Comparative Analysis on Kikuyu
Comparative Analysis on Kikuyu within the SPE and Post-SPE Framework
Term Paper for Foundations of Phonology Course
Introduction
This paper aims at observing the data set of Kikuyu language. Two frameworks are compared, i.e. the Sound Pattern of English (henceforth; SPE), and Post SPE (Autosegmental Phonology). Within the discussion, approaches form other theories will also be highlighted; however, the main point addressed in this paper is to evaluate the data set of Kikuyu on the basis of phonological theories within the two frameworks involved.
Based on SPE theory, the lexical entries should consist of sufficient information for the phonological rules in order to identify its phonetic forms for each context. In other words, each lexical entry is entered as a set of phonological distinctive features. Furthermore, the underlying representation (UR) is considered as an abstract representation in comparison to a surface representation (SR). Along the paper, we will discuss both frameworks together with feature notations and then we will analyze the data set segmentally to obtain the rules governing the language of Kikuyu.
In the next section, we will attempt to analyze the variability of the coordinated articulary apparatus with the spirit of the Post-SPE framework driven by a number of questions in the optimization of the data set analysis of with respect to the framework inquestion. During the analysis, we will not consider some basic requirements such as No Crossing Constraint and Linking Constraint in order to be consistent with the well-formedness condition of Post-SPE framework. Furthermore, we will also approach the so-called geometry of phonetic representations accompanied by sufficient examples to figure out any possible solution.
One of the central issues addressed within this paper is the analysis on the shifting of nasal + consonant with respect to the given data set. We will try to compare two theories in question and observe how those theories could account for the changes in the data set. At a later stage, we will see which theory casts better analysis of the given data than the other does.
Some literatures will be considered especially those from the textbook of Phonological Theory: The Essential Readings by Goldsmith, J. such as The Sound Pattern of English by Chomsky, N. and Halle, M., among others, as well as relevant sources which might give us more information about the language of Kikuyu. Now, let us go through the analysis from the first framework, i.e. the SPE framework.
The SPE Framework
The SPE framework is believed to be the basis of Generative Phonology since theories within this framework were influenced by the views from generative linguistics. Chomsky and Halle (in Goldsmith, 1999 : 17-19) states that a speaker’s knowledge of his language consists of knowing the lexical items of the language and each lexical entry must contain specified features, which determine the phonetic form of the item in all contexts, i.e. the item’s phonological features. Furthermore, such phonological features are classificatory devices, they are binary, as are all other classificatory features in the lexicon, for the natural way of indicating whether or not an item belongs to a particular category is by means of binary features.
There will be two levels of representations that will be discussed in the SPE framework; underlying representation (i.e. lexical or morphophonemic sequence) and the surface form (i.e. phonetic output form). Given the authors’ aim at maximizing the ‘simplicity’ of the grammar, it follows that underlying representations should be as abstract as possible and avoid redundant, or non-distinctive, features. Minimized underlying representations are indeed a requirement to ensure the generality of the overall linguistic system.
Within this framework, we will analyze the data set on the language of Kikuyu in the spirit of morphological analysis, overview on any possible alternation within the dataset, as well as identifying the underlying representation (UR) from the given data set. Following this analysis, we will try to observe possible rules with feature notation of the given data in order to get a generalized rule ordering within the data set. For this analysis, we also refer to the International Phonetic Alphabet (IPA) chart, especially for the consonants chart and their features. Let us try to analyze the data set from the morphological analysis with the given data set below.
The data set of Kikuyu is given below in table 1:
Imperative---1 sg.Imperfect--English Meaning
ßura------mbureet?--------'lop off'
ßaara-----mbaareet?-------'look at'
t?ma------nd?meet?--------'cut'
toma------ndomeet?--------'send'
reha------ndeheet?--------'pay'
ru?a------ndu?eet?--------'cook'
cina------??ineet?--------'burn'
koma------?gomeet?--------'sleep'
kera------?gereet?--------'cross'
?ora------?goreet?--------'buy'
?aja------?gajeet?--------'divide'
From table 1, we can see that there are two forms to observe, the one being Imperative and the other being the 1-sg-Imperfect from data set of Kikuyu language, which is followed by its meaning in English. The table shows that for each given word, several letters remain unchanged, which is typed in Bold inside the word. Most of them seem to have a Vowel and Consonant order (VC) and a long vowel one (i.e. VVC) such as in ßaara. Furthermore, we can also see in the Imperative column in which all words are always ended with a, marked after a slash sign (-). Subsequently, the words in the first singular Imperfect column are always ended with an ‘eet?’, which is also separated by a slash (-). A complete data set analysis is given in table 2 below.
Table 2
Imperative-------------1 sg. Imperfect------------English Translation
ßur - a--------------------mbur - eet?----------------------'lop off'
ßaar- a--------------------mbaar- eet?----------------------'look at'
t?m - a--------------------nd?m - eet?----------------------'cut'
tom - a--------------------ndom - eet?----------------------'send'
reh - a--------------------ndeh - eet?----------------------'pay'
ru? - a--------------------ndu? - eet?----------------------'cook'
cin - a--------------------??in - eet?----------------------'burn'
kom - a--------------------?gom - eet?----------------------'sleep'
ker - a--------------------?ger - eet?----------------------'cross'
?or - a--------------------?gor - eet?----------------------'buy'
?aj - a--------------------?gaj - eet?----------------------'divide'
From table 2, we can see that the unchanged letters, which are Bold typed above, are the stems or could be part of the stems of the word in underlying representation. Furthermore, we can also see suffixes, which indicate the Imperativeness or the given words that are signaled as the final letter ‘a’ at the final position of the word. From the regularity of the final letters ‘eet?’, we can say that the given words must be classified as suffixes indicating the 1 sg. Imperfect form of Kikuyu language. We will discuss the underlying forms of the morphemes regularity in a separate discussion in a later part. Now let us see the structure of the nasal sound which occurs before the stems.
It is conceivable that an alternation is defined as a morpheme, which has two different sound shapes, which can be analyzed by a phonological process. From the data set of Kikuyu, the some alternations can be observed as indicated in the following table (see table 3). The alternations can be identified easily in that ß becomes b; t /r becomes d; c becomes ?; k / ? becomes g. Those alternations can be viewed as the alternations in the language of Kikuyu whose phonological process will be explored in rules.
Table 3
Imperative------------1 sg. Imperfect
In a.b.ß-------------------------mb
In c.d.t-------------------------nd
In e.f.r-------------------------nd
In g.c---------------------------??
In h. i. k-----------------------?g
In j. k. ?-----------------------?g
Furthermore, we can also see that a nasal consonant is inserted before the changed consonant, e.g. m; n; ? and ?, which suggests that the morphological process goes along with the phonological process. Such insertion shows us important points for the data set in the language of Kikuyu. We will discuss such phenomenon in greater concern in later part. However, there is one thing to say about this phenomenon in the insertion of nasal consonant in the 1 sg. Imperfect groups could be analyzed as certain prefixes embedded which might express the tense of a verb. Whenever the alternation is constructed in an opposite way, i.e. b becomes ß in data set, this lead to an ill formed construction. Such a case is also falsifiable from data h. and j. in which ?g would become k and ? respectively following the case. Therefore, we will consider the sequence of alternation as from Imperative to 1 sg. Imperfect. The reanalysis of the stems of both forms is illustrated in table 4 below. Both prefixes (nasals) and suffixes (– a and – eet?) are discarded in table 4 so that we can get the stem of each verb.
Table 4.
Imperative--------------1 sg. Imperfect--------------English Meaning
ßur---------------------bur-------------------------------'lop off'
ßaar--------------------baar------------------------------'look at'
t?m---------------------d?m-------------------------------'cut'
tom---------------------dom-------------------------------'send'
reh---------------------deh-------------------------------'pay'
ruc---------------------du?-------------------------------'cook'
cin---------------------?in-------------------------------'burn'
kom---------------------gom-------------------------------'sleep'
ker---------------------ger-------------------------------'cross'
?or---------------------gor-------------------------------'buy'
?aj---------------------gaj-------------------------------'divide'
What we have observed so far indicates that phonological components are obtained by mapping from the underlying representation (UR) to the surface (phonetic) representation (SR). This mapping phenomenon can be observed by rewrite rules which will be discussed in a separate part. In other words, the data set of Kikuyu we have so far could be considered as the Surface Representation. In the following paragraphs, we attempt to identify the underlying representation of the Kikuyu language.
As noted above, we have seen that the unchanged letters in table 2 could be analyzed as the stems or part of the stems of the words in the underlying representation. On the basis of minimization of the underlying representation we will attempt to rule in the consonant before the unchanged letters because it seems to be implausible to predict the consonants such as ß, t, r, c, k, ? by rule. Furthermore, we have also observed that the consonant ß can shift to b not the vice versa. Referring to the SPE theory, minimizing the underlying representation means that anything, which can be predicted by a rule, should be eliminated from the underlying representation. For instance, the shifting from ß to b can be viewed in table 4 which is exemplified by the shifting is from ßur to bur. Such process also applies to all other words in the data set. By definition, we could get something like /ßur/ to be the actual stem for underlying representation of the word, which means ‘lop off’ in Kikuyu. The stems in underlying representation in the data set are presented in table 5 below and the Underlying Representations for the Imperative and 1 sg. Imperfect are represented in table 6.
Table 5
UR Stem English Meaning
/ßur/ 'lop off'
/ßaar/ 'look at'
/t?m/ 'cut'
/tom/ 'send'
/reh/ 'pay'
/ru?/ 'cook'
/cin/ 'burn'
/kom/ 'sleep'
/ker/ 'cross'
/?or/ 'buy'
/?aj/ 'divide'
Table 6
Imperative UR 1 sg. Imperfect UR English Meaning
/ßur - a/ /Nas - bur -eet 'lop off'
/ßaar - a/ /Nas - baar-eet 'look at'
/t?m - a/ /Nas - d?m -eet 'cut'
/tom - a/ /Nas - dom -eet 'send'
/reh - a/ /Nas - deh -eet 'pay'
/ru? - a/ /Nas - du? -eet 'cook'
/cin - a/ /Nas - ?in -eet 'burn'
/kom - a/ /Nas - gom -eet 'sleep'
/ker - a/ /Nas - ger -eet 'cross'
/?or - a/ /Nas - gor -eet 'buy'
/?aj - a/ /Nas - gaj -eet 'divide'
In the framework of SPE, we have been familiar with the terms such as abbreviatory conventions, conciseness, Minimize UR, Rule format and Evaluation measures, etc. They will be considered here under IPA consonant chart and feature table where relevant information is given in table 7 below:
Table 7
(Imp = Imperative) (1sg = 1 sg. Imperfect)
Group A (data a. b.)
ß - bilabial fricative (Imp)
b - bilabial plosive (1sg)
m - bilabial nasal (1sg)
Group B (data c. d. e. f.)
t - alveolar plosive (Imp)
r - alveolar fricative (Imp)
d - alveolar plosive (1sg)
n - alveolar nasal (1sg)
Group C (data g.)
c - palatal plosive (Imp)
? - palatal plosive (1sg)
? - palatal nasal (1sg)
Group D (data h. i. j. k.)
k - velar plosive (Imp)
? - velar fricative (Imp)
g - velar plosive (1sg)
? - velar nasal (1sg)
From the distribution in table 7, we can draw some important information in the surface level. In Imperative classification (Imp), we can observe that the fricatives are plosives whereas in (1sg) group, we only observe the plosives ones. This observation is helpful for arriving at the deduction that under certain environment, fricatives/plosives are interpreted as (à) plosives. Furthermore, within each group, we can identify that the same place of articulation is shared, i.e., bilabial / alveolar / palatal / velar. This observation will contribute to identify the relation between the changed consonants and the added nasal sound types.
In the mean time, within each ‘1sg’ group, we can also observe that the nasal sound always precedes the plosive sound. This observation is useful for understanding whether the prefix [Nasal] functions are in a linear order. By applying the minimized major feature for these consonants, we can generate some crucial feature notations as noted below.
1. Fricatives [-son, +cont]
2. Plosives [-son, -cont]
3. Nasals [+son, -cont]
Therefore, now we can observe the assimilation of the feature [cont] in the course of transferring from fricatives to plosives and probably the dissimilation of the feature [son] between nasals and plosives. In order to satisfy the conditions of Minimize UR and the Evaluation measure, we could observe each sound in detail and add the feature [voice] where we can see that all plosives and nasals are [+voiced] as illustrated in table 8 below.
Table 8
Features - consonant Features-place of articulation
Group A (data a. b.)
ß - [-son] [+cont] [+voiced] (Imp) [+ant] [-cor]
b - [-son] [-cont] [+voiced] (1sg) [+ant] [-cor]
m - [+son] [-cont] [+voiced] (1sg) [+ant] [-cor]
Group B (data c. d. e. f.)
t - [-son] [-cont] [-voiced] (Imp) [+ant] [+cor]
r - [+son] [+cont] [+voiced] (Imp) [+ant] [+cor]
d - [-son] [-cont] [+voiced] (1sg) [+ant] [+cor]
n - [+son] [-cont] [+voiced] (1sg) [+ant] [+cor]
Group C (data g.)
c - [-son] [-cont] [-voiced] (Imp) [-ant] [+cor]
? - [-son] [-cont] [+voiced] (1sg) [-ant] [+cor]
? - [+son] [-cont] [+voiced] (1sg) [-ant] [+cor]
Group D (data h. i. j. k.)
k - [-son] [-cont] [-voiced] (Imp) [-ant] [-cor]
? - [-son] [+cont] [+voiced] (Imp) [-ant] [-cor]
g - [-son] [-cont] [+voiced] (1sg) [-ant] [-cor]
? - [+son] [-cont] [+voiced] (1sg) [-ant] [-cor]
From the features distribution above, we can observe several important generalizations. First, as we can see in the left column, consonants could adjust themselves to the nasal during the shifting to plosive by preserving [-cont] and [+voiced]. Then, the consonant r in Group B is the only consonant in Imp which has feature [+son], therefore, we have to shift it to become [-son] as well in the course of shifting to plosive. However, this is definitely not a kind of adjustment to the nasal since it is conceivable that all nasals are found to be [+son].
Second, in the right column, we can observe that the insertions of prefixing nasals adjust themselves to the consonants and become bilabial / alveolar / palatal / velar nasal respectively in place of articulation. Furthermore, we attempt to figure out the adjustments above as assimilation, as well as the “dis-adjustment” of r as dissimilation. In sum, we can obtain two important rules in the language in question.
Rule A:
[-cont]
[+cons] à [+voice]/[+nas] ______
[-son]
Such rule entails the adjustment of the consonants to the nasals. Under this rule, ß becomes b; t and r becomes d; c becomes ?; k and ? becomes g due to the prefixing nasal. Then, all fricatives become plosives as given the following rule.
Rule B:
[+nas] à [aPlace]/______[aPlace]
In rule B, the same place of articulation is construed by the notation [aPlace] represents here. This rule signals the adjustment of the nasals to the consonants. By using this rule, nasal will become m whenever aPlace is bilabial; n whenever aPlace is alveolar; ? whenever aPlace is palatal and ? whenever aPlace is velar. The origin nasal in prefixing cannot be observed from the given data set. For instance, if it is [+nas, +cor] then it should be a consonant ‘n’.
Along the previous part, we have observed some ordering rules in the language of Kikuyu. Further in this part, we will try to figure out the ordering of the two rules we have mentioned in advance. Based on the SPE framework, we have seen that all morphological rules apply before all phonological rules. Although Kiparsky and many others believe the other way against this framework, we try to see whether this rule is indeed workable within the spirit of SPE framework.
Here, the morphological rules are viewed as infix insertion, i.e. prefix /Nas/; suffix /-a/ and /–eet?/ in the given data set which will apply in the first place and the phonological rules given in Rule A and Rule B above will apply in the second place. Now the question turns to which phonological rule applies in first order, being Rule A or Rule B. Suppose we take examples in data d and e from the given data set. First, let us look at data d. it seems that we will not find any difference in data d with respect to the ordering of two rules and hence the result looks like the same. Then, try to compare with data e. Let see what happens.
Table 9
Data d = from [toma] to [ndomeet?]
If Rule A precedes Rule B
Morphological Rule /Nas/ + /tom/ + /eet?/ =UR
Rule A /Nas/ + /dom/ + /eet?/
Rule B /n/ + /dom/ + /eet?/ =SR
If Rule B precedes Rule A
Morphological Rule /Nas/ + /tom/ + /eet?/ =UR
Rule B /n/ + /tom/ + /eet?/
Rule A /n/ + /dom/ + /eet?/ =SR
Data e. from [reha] to [ndeheet?]
If Rule A precedes Rule B
Morphological Rule /Nas/ + /reh/ + /eet?/ =UR
Rule A /Nas/ + /deh/ + /eet?/
Rule B /n/ + /deh/ + /eet?/ =SR
If Rule B precedes Rule A
Morphological Rule /Nas/ + /reh/ + /eet?/ =UR
Rule B /n/ + /reh/ + /eet?/
Rule A /n/ + /deh/ + /eet?/ =SR
As we can see above, the comparison between data d and e indicates the same thing for ordering of two rules as well. This suggests that the ordering of rules do not have something to do with the result. It further suggests that the nasals and the derived consonants could have a relatively strong connection with each other, i.e. one co-exists with the other.
Within the SPE framework, the data set has been analyzed in the form of features of lexical individual segments. We tried to apply the rules we have and they seems to work well. This gives further indication that the SPE framework can be considered as effective tool in explaining the phenomenon in data set of Kikuyu language. However, it seems that the relationship between the two rules is not transparent. The reason for this might be the difficulty in generating the only one rule instead of two explaining such phenomenon in the SPE framework.
The Post-SPE Framework
Autosegmental phonology was introduced by John Goldsmith (1976) and re presented a great step forward in linguistic research. In the classical generative theory developed by Chomsky and Halle, phonological components were linear sequences of segments which themselves consisted of feature bundles. One of the downfalls of the
SPE framework resided in the assumption that every segment had to correspond to exactly one feature specification and vice-versa (every feature specification had to correspond to exactly one segment). Consequently, many phonological phenomena (related to stress, lengthening, rhythm and intonation for instance) were left unexplained.
SPE simply had no way of providing a proper account of prosodic phenomena. As stated by Goldsmith himself ((1999: p.137), “Autosegmental phonology constitutes a particular claim about the geometry of phonetic and phonological representations. it suggests that the phonetic representation is composed of a set of several simultaneous sequences of these segments, with certain elementary constraints on how the various levels of sequences can be interrelated or ‘associated’. The main innovation brought about by Post-SPE framework is the fact that supra segmental features, such as stress or tone, are no longer confined to exactly one segment but can be shared by two or more segments and vice-versa. Some features, while remaining associated with a segment, are now handled separately. Many phonological phenomena can then be analyzed in terms of a restructuring or reorganization of the autosegments in a representation.
Different from the first section, here we will analyze the data set of Kikuyu within the Post-SPE framework, or also known as the Autosegmental Phonology but within this paper we rather use the term Post SPE just for the sake of easiness in comparison. In this section, we will also consider the use of the IPA chart and feature notations for the discussion in the Post-SPE framework. Also in this part we might still view some rules from SPE framework and will be reviewed in the spirit of Post-SPE framework. Phenomenon such as assimilation and dissimilation will be approached to address the problems we found in the data set of Kikuyu Language. Of course, other theories within the Post-SPE will also be introduced. Let us start our discussion with the identification of features and spreading in this framework.
In this framework, features will be viewed as independent features of their segments in order for them to be represented as auto-segments. We can observe in the data set of Kikuyu language that a phonological process can influence more than one consonant at a time. This might lead us to review the data we had in table 3 where ß – mb; t – nd; r – nd; c – ??; k – ?g; ? – ?g, with the given two rules. We have mentioned earlier about assimilation in SPE, i.e. an alternation which copies a feature specification from the closest segment. Take Group A and Rule B as an example:
[+nas]à [bila Place] / ______ [bila Place]
Nà m / ______ ß / b
In the Post-SPE framework, assimilation is identified as the spreading of features to the neighbouring X-slots. This means that an X-slot is associated with the two X-slots, resulting assimilation as illustrated below:
NX NX mX
|àÂ¥|àÂ¥| [+ant][+ant] [+ant]
[-cor] [-cor] [-cor]
For the sake of easiness, some tiers are excluded. Here, X represents the changed consonant in Kikuyu. The nasal is considered another X slot, which is marked as N. After spreading, we see that the two X-slots share the feature of [+ant] and [-cor]. The feature from the neighboring X-slot could be deleted after the course of spreading. The replacement of Rule A in the Post SPE framework is a bit more complicated than Rule B since deletion is involved. As for instance, consider Group A we have noted above, within the SPE framework, we will get the shifting below.
[+cons]à [-cont] [+voice] [-son] / [+nas]______
ßà b / N______
Then, let us apply this within the Post SPE framework to analyze this phenomenon where we have two X-slots represent Nasal and Consonant respectively during the course of alternation. An example of shifting from Nß to Nb is given below.
X X X X
| | à | /
[+son] [-son] [+son] [-son]
[-cont] [+cont] [-cont] [-cont]
[+voiced] Â Â Â Â Â Â [+voiced]Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â [+voiced]Â Â Â Â Â Â Â Â Â Â Â Â Â [+voiced]
For the sake of easiness, some tiers are reduced. We can see in the shifting process that [-son, +voiced] features are preserved during the shifting and [+cont] feature becomes [-cont]. In the SPE framework, it is conceivable that morphological rules apply before phonological rules, so we have a Nasal slot before Consonant slot in 1 sg. Imperfect in Kikuyu. When we analyze further in the Post SPE framework during the shifting, [+son and -son] features reduces to [-son], then [cont and voice] features shift to [-cont, +voiced]. This suggests assimilation to the nasal feature. All consonants following the nasal are still [-cont, +voiced] but this is still regarded as process of assimilation. Therefore, the same place of articulation of nasal adapting to the following consonant is considered a process of assimilation.
Different from assimilation, let us discuss about the dissimilation in the given data set for the feature [son]. In table 3, we have seen that in data e and f, r becomes nd. The analysis under Post-SPE is figured out in the following diagram.
ràNÂ +Â Â Â ràNÂ Â +Â Â ràNÂ +Â dànd
XXÂ Â Â Â Â Â Â XXÂ Â Â Â Â Â Â Â Â Â XXÂ Â Â Â Â Â Â X
|à|Â Â Â Â Â Â |à|Â Â Â Â Â Â Â Â Â Â -|Â Â Â Â Â Â Â /
[+son][+son]Â Â Â Â [+son][+son]Â Â Â Â Â Â Â Â [+son][+son]Â Â Â Â Â Â Â Â Â Â Â [-son]
[+cont][-cont] [+cont][-cont] [+cont][-cont]Â Â Â Â Â Â Â Â Â [-cont]
[+voiced] [+voiced][+voiced]Â Â [+voiced][+voiced]Â Â [+voiced][+voiced]
In the above diagram, we have reduced some tiers for the sake of easiness. When the lines are associated, the line behind X is deleted, which is marked as (-). This result the removal of the feature [+cont] [+voiced] and the slot receives new features from features under the N (nasal) slot. However, we see that feature [+son] turns into [-son] during the course of shifting. Therefore, this leads to an assimilation process towards nasal sound since nasal sound is [+son] and all other consonants do not change their [-son] features. This is problematical. Therefore, we might approach it within the spirit of “Geometry of Phonological Features” by Clement. He suggests that there are three types of assimilation; total, partial and single-feature, relying much on the position of the spreading element in the tiers. We see that [son] and [cont] generate on the same tier, i.e. manner tier. Therefore, classifying this phenomenon as partial and single feature assimilation seems to be implausible since the assimilation of the data set has involved more than one feature. It seems that dissimilation with such theory cannot be preserved. Now let us try to account for the phenomenon with the X theory. Consider the following diagram with an example the word ßura in data a.
Data a:
    s                          s             hierarchical syllable structure
   /   |               /  |
   /R               /   R             (Rhyme)
   /   |             /   |
    ON             ON             (Onset - Nucleus - Coda)
   | |              |             |
    XX             XX             skeleton (no feature [±syllabic] in X theory)
   | |             |            |
    ßu             ra            autosegmental features
The two segments such as nasal and plosive behave like as if one segment. These segments can be considered as pre-nasalized consonants which might be marked in other forms. Similar phenomenon can also be found in other languages such as Bantu language. With respect to Kikuyu, it seem to be implausible for such a language to have a Nasal+Consonant (NC) cluster in 1 sg. Imperfect being a prenasalized consonant. Since the NC in the data set of Kikuyu is actually one X-slot it seems to be generalizeable that such a language does not allow two [+son] in one slot and one of them has to go [-son], which is in this case the consonant one. In other words, in NC cluster nasal sound is pronounced with [+son] and NC share the same [-cont, +voiced] features. Further, on this issue, we may need to address such phenomenon with another theory, which is still in the spirit of post SPE, i.e. the Mora theory
The discussion within this approach might give better understanding on the phenomena of mono segmental NC cluster. In Mora theory, however, it seems to be more difficult to propose further analysis on the given evidence for the changes, which have been involved in the language of Kikuyu. In this theory, an onset is what might be able to be changing in this language since an onset consonant is irrelevant to mora due to the absence of weight. It could be arguable that since the onset consonants do not count for timing (Van Oostendorp), the slot might only be one slot for the NC sounds instead of two slots for such a language. Consider the diagram below, which might suggest the first step of morphological alternation within the Mora theory.
       s...                             s...                                   s...
      /             |                 //             |                     /|
       /µ             à               //             µ        or            /µ
       /             |                 //             |                      /|
       CV            C              CV                     CV
        ß             u                  Nßu                                Nßu
Within this section, we attempt to account for the occurrence of spreading and assimilation within the Post-SPE framework. Certain phenomenon, which used to be a bit complicated to explain under the previous framework of phonological representations, is getting much more revealed if we try to observe any phenomena from the perspective of the Post SPE framework. The observation on the phenomenon of the Kikuyu data set seems to be better explained under the latest theory yet the nature of the data set should be preserved in such a good way, as Goldsmith proposes.
Conclusion
So far, we have tried to account for the phenomenon of Kikuyu language within the two frameworks, i.e. SPE and Post SPE. We have seen both strengths and weaknesses. By and large, the main difference between the two frameworks, SPE and Post SPE is that in SPE, data set of Kikuyu is analyzed within segmental level whereas in the later framework, the data set is analyzed in an auto-segmental level, hence it is called Autosegmental phonology. Further distinctions can also be captured within the notion of assimilation. In SPE framework, assimilation is construed as kind of copying process whereas in the later framework, assimilation is analyzed as spreading. Furthermore, in SPE we seems to have one-to-one which is defined as mapping whereas the later framework, the X position can be associated with zero, be it one or two autosegments.
In the SPE framework, we obtain the underlying representations on the basis of the features analysis in which at the later stage, we could generate two rules to represent the phenomenon of shifting in Kikuyu data set. In the later framework, we do not deal with rules, instead spreading of association lines and assimilation to account for the problem are involved. When we attempt to incorporate the two rules in SPE framework into one single spreading in the post SPE framework, some obstacles such as dissimilation of [son] could be encountered. We tried to propose some possible solutions to the problem in question; however, we seem to have some remaining problems. For example, when we try to propose one assumption of mono-segment instead of bi-segment for NC cluster within the Post SPE framework, but still there are some questions left.
In conclusion, both frameworks have their own strengths and weaknesses. In the SPE framework, we found that the rules are considered as well explained to deal with the given phenomenon. However, this framework is still unable to reveal the inner relationship among the changing elements in Kikuyu. In the later framework, the analysis is pretty much closer to the inner relationship but we see that there some exceptions that need to be considered. If we have to choose, we would say that the Post SPE framework seems to be much more helpful than that of the first framework regardless some exceptions exist. Following Goldsmith, the Post SPE framework contributes greater in figuring out the nature of Kikuyu language than the first one since it could approach some conditions of Kikuyu language better.
References
Chomsky, N. and M. Halle (1968). “Phonetic and Phonological Representation”. In
Goldsmith (1999): Phonological Theory: The Essential Readings. Blackwell Publishers(pp.17-21).
Chomsky, N. and M. Halle (1976). The Sound Pattern of English. New York: Harper and Row.
Clements, G.N. and S.J. Keyser (1983). CV Phonology: A Generative Theory of the Syllable.
Cambridge: MIT Press. Van Oostendorp, M (2005). Mora Theory. p1-8
Goldsmith (1999): Phonological Theory: The Essential Readings. Blackwell Publishers.
Goldsmith, J. (1976). An Overview of Autosegmental Phonology. Phonological Theory: The Essential Readings. Blackwell Publishers.
Class Handouts
Session 1 – Classical Generative Phonology (2008.Sept.12)
Session 2 – Summary of standard features (2008.Sept.19)
Session 4 – Autosegmental phonology I – features (2008.Oct.3)
Session 5 – Autosegmental phonology II – the CV skeleton (2008.Oct.10)
Session 6 – Autosegmental phonology III – the mora (2008.Oct.17)
Order Now