Truncation of Some Akan Personal Names

This paper examines some morphophonological processes in Akan personal names with focus on the former process. The morphological processes of truncation of some indigenous personal names identified among the Akan (Asante) ethnic group of Ghana are discussed. The paper critically looks at some of these postlexical morpheme boundary processes in some Akan personal names realized in the truncated form when two personal names interact. In naming a child in a typical Akan, specifically in Asante‟s custom, a family name is given to the child in addition to his/her „God-given‟ name or day-name. We observe truncation and some phonological processes such as vowel harmony, compensatory lengthening, etc. at the morpheme boundaries in casual speech context. These morphophonological processes would be analyzed within the Optimality Theory framework where it would be claimed that there is templatic constraint that demands that the base surname minimally surfaces as disyllable irrespective of the syllable size of the base surname. The relatively high ranking of this minimality constraint, we claim in this paper, forces the application of the compensatory lengthening rule to ensure satisfaction of that constraint in the truncated forms.


INTRODUCTION
This paper studies the morphophonology of some indigenous Akan (Asante) personal names (henceforth APN), which generally falls under onomastics.The present study focuses on the analysis of how these personal names are realized at the phonetic level of the Akan grammar.It throws more light on what happens at the morpheme boundary when personal names (dayname and family/surname) interact.During naming ceremony of a child in a typical Asante custom, a family name is given in addition to the "God-given" or day-name so that a male child born on Friday will be called Kofi1 , which hypocoristic form is Koo and the father may decide to name him Nimo, thus making up his complete personal name Kofi Nimo.In a discourse situation; in rapid speech, these two separate personal names systematically undergo some morphophonological alternations at the phonetic level of the language.Some of these systematic morphophonological processes include truncation, compensatory lengthening, vowel harmony, and invariably, tonal changes.The current study critically examines the first of these three processes.It will be shown that at the morpheme boundary, it is usually the surname that tends to undergo those systematic alternations while the day-name preserves its invariable disyllabic structure in its hypocoristic form.
The present paper focuses on the morphological process of truncation which ensues when these two personal names interact in an utterance i.e. rapid speech.In Akan, virtually all day-names have their hypocoristic forms formed in isolation.However, surnames usually do not have such systematic truncated forms as there are in many other languages such as English (Lappe, 2002) and Spanish (Piñeros, 1998).

THE AKAN SYLLABLE STRUCTURE
The Akan language, like many other languages, cross-linguistically, operates a simple open syllable system since the language does not permit nonsonorants at coda position.The language has three basic syllable structures such as vowel (V), vowel consonant (CV) and consonant (C).The C is always realized as syllabic consonants such as syllabic nasals /m, n, ŋ/, liquid /r/ as well as the semi-vowel /w/.For example, a CVC word kan "count" is syllabified into CV.C.Each of these syllables constitutes a tone-bearing unit.Akan verb stems, in particular, mostly have the CV syllable structure (cf.Dolphyne, 2006) as in the following examples; kɔ "go", fa "take", si "alight", etc..It is only the syllabic consonants that can occur morpheme-finally in Akan aside from vowels.For a detailed discussion on this, see Dolphyne (2006), (Abakah, 2004(Abakah, , 2005)), among others.Dolphyne (2006) further postulates that each vowel in Akan constitutes a syllable.Therefore, a sequence of vowels of the same or of different qualities should not belong to the same syllable.For example, the following words are represented in the following syllable structure at the underlying level of representation in (1).names in normal or causal speech context.For example, I gave Kofi + Opoku separately and each consultant would say them in speech context to realize a form like [kʷoo#pʷokʷu].To be able to correctly predict the position of the syllables that are targeted in the ensued truncated form, I asked the consultants to repeat each combination three (3) times.I did Praat 5 analysis thereof of the truncation.
The focus of the analysis was primarily on base Akan personal day-name + family name construction and their accompanying morphological process of truncation in the truncated forms.

AN OVERVIEW OF PERSONAL NAMES IN AKAN
The study of proper names is cross-linguistic and has been done across cultures and languages such as in Yemeni by Ghaled Al Zumor (2009), in Arabic by Auda (2003), in Javanese by Widodo and Saddono (2012), in Indian by Sharma (2005), in Spanish by Piñeros (1998), in German by Wiese (2001), in English by Lappe (2002), in Nigerian by Lamidi and Aboh (2011), etc.There have been a number of studies on Akan (personal) names within the various fields of study ranging from anthropology, communication, linguistics, etc. over the past years.Such studies have been done by scholars on Akan such as Abakah (2004), Ansu-Kyeremeh (2000), Obeng (1997Obeng ( , 2001)), Agyekum (2006), Opoku (1967), Sarpong (1977), Baduike (1976) among others.Among these various studies, it is Obeng"s (1997) work that look at grammatical study of the phenomenon.While Obeng"s (idem) study focuses on wordinternal (morphophonology) formation of hypocoristics of day-names in Akan, as well as sociolinguistics of those hypocoristic forms, others, such as Agyekum (2006), Ansu-Kyeremeh (2000), Opoku (1967), Sarpong (1977), etc. have studied Akan personal names from a cultural and sociolinguistic points of view.From the brief discussion mentioned, it can be concluded that though this area of study is a very fertile grounds for linguistic study, exhaustive research in the grammar domains such as in morphology, phonology, and syntax are as yet to be done.This paper, therefore, attempts to contribute to the study of the APN from the perspective of morphology and phonology.It covers the truncation of both dayname and family (surname) name in an utterance context with focus on investigating what goes on at morpheme boundary of these two personal names at the phonetic level.
Hypocorism can simply be defined as a pet-name; a diminutive or abbreviated name (Robinson, 2006, p. 663).In Akan, almost all personal day-names have their hypocoristic variants.Obeng (1997) has listed the hypocoristic forms of all the seven (7) Akan personal day-names for both males and females as below.For the purpose of the present study, I adapt only the Twi variants and present them as follows in (2).
(2) Akan personal day-names and their hypocoristic variants (Obeng, 1997, p. 43 Boersma (2012, p. 1), is a "computer program for analyzing, synthesizing and manipulating speech and other sounds, and for creating publication-quality graphics".So for the present study, I used it to determine the quality of the vocalism of the manipulated forms as well as to verify the tones of some forms. 6As Obeng (1997)   From (2) we observe that it is usually the male personal day-names that undergo full hypocorization.The female counterparts compound with an address name "awuraa" lady.Again, we observe that [kèé] is invariably used for virtually all male personal day-names with an initial /k/ sound.It is the male hypocoristic variant [kʷòó] that usually surfaces in the complete personal name format: personal day-name + surname/family name.With the females, however, none of the hypocoristic forms surface in the same format in the Twi dialect.Again, it is worthy to note that the above hypocoristic forms are usually used in isolative address, but in combinative style i.e. day-name + surname, forms such as [kʷòó] or [kʷàá] are employed as we shall see in the subsequent data.

DATASET 1: THE APNs
It is at the intermediate level that systematic morphophonological processes such as truncation, vowel harmony, compensatory lengthening, etc. take place.The truncatory process does not occur at random, but a careful observation of the data will reveal that the processes, especially deletion, are consistent and follow some systematic patterns which we would discuss in detail shortly in this section.The commonest morphological process observed in surnames is truncation.Truncation, as Kager (1999) puts it, "is a general term for any morphological category that is derived by a systematic phonological shortening of a basic form" (Kager, 1999, p. 259).It is observed at two main places; (a) at morpheme boundary and (b) within morpheme or wordinternally.The former is more productive in the language, therefore, the present paper will focus on it in this section.Aside from this morphological process, there are some phonological rules that apply as well including vowel harmony and compensatory lengthening, which will also briefly engage our attention in this section.

MORPHEME BOUNDARY TRUNCATION IN AKAN
In Akan, sometimes there is an elision, particularly of the affixal morpheme or initial V or CV syllable of the surname at morpheme boundary in the truncated forms of day-name + surname format of personal name.This morpheme boundary deletion is motivated in the general phonology of Akan as has been observed in compounding in Akan (cf.Marfo, 2004;Abakah, 2004Abakah, , 2005;;Dolphyne, 2006;etc.).The truncation process usually targets the first syllable(s) of the surname at morpheme boundary.We will illustrate this in the following subsections.
As would be observed throughout the rest of this subsection, the main targeted base form for truncation is invariably the surname and the targeted syllable is usually the initial syllable of some disyllabic surnames; and in some cases, the first two syllables in polysyllabic surnames.The discussion begins with truncation of trisyllabic surnames. 7As we can observe from the table in (3), all the personal day-names that orthographically begin with kwa-are shortened into [kʷàá] in the hypocoristic form.Kwadwo, just like Kofi, has [kʷòó] as an additional hypocorism.The additional variant form, we suspect, might have come from that of the Fante variant of Friday-born Kojo [kʷòó]. 8The hypocoristic form of Thursday female personal day-name is [jàájàá]; a reduplicated form of the base /jaa/.
Volume 15 (1), February 2015(http://dx.doi.org/10.17576/GEMA-2015-1501-09)ISSN: 1675-8021 From (3), it is observed that while in (3a-3j) it is only the first V of the surname that deletes, in (3k-3o) it is the initial CV syllable that truncates.This pattern is generalized as follows; Generalization 1: Day-name + trisyllabic surname: Delete the first syllable of the trisyllabic surname CV/V → Ø / + #___ In the generalization 1, we observe that the initial CV/V of a trisyllabic surname deletes at the morpheme boundary when it comes into contact with a hypocoristic variant of day-name at the surface representation.

TRUNCATION OF POLYSYLLABIC SURNAMES
The number of syllables of the surname that undergoes is not as crucial as what remains after the truncation.Throughout this paper, we will observe that there seems to be a minimality requirement of two syllables imposed on all the surnames at the phonetic level after this truncatory process.This disyllabic minimality requirement serves as the template for the truncated names (both day-name and surname) in the truncation.This is consistent with the general minimum size of personal names in Akan, which is two syllables.In (4), truncation of polysyllabic base surnames in Asante is discussed.
( In (4a-4d) the first two syllables; V.CV of the surnames delete at morpheme boundary while in (4e-4j) it is only the first V of the base surnames that delete in the truncated forms.These two observations is generalized as follows; Generalization 2: (i).Day-name + polysyllabic surname: Delete the first two syllables of the surname (as seen in 4a-4d).
V → Ø / + #___ The difference between the truncation observed in (4a-e) and that observed in (4f-j) cannot be attributed to a specific phonological rule.The only systematic observation about the former examples is in their tonal pattern, i.e. the targeted initial two syllables bear Low-Low (LL), while in the latter examples we observe two different patterns: Low-High (LH as in 4f-i) and High-High (HH as in 4j) on the initial two syallbles.The shape of the base surname does not offer any explanation to the discrimination of the number of the initial syllables that truncate at the surface representation.Again, the context within which the truncation occurs is not defined, hence, the truncation occurs in a context-free environment.
In the next subsection, how disyllabic base forms of surnames are truncated is discussed.

DATASET 2: COMPENSATORY LENGTHENING
In the last set of data on truncation in the APN, the truncation of disyllabic surnames will be discussed.It has been observed that during truncation, irrespective of the syllable size of the base surname, the last two syllables survive the deletion as has been consistently seen thus far.In (5), cases where the truncation rule applies to disyllabic base surnames and how the grammar compensates for the lost syllable in the truncated form is presented.However, since it has been observed thus far that the truncated surname invariably has minimally two syllables, there is always compensatory lengthening of the last vowel of the day-name to ensure the fulfillment of that minimality requirement on the templatic shape of the surnames.The vowel used for the compensatory lengthening systematically carries tone left over by the truncated syllable as seen in the appendices A & B.
( Truncation of disyllabic surnames as we have seen in ( 5) involves two processes; deletion and subsequent compensatory lengthening.The latter process is very crucial since its failure to apply will result in ill-formedness as seen at the extreme right of the data in (5).Here tone plays very important role in determining the number of syllables that surfaces.In the data, we consistently perceive a gliding tonal configuration of LH#LH in the truncated form.This observation strengthens our earlier claim about compensatory lengthening across the morpheme boundary.We account for this observation in the following generalization.Generalization 3: Day-name + disyllabic surname: Delete the first syllable of the surname; spread all features of the final V of the day-name to compensate for the lost syllable.
Having looked at the data on contexts that lead to compensatory lengthening within personal names, attention is now turned to some exceptions to this truncation rule in some disyllabic surnames akin to the above.

NON-UNDERGOERS OF TRUNCATION
It is not in all cases that there are morpheme truncations when personal names of Akan interact in both isolative and combinative styles.In this section, some exceptions to this truncation rule in disyllabic and polysyllabic surnames is discussed.The data in (6) presents an exception to the generalization made thus far about truncation.This happens when the surname is disyllabic.As an alternative to the dual processes of first deletion, and subsequent compensatory lengthening, speakers maintain the base syllable size of surnames intact thus satisfying the templatic constraint.

DATASET 4: NO TRUNCATION OF POLYSYLLABIC SURNAMES
The exceptional cases are not idiosyncratic to disyllabic base forms, but also to polysyllabic surnames as can be seen in ( 7) below.Again, the application of the truncation rule would not have resulted in any violation of the templatic constraint.
( In (7) above, we cannot readily account for why the surnames fail to undergo truncation since they possess similar shapes as those we observed in (3), which lost their initial syllable(s) in the truncated forms.
As noted earlier, we also observe vowel harmony in the truncated names.Regarding the directionality of the vowel Advanced Tongue Root (ATR) harmony, it is assumed in this paper that it is largely leftwards i.e. from right to left.The vowels in the surnames, to a large extent, determine/control the ATR value those of the day-name and the opposite is not the usual case.In the next subsection, we briefly look at the directionality of ATR spreading in APN.

ATR HARMONY IN THE APN
There have been extensive studies on Akan vowel harmony in the literature already and among such works are Welmers (1946), Schachter and Fromkin (1968), Clements (1981), Stewart (1967Stewart ( , 1983)), O"Keefe (2003), Dolphyne (2006), Boadi (2009), Ballard (2010), Abakah (2013) among others.In this subsection, I will briefly look at the Akan [+ATR] value spreading as well as the directionality of its spreading.The leftward ATR spreading is observed from the data where [+ATR] value spreads from a vowel to the right usually to the immediately preceding [-ATR] vowel.That is, it spreads from the surname to the day-name.In a construction of CV 1 .V 2 # CV 3 .V 4 ; where V 2 is [-ATR] and V 3 ; [+ATR], it is the V 3 that spreads its value to the preceding vocalic segment(s).But when V 2 is [+ATR] and V 3 [-ATR], the harmony is blocked.Again, it is worthy to note that when V 1 and V 2 share the same quality without any intervocalic consonant, both harmonize with the following vocalic segment in terms of ATR value.These are further illustrated in ( 8 From ( 9), it appears the Noun-Noun compounding obeys the disyllabic minimality requirement imposed on Noun 2 as all the base trisyllabic Noun 2 surface as disyllabic in the compounding.

FORMALIZATION OF THE TRUNCATION
Patterns of the various morphological and phonological processes have been observed in the discussion, thus far.The processes discussed thus far include truncation, vowel harmony, and compensatory lengthening.We now formalize the generalizations made about those processes within the phonological theory; Optimality Theory (Prince & Smolensky, 1993).
With its hierarchical ranking of universal set of constraints, we are able to account for these processes with a set of ranked constraints which we can base on and make predictions for a similar process in other grammars.Again, by reranking some constraints, we will be able to adequately account for even the non-truncation cases discussed in ( 6) and ( 7) all with the same set of constraints, which the other earlier generative models would be able to adequately do.
In doing the optimality theoretic analysis of the truncation, we follow McCarthy & Prince"s (1995) Correspondence Theory of reduplication and its offshoot; Output-to-Output (OO)correspondence for morphological truncation by Benua (1995), McCarthy (1995), Kager (1999) among others.From the data discussed thus far, the base personal names represent an output, not input, just like their truncated forms.Correspondence in truncation is, therefore, assumed to be between elements (segments) of the base (B) and its truncated form (T), which is akin to that between the base (B) and the reduplicant (R) in reduplication.It is worthy to note that both the B and T are separate outputs and as such, each constitutes a free standing form.Hence, the need to analyze the patterns observed based on the correspondence between two output forms, but not between an input (I) and an output (O) as was the "original" motivation for the correspondence theory, though there is IO relation between the I and the B of the T. Before we begin the definition and evaluation of our working constraints, let us define what Correspondence is following McCarthy & Prince (1995), which is further expatiated by Beckman (1997) as follows.
(10) Correspondence (McCarthy & Prince 1995;Beckman, 1997, p. 13) Given two related strings S 1 and S 2 (underlying and surface), correspondence is a relation R from the elements of S 1 to those of S 2 .An element α Є S 1 and any element β Є S 2 are referred to as CORRESPONDENTS of one another when αRβ.
The employment of the OO-correspondence theory, as just explained, is necessitated by the need to map segments in the B (Output) to those in the T (Output).At the surface representation, usually there is truncation of syllables, particular initial syllable(s) of the surname.We employ constraints from the well-formedness family, as well as constraints on correspondence (from the faithfulness constraints family) such as MAX, DEP, and IDENT to account for the truncation patterns observed especially at morpheme boundary of some APNs in this paper.When we rank the former constraints above the latter, it will ensure that in deleting syllables, no segment is left behind.It is only in fewer cases that base forms are faithfully represented at the phonetic representation.The reverse of the ranking just discussed will typologically produce an output that does not attain a faithful mapping between base forms and the output forms in terms of segmental correspondence, not necessarily featural correspondence, when personal names interact at the phonetic level of the Akan grammar.We define our working constraints (constraints are taken from Benua 1995, Kager 1999;etc.)as follows; Volume 15 (1), February 2015(http://dx.doi.org/10.17576/GEMA-2015-1501-09)ISSN: 1675-8021

FAITHFULNESS CONSTRAINTS FAMILY
The following are some of the faithfulness constraints employed for analysis in the present paper.(Kager, 1999. pp. 264-272) (11) MAX-BT -Every element of B has a correspondent in T. form.(No truncation of elements).
This constraint has to be completed and dominated, since we have observed that in the truncated forms, it is either a syllable or two is deleted or the base is left intact in the truncated form.Therefore, a higher ranking of this constraint will account partially for only the instances where no truncation takes place as we just discussed.
(12) DEP-BT -Every element of T has a correspondent in B.
(No intrusion of elements).
This constraint that penalizes any addition to the T has to be ranked relatively high to severely punish any candidate that surfaces with more syllables than there are in the B. We have generally observed about the Ts that they are maximally identical to their Bs in term of syllable structure or they have less number of syllables.
Since we observed that truncation usually targets initial syllables in the APN, the requirement for identity between elements in B and those of the T is not segmental in the general sense, but in terms of syllable structure.Thus, we define our identity constraint as follows; (

13) IDENT-BT[σ] -
There must be identity between B and T in terms of syllable structure.
The identity requirement does not necessarily refer to the structure of the syllable in that sense of the word, but rather quantity of the syllables.For instance, if a CV syllable in the B phonologically manifests itself as a V in the T, we consider it identical i.e. monosyllabic, hence, the two become correspondents.For example, the B surname; [a.ma#ni.mo]surfaces as [a.me#e.mo] in the T.This constraint has to be crucially dominated by some markedness constraint that would set minimally requirement for the T. On the other hand, its higher ranking would block truncation from applying especially to polysyllabic base surnames.

MARKEDNESS CONSTRAINTS FAMILY
The following are some markedness constraints used for this analysis in this subsection.We begin with a templatic constraint that demands that a T be minimally equal to a disyllable (a foot) and again we follow Itô (1990), Kager (1999) and define this constraint as follows; (14) TRUNC=σσ -A truncated form minimally equals disyllable This constraint requires a morphological category; a T, to be minimally disyllabic.It has independent motivation for personal names in general in Akan as the minimal syllable a name consists of two syllables.A highly-ranked TRUNC=σσ will severely punish any candidate that deletes personal name (particularly the targeted surname) to something less than a disyllable in the T. The high-ranking of this constraint is evident in the fact that even where there was initially a truncation of the initial syllable of a base disyllabic surname, the lost syllable was compensated for by compensatory lengthening in the end to ensure the satisfaction of this minimality requirement.The minimality constraint as being used for the analysis in this paper follows the idea by McCarthy and Prince (1990, p. 25) that "the minimality constraint… sets a lower bound in terms of moras".In other words, as was seen in the data presented, Akan does not place maximality requirement on the truncated forms of the base surnames as it does with the disyllabicity on surnames.To account for the direction of vowel harmony discussed thus far, we employ the following sequential markedness constraints; (15) *-ATR, +ATR -Assign violation mark to a [-ATR] vowel followed by [+ATR] vowel.
The evaluation of candidates on this constraint is only local.It accounts for the value that spreads and at which direction the value spreads.It demands that whenever the vocalic segment in the initial syllable of the surname to the right has the value [+ATR], the final syllable of the day-name should not be specified with [-ATR].

ANALYSIS OF SYLLABLE TRUNCATION IN THE APN
In this subsection, I present an OT analysis of how APNs, particularly day-name + surname are realized at the phonetic level of the grammar.Let us note that our input forms for the tables represent the B of the names, which themselves are also outputs.

ANALYSIS OF TRISYLLABIC SURNAMES
Candidate (a) is always the faithful output; a mirror image of the B. In table 1, it fatally violates the constraint *+ATRC 0 -ATR, the constraint that penalizes the sequence of [+ATR] and [-ATR] across morphemes without any intervening consonant at the surface representation i.e. in the truncated forms.It also violates another well-formedness constraint similar to the previous one; *V#V, which bans a sequence of two vowels at the morpheme boundary.The difference between the two is that while the former specifies the ATR values of the vowels, the latter does not.Both candidates (c) and (d), which systematically truncate both the left and the right peripheral vowels of the B of the surname severely violate a highlyranked constraint; TRUNC=σσ since in their respective Ts only monosyllables surface.The competition for the optimality then falls to candidate (b), which minimally violates IDENT-

ANALYSIS OF POLYSYLLABIC SURNAMES
The truncation pattern observed in polysyllabic surnames is similar to that of the trisyllabic surnames with the only difference being that unlike in the trisyllabic surnames where only the initial syllable systematically deletes, in the polysyllabic surnames the first two syllables can be affected by this truncation as observed in (4a-4d).the other candidates, it is ruled out of the competition for optimality on grounds that it incurs a severe violation mark for violating *V#V which is relatively ranked higher than the other two constraints violated by the other candidates.
Having discussed how trisyllabic and polysyllabic surnames are truncated in tables 1 and 2 respectively, the attention is now turned to analysis of truncation of disyllabic surnames in table 3.

ANALYSIS OF DISYLLABIC SURNAMES
As we saw in (5), in some disyllabic surnames where there is truncation, because of the highranking of minimality constraint TRUNC=σσ, the grammar resorts to compensatory lengthening to ensure that truncated forms surface minimally as disyllable.With this foreknowledge, let us attempt an analysis of this pattern in table 3 before we conclude our discussions in this paper.
Volume 15 (1), February 2015(http://dx.doi.org/10.17576/GEMA-2015-1501-09)ISSN: 1675-8021 To account for some instance of ATR spreading in some base forms, we would need to introduce a directionality constraint that would require [+ATR] spreading to be leftward, as was observed in the data throughout.This constraint is SPREAD[+ATR]-L and is defined as follows. (
The spreading is within a domain i.e. to the immediately preceding syllable but not any syllable before it.Usually the initial V of the day-names preserves its ATR value i.e. it is immune from undergoing [+ATR] assimilation.Therefore, the constraint punishes a candidate that does not spread the value to the V in the syllable just across the morpheme boundary.This constraint has to dominate both *+ATR C 0 -ATR and *V#V so that a candidate will incur a more severely violation in not spreading [+ATR]

CONCLUSION
In conclusion, this paper has investigated APN, which generally falls under onomastics.It has focused on some morphophonological processes that are realized at the morpheme boundary of day-name + surname format of the APNs.It has been observed that in the truncation of some base personal names, some common morphophonological processes such as truncation, compensatory lengthening, vowel harmony, etc. occur across base forms.These processes do not occur randomly; but they systematically follow some patterns in terms of the quantity of the syllables of the surnames that can delete in a particular surname and which of the ATR values can spread and to which direction.Regarding the syllable size targeted for truncation, we observed earlier in this paper that whenever the base form of the surname is more than two syllables, in some, the targeted initial syllable may be truncated leaving behind the minimal two syllables.However, we observed again that when the base surname is disyllabic, it preserves its syllable structure in the truncated form through compensatory lengthening to satisfy a high-ranking templatic constraint imposed on all truncated forms of APNs.The templatic constraint employed to account for such minimality requirement on surnames was TRUNC=σσ.This constraint is so highly-ranked, as we have observed, that even when there is initial deletion of the targeted initial syllable, there is an invariable compensatory lengthening to ensure that the surname preserves its disyllabic minimality in structure in the truncated form.Marfo (2004) has made similar observations about compounding of (different) class of lexical items particularly Noun 1 -Noun 2 compounding in which in trisyllabic Noun 2 (as seen in 9b-9e), it is only the initial syllable that truncates.And when the Noun 2 is disyllabic as shown in (9a), just as we discussed about the APNs, seems to block the truncation rule.However, unlike in the personal name, the disyllabic minimality constraint is weak on some Noun-Adjective compounding that there are examples such as asɛm "matter" + papa "good" → asɛmpa "good news", kwasɪa "fool" + ɔkʊ "war" → nkwasɪakʊ "useless war".We can thus conclude that just as in the hypocoristic form of day-names (cf.Obeng 1997Obeng , 2001)), the surname also surfaces minimally in disyllable structure.What this implies is that at any given time, a complete typical APN is minimally of four syllables i.e. quadrisyllabic; two in the day-name and two in the surname.A similar disyllabicity claim has been made about the German and Spanish truncated names by Wiese (2001) and Piñeros (1998) respectively.According to Wiese (2001), German base names such as Andreas, Dagmar, Bastian, for example, are truncated into Andi, Daggi, Basti respectively.In a similar fashion, Piñeros (1998) reports that some Spanish base names: Alexandrina, Ernesto, Braulio have the following as their respective truncated variants; Dina, Neto, Balo.Again, according to Weeda (1992) and McCarthy & Prince (1998), as reported in Lappe (2002), English also realize disyllabicity in its truncated names Volume 15 (1), February 2015(http://dx.doi.org/10.17576/GEMA-2015-1501-09)ISSN: 1675-8021 in the y-suffixed hypocoristic variants such as Tommy for Thomas, Barbie for Barbara, and Aggie for Agatha.Cohn (2005) has also observed word minimality on truncated forms of Indonesian personal names and terms of address.According to Cohn (idem.),though there is bisyllabic minimality requirement on content words in Indonesian, such a restriction is weak in the truncation of personal names and terms of address where even a monosyllabic form may ensue as a truncated form.This is opposite to what obtains in the general morphology of Akan where there is no such minimality restriction on the syllable size words in other domains in the language.
Finally, it can be concluded that the findings in this paper have lent credence to the claims in the literature, particularly by Weeda (1992), Lappe (2002), among others that the syllable structure of truncated names is, to a large extent, similar across languages.That is, they are generally two syllables in minimum.

TABLE 1
], a constraint that is violated twice by all the other candidates except candidate (a) which satisfies it.In table 2, truncation of polysyllabic surnames in Akan is analysed.The same set of constraints and the same constraints ranking as previously shown in table 1is employed.

TABLE 3 .
Truncation of disyllabic surnames where suboptimal candidate emergesFrom table 3, with the same ranking of the same set of constraints as previously used, we get suboptimal candidate emerging.Based on table 3, candidates (c) and (d) are ruled out of the competition for incurring a severe violation of the highly-ranked TRUNC=σσ.Both candidates present only one syllable of the surname in the truncated form.This is a clear and fatal violation of the minimality constraint.Candidate (c) additionally violates the next highly-ranked *-ATR,+ATR, a constraint which is equally violated by candidates (a) and (b) as well.The race for optimality is then decided by *V#V, a constraint that is severely violated once by candidate (b); the expected most harmonic output.This situation leaves candidates (a), the suboptimal output, to emerge.To resolve this problem, we would need to account for the directionality of spreading of [+ATR] value.
value leftwards than it surfacing with contiguous vowels of the same ATR value or specified with [+ATR] and [-ATR] values, respectively at morpheme boundary.Now with this constraint added to the previous set of constraints, let us reanalyze truncation in disyllabic surname [ni.mo] which we unsuccessfully attempted in table 3 as follows, in table 4 below.

TABLE 4
With the introduction of SPREAD[+ATR]-L, we realize that candidate (a) which emerged an unexpected winner in table 3, now severely violates our new constraint for refusing to spread the [+ATR] value of the vowel in the syllable to the right of the day-name.The violation of this crucially-ranked constraint in table 3 has settled competition for the optimality between candidate (a) and candidate (b).The latter candidate outcompetes the former for satisfying the next highly-ranked constraint after *-ATR,+ATR; which both candidates equally violate.