cf. Data page

III. Analysis

Some aspects of Igbo phonology, namely, the vowel system and vowel interactions, must be considered before the data can be analyzed. I have not consid ered or marked tone in these data, partly because it can be shown to be independent of vowel quality (i.e., anticipatory assimilation can take place between two vowels in hiatus, but only quality, and not tone, assimilates: ÉÍ+è&Eacut e; becomes ÉéèÉ), and partly because the data in Enzeanya does not mark tone. The statements I will make involve vowel quality, and so I believe that tone can be safely ignored for the purposes of these data.

The vowel system of Igbo consists of eight vowels, with two distinct harmonizing (sub)sets (cf. Ladefoged 37)

Contrary to some older formulations (Carnochan 156), Igbo does not harmonize at the word level, but rather at the morphemic level (Onwuemene 11). Any vowel can be described using the features [back], [high], and [ATR ]. A general statement can be made that if two adjacent syllables have differing values for [ATR], they are from different morphemes. So even without referring to the grammar, we know that the name kanOsike is at least trimorphemic, since there are two po ints in the word where [ATR] changes. This is one way to determine where grammatical distinctions may be restricting hypocoristic formation.

There are various other rules relating vowel harmony to individual morphemes (Emananjo 17-27). The most im portant for are purposes are those instantiating the special status of the group described by the features [+bk +hi], namely /u U/. One example which operates independently of the data at hand will suffice to show this special status. There is a reduplica tive process in the formation of the simple gerund from a verb stem, which takes the following forms (Emenanjo 19):

Verbal root	Gloss	Simple gerund	Gloss
-si	cook	osisi	t he cooking
-je	go	ojije	the going
-bu	carry	obubu	the carrying
-go	buy	ogigo	the buying
-pI	squeeze	OpIpI	the squeezing
-zA	sweep	OzI zA	the sweeping
-zU	train	OzUzU	the training
-c hO	seek after	OchIchO	the seeking after

The shaded cells indicate the effect that final /U/ and /u/ have on the previous consonant: they require assimilation. The "elsewhere" or unmarked segments are /I/ and /i/, and [ATR] is also predictable from the environment (spreading from the stem vowel).No other segments show this feature across consonants.

To relate all of this to the pros odic restraint on hypocoristic construction, we consider the alternate form chukwu and chi. Their semantics are very similar, and Praise used the two forms almost interchangeably as subconstituents of names. She often said forms like Chukwujioke and Chiji oke were "the same thing." Chukwu is actually bimorphemic, consisting of chi "God" and ukwu "great." There is no related free form *chu. (cf Welmers 16); it is strictly a harmonic form resulting from the presence of kwu.

This is where the interpla y between harmony and hypocoristics becomes complicated: there are hypocoristic forms cited by both Ezeanya and Okoh which the formant chu without what is apparantly its liscensing morph kwu.. E.g., chude for chukwudebelu, chuke for chukwukelU. It is wort h noting here that Praise actually had different intuitions with regard to the grammaticality of the form chude: She considered it ungrammatical, and was surprised that it was cited as such in Ezeanya. Nevertheless, she felt that chike was an acceptable h ypocoristic for Chukwukelu. We will return to this problem after considering the general structural constraints on hypocoristics.

Igbo syllables are of the following types: V, CV, CVN, and N (the syllabic nasal). Vowels in hiatus across morpheme b oundaries may or may not undergo anticipatory assimilation in quality (but not tonally, as mentioned above). If we define § as a syllable as stated above, and $ as either N or V (a mora), we can state the structural template for hypocoristics as foll ows: [($)§§]. This says, a hypocoristic has 2 syllables and an optional initial moraic segment. Following Broselow and Marantz (1983) cited in Katamba (185):

In this case, the final segment of the word is selected, and so association is from right to left (There is a tendency in male names, according to Praise, to circumscribe that portion of the name which is not Chukwu.. She was quite emphatic in pointing this out; I got the impression that it is avoided as a rule, because it is so com mon. As a female nickname she said Chichi and Chi are very common, and can be use for anyone girl whose name contains Chukwu.):

	
	C VC VVCVCV	   	   $§ §	
	| ||\||||||		      ||\|\
	chukwuekeka		chukwuekeka
 

Here is the evidence for the mora plus two syllables rather than a trisyllabic template, as the preceeding high vowel is not included to make *wekeka.

There also seems to be a dispreference (probably related to a semantic restraint against splitting morpheme s) for taking the optional moraic segment from a previous morpheme. Here's an example where the final /u/ in the ubiquitous Chukwu is not used to fill the optional moraic slot in the template (* shows where the optional moraic segment has been stray-erase d):

 
	C VC VCVCV			$§ §
	| ||\|||||		     ||\|\
	chukwubiko		chukwubiko
 

Here, an unnattested Ubiko would presumably also match the template (cf amechi, amaka, eloka, etc.), and only semantics is available as an e xplanation.

The previously mentioned rules of assimilation in vowel hiatus also are also relevant to hypocoristics. There is a tendency to shorten vowel clusters, even if anticipatory assimilation has already taken place (+ is a morpheme boundary; the assimilation takes place here):

								
	VCV VC V					   $§  §	
	||*\|| |		phonetically:	||\*|\ 
     ama+echi		[ameechi]	    ameechi
           

The second e is dropped. (The tonal tier, if included here, would be unaf fected by this assimilation.) A similar tendency can be seen in the following example, which has syncope, so we use left-right mapping when forming the hypocoristic:

 
	VCVVCV		$§ §
	||||||		||\*
	ijeoma		ijeoma
&nb
sp;

Thus the tactics of Igbo prefers Ije to Ijeo. An alternative solution to this situation produces another grammatical hypocoristic for this name (Ijeoma is Praise's best friend, so her judgements about the grammaticality or ungrammaticality o f all possible outputs for this form are quite strong.). This one involves reduplication, but only of the first complete syllable:

 
	VCVVCV		$§ §
	||||||		*|\|\
	ijeoma		ijeje = jeje

This process, too, is quite productive, yielding forms like keke from kelechukwu, and chichi, from any form which contains chukwu.

Let us turn again to the appearance of chu in hypocoristics, this time considering it from in the light of the template, which I hope is now soundly established. Remebering that chu only appears in the environment of a [+hi +bk] vowel /u/ or /U/ in the following syllable, we have some good evidence for segment copying, because there are hypocoristics which contain chu before other syllabl es:

	
	C VC VCVCVCV		$§     §
	| ||\|||||||		*|\  * |\ * *
	chukwudebelu		 chukwudebelu	= chude
 

So it appears that the formation of the hypocoristic takes place after the application of vowel harmony. B ut the existence of an alternative grammatical form chide shows that this is not always the case. Indeed, the ordering seems to be unpredictable, as Praise's judgements in this sort of situation differed from those of Ezeanya.

As a final issue, I would like to address the forms which don't obey the template. Some of these Praise hardly considered hypocoristics: anOsike for kanOsike, she explained differed only in the presence of kà, a highly grammaticalized morpheme whos meaning was so abs tract she had difficulty summing it up in a single word. Similarly, the previously mentioned synonymy between chi and chukwu are discounted by Praise as being "pretty much the same word." Thus chukwujioke and chijioke are barely considered by her to be sh ortened forms. nkemfUlUewlum has several forms, but it should be noted that the final /m/ in nkem and the initial /m/ (actually a labiodental nasal) are a single morph meaning "I" or "me," and are not the result of splitting a morpheme. The unusual form, nkemfUlU, with 4 instead of two syllables, is explained by Praise as having a separate meaning "I took what I saw." She also offered explanations for the other forms in the final column of the data sheet, all which relied on semantics. She has the intuiti on, in other words, that these forms are atypical of hypocoristics.

IV. Works Cited

cf. Works Cited