reason why palatalization will not occur is because the environment for the rule cannot be created within the lexicon. The other reason will be considered in the next section. 15 The derivations referred to earlier are given below: (1) kaji chicken [kaz] level 1 ( [kaz]i) affixation [kaji] palatalization level 2 level 3 (2) abinci food [abu] [Xi] level 1 [ci] palatalization level 2 [abi] [ci]……………………….u i [abi] n [ci]……………………. 4 [abinci] ……………………… Compounding level 3 The status of X given above will be dealt with in the next section. (3) maigida [gid] level 1 [gid a] Affixation level 2 [mai] [gida] Affixation [maigida] Affixation level 3 Post-lexical [megida] ………………. rule 5 (4) gidaje houses [gid] level 1 level 2 level 3 (5) level 1 level 2 level 3 (6) level 1 level 2 [gid] ade] [gidaje] macizai [Xiz] ciz [ma [ciz] ai] [macizai] alamura [alamar] - Affixation and consonant copying Palatalization Palatalization Affixation Affairs 16 level 3 ( [ alamar]a) Affixation [alamura] ………………. 3 In light of compounds like gidajenabinci ‘restaurants’ and abubuwanso ‘wishes’ we have to revise our earlier proposal according to which compounding is to be restricted to level 2. The examples given here indicate that compounding should be allowed to take place at level three after the plural formation. So far example, we have (6) level 1 level 2 level 3 4. gidajen abinci [gid] [abu] [Xi] [ci] Palatalization [abi] [ci] u i [abi] n [ci] …………….. [abinci] Compounding [gid] ade] Affixation plr. [gidaje] Palatalization [gidaje] n [abinci] …………… 4 gidajen abinci Compounding PALATALIZATION IN NON-DERIVED ENVIRONMENTS In some of the examples considered so far, it will be noticed that palatal consonants occur before front vowels in non-derived environments. An example of this is ci ‘eat’ (with the consonant marked X in the underlying representation). Palatalization applies in this case contrary to the strict cycle condition which forbids a cyclic rule from applying in non-derived environments. We had also earlier referred to the occurrence of palatal consonants before back vowels. In this section we shall account for the apparently deviant distribution of palatal 17 consonants. We will have to account for words such as kujera ‘chair’, fitila ‘lamp’ jaki ‘donkey’. Forms such as ‘kujera’ and ‘ci’ provide evidence for the need for underspecification. Following Kiparsky’s proposal, only structure changing rules are blocked from applying in non-derived environments. If consonants occurring before front vowels are underspecified for their anterior feature, palatalization rule can apply, since its application here will be structure building rather than structure changing. Structure building rules can apply in non-derived environments. Given this analysis, only these palatal consonants which occur before back vowels are underlyingly patalal, as in jaki ‘donkey’ cali ‘netbag’. Underspecifying coronal consonants occurring before front vowels allows us to capture the palatalization process in derived and non-derived environments as one process. So far instance, the coronal consonant before the front vowel in ‘kujera’ will be listed as - son + cor - cont + voice o ant The lexical redundancy rule [+ cor] [-ant] / - v [back] provides full specification for the consonant, giving us ‘kureja’. The rule applies by simple filling in the value of the anterior feature. 18 Given the application of palatalization as a structure building rule here, we can account for the occurrence of alveolars before front vowels by positing that the application of the rule is blocked because the consonants are fully specified in the lexical entry. Following Kiparsky’s convention of construing a lexical entry as a lexical identity rule, the application of palatalization in a form like ‘fitila’ will be blocked by the Elsewhere Condition. The lexical identity rule ‘fitila’ is a more specific rule than the palatalization rule and since the outputs of a conjunctive application of the rules are distinct, the more general rule i.e. palatalization rule, will fail to apply. Thus we derive the correct form ‘fitila’ rather than *ficila. Justification for this analysis is found in the fact that there are many words of foreign origin among the words in which alveolars occur before front vowels. For example: asibiti rediyo minti hospital radio minute These words are, among others, to be analyzed as being fully specified when they entered the grammar of Hausa. Interestingly, most of these words form their plural by adding the suffix oci (with the consonant copied from the stem). Palatalization applies without exception in the suffix. For example: asibiti minti asibitoci mintoci 19 The fact that palatalization applies in the suffix is consistent with the fact that pluralization in the form given above had been in the grammar before the adoption of the new lexical items. 5. SUMMARY: In this analysis we have attempted to account for the organization of the lexicon of Hausa within the theoretical framework provided by lexical phonology. We have shown that the palatalization rule in Hausa is a cyclic rule and that by organizing the lexicon into different levels, we can account for the differences in the word formation processes. Based on the notion of underspecification and the Elsewhere Condition, the apparently deviant distribution of palatal consonants has been made to follow from more theoretical considerations. In conclusion we would suggest that an analysis based on lexical phonology should be made complementary to the actual historical study of Hausa language. 20 REFERENCES Abraham, R.C. (1941): A Modern Grammar of Spoken Hausa. Hertford, England: Stephen Austin. Archangeli, D. (1984): An Overview of the theory of Lexical Phonology and Morphology. Cambridge Mass: MIT Press. Brauner, S. and M. Ashiwaju (1966): Lehrbuch der Hausa-Sprache. Munchen: Max Hueber Verlag. Centre for the Study of Nigerian Languages (1977). Sabon Kamus na Hausa Zuwa Turanci (Modern Hausa - -English Dictionary) Zaria: Oxford University Press. Durand, J. (1990): Generative and Non-linear Phonology. New York: Longman. Giegerich, H. (2005): Lexical Strata in English. Cambridge: Cambridge University Press. Greenberg, J.H. (1941): Some Problems in Hausa Phonology. Language 17.4 31623. Gregersen, E.A. (1967): The Palatal Consonants in Hausa. Journal of African Languages 6. 170-84. Katamba, F. (1989): An Introduction to Phonology. Essex: Longman. Kiparsky, P. (1982b): From Cyclic to Lexical Phonology in Van der Hulst and N. Smith. The Structure of Phonological Representations (Part 1). Dordrecht: Foris Publications. Mohanan, K.P. (1982): Lexical Phonology. MIT Dissertation IULC. Mohanan, K.P. (1986): The Theory of Lexical Phonology. Dordrecht: Reidel. Robinson, C.H.: (1913): Dictionary of the Hausa Language Vol. 1 Hausa – English Cambridge: Cambridge University Press. 21