5.3.3.1 The MCA interpretation of “uvularized” and “epiglottalized” consonants
In classifying the consonants with the “guttural release type”, Miller-Ockhuizen (ibid.) has introduced two features for secondary articulation that are new to Khoisan language descriptions, namely, “uvularization” and “epiglottalization”. Below, discussing these two features, I re-interpret the “uvularized” and “epiglottalized” consonants by using MCA.
In her interpretation, “uvularization” applies to five non-click stops with uvular fricative release (/t t d dz d/) and to eight clicks with uvular fricative release (/|
! |! /), as seen in Table5.23. It should be pointed out that uvularization is not only new to Khoisan language descriptions, but also uncommon in the world’s languages. Thus, Maddieson’s (1984) survey on the phonemic inventory of the UPSID languages does not include uvularized consonants; Ladefoged and Maddieson (1996) do not use uvularization in their linguistic phonetic investigation of the world’s languages; the IPA framework does not prepare notation for this type of secondary articulation; Odden (2005: 145) states that uvularized consonants are rare, and does not mention examples of them except for “uvularized clicks... attested in Ju|’hoansi”, which is apparently after Miller-Ockhuizen’s interpretation. Therefore, Miller-Ockhuizen’s interpretation creates a series of consonants in Ju|’hoansi that require an unestablished idiosyncratic feature.
MCA, avoiding the use of this idiosyncratic feature, interprets her “uvularized” consonants as the consonant clusters in which the cluster offset is an independent phoneme //, i.e. those of the plain+// series and those of the voiced+// series. Thus, under MCA, Miller-Ockhuizen’s /t t | ! / are interpreted as /t t k| k! k k/ (i.e. the stop clusters of the plain+// series), and her /d dz d |!
/ are interpreted as /d dz d | ! / (i.e. the stop clusters of the voiced+// series). Note here that the “Guttural Co-occurrence Constraint” involved in these two series can adequately be stated under the MCA interpretation: the constraint
276 that a guttural vowel does not follow a consonant of these series of consonant clusters is accounted for by the cluster offset // which is a guttural consonant under the MCA interpretation.
“Epiglottalized consonants” in her interpretation include two non-click consonants /t k/ and eight click consonants /| ! | ! /, as shown in Table 5.23. These sounds have traditionally been described as affricated ejectives. For example,
Dickens (1994: 12) transcribed them as [tkx’ kx’] and [|kx’ !kx’ kx’ kx’ |kx’ !kx’
kx’ kx’], respectively, and Ladefoged and Maddieson (1996: 275-278) described the click accompaniments of the eight Miller-Ockhuizen’s “epiglottalized” clicks (i.e. /|
! / and /| ! /) as involving affricated ejectives, transcribing them as [k|x’ k!x’ kx’ kx’] and [|x’ !x’ x’ x’]. In fact, even in her own earlier work (Miller-
Ockhuizen 2000: 324), the four “epiglottalized” clicks /| ! / were transcribed as
[|kx’ !kx’ kx’ kx’], and described as “clicks with uvular affricated ejection” (note here that she transcribed “uvular affricated ejection” by using [kx’], which should be [q’] in the IPA). As regards their place of articulation, she interpreted the friction of the ejection as uvular on the grounds that it has higher F1 than voiceless unaspirated clicks (Miller-Ockhuizen 2003: 18). Miller-Ockhuizen (2000: 313-314) phonetically transcribed /t/ and /k/ as [tkx’] and [kx’] (i.e. [tq’] and [q’] in the IPA), respectively.
Abandoning the previous view, Miller-Ockhuizen (2003) claims that epiglottalized clicks involve “... no constriction at the glottis, and they are not ejectives” (p.18), on the grounds that in preliminary fiberscopic experiments with her own pronunciation, “epiglottalized clicks are produced with a constriction in the pharyngeal region ..., formed with pharyngeal narrowing and false vocal fold contraction” (p.55). In order to confirm this claim, however, the articulation of these sounds produced by native speakers must be investigated, as she herself states (ibid.: 55). In addition, she argues that her interpretation of /| ! / as voiced epiglottalized clicks “... allows us to
277 maintain the very strong universal that there are no voiced ejectives found cross- linguistically” (ibid.: 18-19). However, as seen in Table 5.23, her interpretation includes two non-click voiced ejective consonants /ds’ d’/, which are obvious counter- evidence against the same universal.
MCA, assuming that the series of her “epiglottalized” consonants involve the affricated ejective, interprets her /| ! / as the consonant clusters /tq’ k|q’ k!q’ kq’ kq’/ in which the cluster onset is /t k| k! k k/ and the cluster offset is /q’/ (i.e. the stop clusters of the plain+/q’/ series), and interprets her /| ! / as /|q’
!q’ q’ q’/ in which the cluster onset is /| ! / and the cluster offset is
/q’/ (i.e. the stop clusters of the voiced+/q’/ series), by regarding her /k/ as /q’/.
Like the plain+// series and the voiced+// series (i.e. /t t k| k! k k/ and /d dz d | ! /), the plain+/q’/ series and the voiced+/q’/ series involve the “Guttural Co-occurrence Constraint”. This can also adequately be stated under the MCA interpretation: the constraint that a guttural vowel does not follow these series of consonant clusters is accounted for by the cluster offset /q’/, which is a guttural consonant under the MCA interpretation.
5.3.3.2 The MCA interpretation of other Ju|’hoansi consonants The MCA interpretation of the Ju|’hoansi consonants is presented in Table 5.24. In addition to the “uvularized” and “epiglottalized” consonants, Miller-Ockhuizen’s “voiceless nasal aspirated” and “voiced nasal aspirated” clicks are interpreted as belonging to the stop clusters in Table 5.24. Thus, her /| ! / and /| !
/ shown in Table 5.23 are interpreted under MCA as /k|h k!h kh kh/ (i.e. the plain+/h/ series) and /|h !h h h/ (i.e. the voiced +/h/ series) in Table 5.24, respectively. In these clusters, the cluster offset is independent /h/, which is transcribed as // in Table 5.22. The stop clusters of these two series also involve the “Guttural Co-occurrence Constraint”, not followed by a “guttural” vowel. Again, this
278 is accounted for by their cluster offset /h/ that is a guttural consonant under the MCA interpretation. The nasality on the surface forms of these series, which is reflected in
“” and “” before a click symbol in Miller-Ockhuizen’s transcription of / | !
/ and /| ! /, is interpreted as the same nasal venting that is observed in
/k|h k!h kh kh/ in G|ui (see Section 3.3.3.3 in Chapter 3). In Ju|’hoansi, the nasal venting is interpreted as also occurring in the voiced series. Under the MCA interpretation, the narrow phonetic transcription of /k|h k!h kh kh/ and /|h !h h
h/ should be [k|h k!h k h k h] and [|h !h h h], respectively, in which the tilde marks the phonetic detail of the nasal venting.
Table 5.24 The Classification of Ju|’hoansi consonants by MCA. Lb stands for labial, Dt for dental, Al for alveolar, Pl for palatal, Vl for velar, Uv for uvular, Gl for glottal, - Af for affricated, and -Lt for lateral Lb Al Al-Af Pl Dt Pl Al Al-Lt Vl Uv Uv-Af Gl Fricatives Plain s h Voiced z Simple stops Plain p t ts t k| k k! k k Voiced b d d | ! Complex stops Voiceless ejective ts’ t’ k|’ k’ k!’ k’ k’ q’ Voiced ejective dz’ d’ Voiceless aspirated p t ts t k| k k! k k Voiced aspirated b d dz dts | ! Stop clusters Plain + // t ts t k| k k! k Voiced + // d dz d | ! Plain + /q’/ tq’ k|q’ kq’ k!q’ kq’ Voiced + /q’/ |q’ q’ !q’ q’ Plain + /h/ k|h kh k!h kh Voiced + /h/ |h h !h h Simple nasals Plain m n | !
279 Miller-Ockhuizen’s “oral voiceless aspirated” class /p t ts t k | ! / and “oral voiced aspirated” class /b d ds d | ! / shown in Table 5.23 are interpreted under MCA as /p t ts t k | ! / (i.e. the voiceless aspirated series) and /b d dz d | ! / (i.e. the voiced aspirated series) in Table
5.24, respectively. In addition, her “voiceless glottalized/ejected” class /ts’ t’ | !
/ and “voiced glottalized/ejected” class /ds’ d’/ in Table 5.23 are interpreted as /ts’ t’ k|’ k!’ k’ k’/ (i.e. the voiceless ejective series) and /dz’ d’/ (i.e. the voiced ejective series) under MCA. Note that the voiceless ejective series shown in Table 5.24 also includes /q’/, which Miller-Ockhuizen interprets as /k/. It should be pointed out that it is a residual question why the clicks of the voiceless ejective series (i.e. /k|’ k!’ k’ k
’/) are phonetically non-ejectives, while the non-clicks of the same series are all phonetically ejectives. I cannot answer this question at this stage.
Here, I should additionally comment on minor transcriptional differences between Table 5.24 and Tables 5.22 and 5.23. First, the voicelessness of the non-nasal clicks is transcribed with “k” before a click symbol in Table 5.24, while it is transcribed with a single click symbol in Tables 5.22 and 5.23. Second, in Table 5.24, the aspirated consonants are consistently transcribed with the superscript “h”, whether they are voiceless or voiced, while in Table 5.23, they are transcribed with the superscript “” when they are voiced. Finally, the voiced affricates /ds d ds’ d’/ in Table 5.23 are transcribed as /dz d dz’ d’/ in Table 5.24.
5.3.3.3 “Guttural Co-occurrence Constraint” under the MCA interpretation The MCA interpretation of the Ju|’hoansi consonants discussed in the preceding two sections implies a revision of the statement of the so-called “Guttural Co-occurrence Constraint”. As summarized in Section 5.3.3, this constraint groups a class of consonants referred to as “guttural consonants”, namely, //, //, and all consonants
280 presented in Table 5.23 under Miller-Ockhuizen’s (2003) interpretation. Under the MCA interpretation, in contrast, the class of “guttural consonants” is defined as only (i) two plain fricatives // and /h/ and (ii) all complex stops shown in Table 5.24. As already mentioned, in addition to (i) and (ii), the stop clusters of all six series (i.e.
Plain+//, Voiced+//, Plain+/q’/, Voiced+/q’/, Plain+/h/ and Voiced+/h/ in Table 5.24) all behave as “guttural consonants” in terms of the “Guttural Co-occurrence Constraint”. This is accounted for by their cluster offset that is one of the three guttural consonants / q’ h/, under the MCA interpretation. In summary, the MCA interpretation can also adequately state the “Guttural Co-occurrence Constraint” of Ju|’hoansi with a revision of the definition of the guttural consonants.
5.3.3.4 Advantages of the MCA interpretation of Ju|’hoansi consonants To conclude the discussion concerning the application of MCA to Ju|’hoansi consonants, I point out the advantages of the MCA interpretation over Miller-Ockhuizen’s (2003) unit analysis (henceforth MUA).
First, the MCA interpretation can reduce the consonant inventory size. Under the MUA interpretation, the Ju|’hoansi consonant inventory is an example of typologically peculiar large size with 88 consonants, i.e. 31 consonants in Table 5.22 and 57 consonants in Table 5.23. Under the MCA interpretation, these consonants are re- analyzed to 57 single consonant phonemes and 31 consonant clusters, as shown in Table 5.24.
Second, the MCA interpretation can capture the identity of the uvular and glottal accompaniments (i.e. accompaniments for /k! ! k!q’ !q’/ and /k!h !h/ in my transcription, corresponding to those for /! ! ! !/ and /! !/ in her transcription) with the independent uvular and glottal consonants (i.e. / q’/ and /h/ in my transcription, corresponding to / k/ and // in her transcription). Under the MCA interpretation, the co-existence of the accompaniments and the equivalent independent consonants can be transparently expressed as the identity of the independent / q’ h/
281 and the corresponding cluster offsets, as reflected in the transcriptions. In contrast, under the MUA interpretation, it is opaque.
Finally, MCA has an advantage in a cross-Khoisan comparison: it facilitates identifying recurrent consonantal types and series of Khoisan languages, and revealing aspects of common Khoisan constraints of cluster onset and offset, as discussed in Section 5.3.2. Miller-Ockhuizen (2003) does not discuss applications of her unit analytic classificatory framework with “uvularization” and “epiglottalization” to consonant systems of other Khoisan languages. It is questionable how her framework can possibly be applicable to the consonant systems of G|ui and !Xóõ that include more than two series of click clusters with uvular offsets, such as /k!q k! k!q k!q’ k! k!q’/. The distinctions among these six series cannot be adequately described by using only “uvularization” and “epiglottalization”. It is not clear how her framework can be extended for a cross-Khoisan comparison. Therefore, at the present stage, there is no reason to argue in favor of the MUA interpretation over the MCA interpretation from a cross-Khoisan comparative perspective.
I have discussed the advantages of the MCA interpretation of the Ju|’hoansi consonants. At this stage, it is still difficult to finally decide which interpretation is more adequate for describing Ju|’hoansi consonants, the MCA interpretation or the MUA interpretation, because there are questions still to be answered in future research, such as questions concerning phonetic evidence for “epiglottalization” or the cross-Khoisan applicability of the MUA interpretation. However, it should be emphasized that the MCA interpretation can be applicable to Ju|’hoansi consonants, adequately stating important constraints, reducing the typologically peculiar consonant inventory size, capturing the identity of independent / q’ h/ and the equivalent cluster offsets, and facilitating a cross-Khoisan comparison of the consonant systems.
282 5.4 Integration of the clicks and the non-clicks As pointed out by Traill (1997: 103), “...existing analyses of clicks and non-clicks are seldom integrated into a single coherent phonological system...”, providing two separate consonantal inventories. Concluding this chapter, based on the G|ui findings, I discuss the issue of how the clicks and the non-clicks should be cross-classified adequately in terms of the same set of features, in other words, how the two sub-classes of consonants should be integrated.
I have already discussed in detail one important dimension of the cross-classification of the clicks and the non-clicks, namely the classification in terms of the type and series of consonant, which is related to the vertical axes of Tables 3.1, 3.4 and 3.15 in Chapter 3. The discussion below focuses on the other dimension of cross-classification, which is related to the horizontal axes of the tables.
The following three subsections recapitulate the phonological features involved in this dimension, which have so far been described and discussed in this study. They include (i) place-of-articulation (abbreviated to POA) features, (ii) two manner-of-articulation (abbreviated to MOA) features, i.e. [affricated] and [lateral], and (iii) one acoustic feature, i.e. [grave].
5.4.1 Place-of-articulation features Let me first summarize important POA features involved in a cross-classification of the clicks and the non-clicks of G|ui. First, the feature [coronal] is necessary for grouping the three plain non-click stops /t ts c/ and the plain clicks /k| k! k k/ into a phonological class that is required to state the phonological constraint on the cluster onset. Second, the coronal stops are sub-classified in terms of the features [apical] and
[palatal]. The feature [+apical] groups the clicks with the [! ] influxes, i.e. /k! k !
k! k k!’ k’/, into a phonological class, which is necessary for stating the constraints on the vowel discussed in Chapter 4. The feature [palatal] is necessary to distinguish the [+palatal] subclass, i.e. /c c c’ k k k’/, from the [–palatal]
283 subclass, i.e. /t ts d dz t ts t’ ts’ k| k! k | ! k| k! k k|’ k!’ k’/. The feature of [coronal] and the two features for its sub-classification, [+/–apical] and [+/–palatal], all involve cross-classifications of the clicks and non-clicks in terms of place of articulation.
Note here that interpreting the place of articulation of the clicks as [coronal] implies that the anterior closure of the clicks is primary, and that the posterior closure is phonologically less important if not irrelevant. This view is supported by the observation that the clicks phonologically pattern with the non-click coronal stops in terms of the constraint on the cluster onset in G|ui.
Notice, however, that this view is based on MCA, which assumes that the posterior closure does not contrast in terms of POA features. If we assume that the posterior closure can phonologically contrast, it must be re-considered. Let us recall the question of whether clicks of the plain+/q/ series under MCA (/k|q k!q kq kq/) should not be re-interpreted as a non-cluster series of /q| q! q q/ (see Section 3.3.3.3). This alternative interpretation assumes phonological contrasts in terms of POA features not only for the anterior closure release (i.e. influxes), but also for the posterior closure (i.e. [velar] vs. [uvular]). This interpretation requires more unit (i.e. non-cluster) clicks than MCA, and complicates the phonological representation of unit clicks in terms of the POA features, but it can directly express the place of the posterior closure in the phase of influx release which it is presumably uvular also in G|ui. The adequacy of this alternative interpretation must be pursued in future research.
5.4.2 Manner-of-articulation features The MOA feature [affricated] also involves a cross-classification between the clicks and the non-clicks. A typical [+/– affricated] distinction is found between /ts dz ts ts’ q’/ and /t d t t’ q’/, and the same feature classifies the [+affricated] clicks /k| k | k| k k|’ k!’ k’ k’/ and the [–affricated] clicks /k! k ! k! k k!’ k’/.
284 The phonological distinction in terms of another manner feature, i.e. [+/– lateral], is only found between /k! ! k! k!’/ and /k k k’/. However, this feature is important for the interaction between the click and non-click, as described in Section 3.2.5.1 in Chapter 3 (recall that the liquid /r/ in the root-medial position assimilates to the lateral click in the root-initial position, cf. /kara/ [kala] “Acacia erioloba E. Mey” vs. /k!ara /
[k!aa] “Ochna pulchra Hook”). In this sense, the feature [lateral] is relevant not only for the clicks but also for the non-clicks.
The two MOA features, [affricated] and [lateral], are therefore both important for the integration of the clicks and non-clicks.
5.4.3 The acoustic feature [grave] In addition to the POA and MOA features, the acoustic feature [grave] must be introduced for the integration of the clicks and the non-clicks. As already described in
Chapter 3, this acoustic feature phonetically distinguishes the apical influxes [! ]
(i.e.[+grave]) from the laminal influxes [| ] (i.e. [–grave]). Traill (1995, 1997) argues for the phonological importance of this acoustic feature, demonstrating that this feature facilitates naturally stateing two processes, i.e. click replacement attested in some Khoe languages, and vowel assimilation attested in !Xóõ.
Concerning the former, there are two types of click replacement regularly attested in some Khoe languages, i.e. the “!k” type attested in G||ana (see Section 3.4.1 in
Chapter 3), and the “c” type attested in some Eastern Khoe languages (Traill 1980, Traill and Vossen 1997). In the former type, the alveolar clicks change to velar non- clicks, while in the latter type, the palatal clicks change to palatal non-clicks. If we assume that the alveolar clicks and the velar non-clicks belong to the [+grave] class, and the palatal clicks and the palatal non-clicks belong to the [–grave] class, then both types of click replacement can be expressed by using [grave] as below (Traill 1997: 107).
[+click, α grave] [–click, α grave]
285 As he correctly states, “... in click replacement ... clicks turn into cognate non-clicks”, and the phonetic “... basis of the cognation” can be accounted for in terms of the acoustic feature [grave], not in terms of articulatory features.
Regarding vowel assimilation attested in !Xóõ, I discuss the feature [grave] in relation to its specification for the labial click in the next section.
5.4.4 The feature distinguishing the clicks from the non-clicks The features presented in the previous three subsections are necessary but not sufficient for cross-classifying the clicks and the non-clicks. Notice that some contrasts between the clicks and the non-clicks cannot be captured only in terms of these features. Let us, for example, consider what feature involves the distinction between /ts/ and /k|/, or the distinction between /c/ and /k/. As shown in Table 5.25, in order to express the distinction between /ts c/ and /k| k/, another feature must be added to the six phonological features mentioned above. This additional feature, tentatively labeled [“click”] in the table, corresponds to Chomsky and Halle’s (1968: 309) feature [suction] that signifies the velaric ingressive airstream mechanism involved in the clicks.
Table 5.25 Contrasts between /ts c/ and /k| k/ ts k| c k coronal + + + + apical – – – – palatal – – + + affricated + + – – lateral – – – – grave – – – – “click” – + – +
Traill (1985: 206) adopted the SPE feature [suction] in order to distinguish the clicks ([+suction]) from the non-clicks ([–suction]) in !Xóõ. The feature referring to the velaric ingressive airstream mechanism is also used to specify the clicks in other more recent studies, such as Güldemann’s (ibid.) “ingressive” as opposed to “eggresive”,
286 Miller-Ockhuizen’s (2003) “velaric” as opposed to “pulmonic”, and Ladefoged’s (1995) “Velaric” with “[+click]” as opposed to “Pulmonic”.
In his later work, Traill (1997) proposed interpreting clicks as “enhanced versions” (p. 115) of non-click stops, demonstrating that clicks are perceptually salient. He stated that clicks “... exploit all the features of non-click stops but utilise a novel source for the production of these features, namely the noisebursts generated by the velaric suction. (ibid.)” This interpretation has more explanatory power than the SPE-type feature for clicks, because it not only refers to the articulatory property unique to clicks (i.e. velaric ingressive airstream mechanism), but also expresses the perceptual effect of this property. If we adopt this interpretation, the [+/– “click”] distinction between /k| k/ and /ts c/ shown in Table 5.25 can be translated in terms of the enhancement of the same features in /k| k/. For example, /k/ and /c/ share the features, but /k/ is distinct from /c/ in that it involves velaric suction enhancing the same features. Henceforth, I express the feature of enhancement by [+/–enhanced], assuming it is a binary feature distinguishing the two consonant classes.
If we adopt [+/–enhanced], the features for the click vs. non-click integration in G|ui are summarized as in Table 5.26.
Table 5.26 Important features for integration of clicks and non-clicks of G|ui, exemplifying with representative stops t ts c k q q’ k| k k! k coronal + + + – – – + + + + apical – – – – – + + palatal – – + – + – – affricated – + – – – + + – – + lateral – – – – – – – – – + grave – – – + + + – – + + enhanced – – – – – – + + + +
In order to adapt this cross-classification to a wider range of Khoisan consonant systems, we must extend it to include two classes that G|ui does not contain, namely the
287 labial clicks, e.g. /k/, attested in South Khoisan languages, such as !Xóõ, and the affricated palatal non-click stops, e.g. /t/, attested in Ju|’hoansi.
Regarding the labial click, its POA and MOA features are obvious, namely [labial] and [+affricated, –lateral]. If we examine the labial click in terms of the feature [grave], we can find phonological evidence for specifying it as [+grave] in !Xóõ. According to
Traill (1995: 123-125), the labial click [] behaves in the same way as [! ] in terms of vowel assimilation in !Xóõ: the assimilation affects the non-pharyngealized low vowel, i.e. /a a a’/, which is immediately preceded by a click with the influx [| ] and immediately followed by /i/. The low vowel in this context assimilates completely to the following /i/, as exemplified below:
/k|ai / > [k|i] “aardwolf (Proteles cristatus)” /kai / > [ki] “steenbok (Raphicerus campestris)”
(In Traill’s (ibid.) transcription plain clicks “k| k” are “| ”.)
In contrast, the low vowel immediately preceded by a click with the influx [ ! ] does not completely assimilate to the following /i/, resulting in [ ’], as exemplified below:
/ka’i/ > [k’i] “aardwolf (Proteles cristatus)” /k!ai / > [k!i] “sp. of tree (Ziziphus mucronata Willd.)” /kai / > [ki] “old (Class 1)”
(In Traill’s (ibid.) transcription plain clicks “k k! k” are “ ! ”.)
He demonstrates that this assimilation rule is naturally stated by using the feature
[grave], classifying the vowel /i/ and the click consonants with [| ] as [–grave], the low vowel /a a a’/ and the clicks with [ ! ] as [+grave]. The rule is stated as follows:
V[+grave] V[–grave]/C[–grave]_V[–grave]
288 As regards the non-click labial consonants, Traill (1995) does not provide examples, and in his !Xóõ dictionary (Traill 1994), there are no words in which a non-click labial consonant is followed by underlying /a/ followed by /i/. However, the following two loan words found in the dictionary suggest the non-click labial consonants, at least /b/ and /p/, may not involve this assimilation, indicating that they may be [+grave].
bai bili “bible” (p.152) (not [bii bili])
phaipi “hosepipe” (p.153) (not [piipi])
In terms of the feature [enhanced], the labial click is different from the other clicks, and should be specified as [–enhanced]. According to Traill (1997), unlike the other clicks, the labial click lacks saliency in all perceptual tests that he conducted. He interpreted its lack of saliency as being due to the articulatory nature of the labial click, which is “...unsuitable for the conversion of aerodynamic energy into a salient acoustic signal. (ibid.: 115)”∗
Finally, I should comment on the feature specification of [affricated] for non-click palatal stops. Non-click palatal stops do not exhibit a [+/–affricated] distinction in any
Khoisan languages: a language like G|ui containing unaffricated palatal stops, such as /c / etc., does not have affricated palatal non-click stops, and a language like Ju|’hoansi containing affricated palatal stops, such as /t d/ etc., does not have unaffricated non- click palatal stops. Therefore, the positive/negative value of [affricated] for the non- click palatal stops varies according to the specific language, although it is not distinctive in Khoisan languages.
∗ The feature [enhanced] can capture the two perceptually different classes within the clicks. However, since the labial click phonologically patterns with the other clicks, i.e. occurring with cluster offset of uvular and glottal obstruents in !Xóõ (Traill 1985), there must be another feature stating the natural class of all the five clicks. This suggests that in order to describe a phonological system with five click influxes like !Xóõ system, we may need both [enhanced] and the SPE-type feature (e.g. [suction], [click] or [ingressive], etc.) This topic must be explored in future reseach.
289 5.4.5 Conclusion Table 5.27 summarizes the discussion of this section, presenting the set of features for the click and non-click integration.
This set of features includes the seven POA features shown in bold in the table, the two MOA features [affricated] and [lateral], the acoustic feature [grave], and the acoustic- perceptual feature [enhanced]. This set does not include three covering features for (i) the class consisting of [labial] and [coronal], (ii) the class consisting of [uvular] and [glottal], and (iii) the class consisting of apical clicks and uvular consonants. The first feature states a natural class of consonants for the cluster onset, and the second states another natural class for the cluster offset. Traill (1985) referred to the first covering feature as “anterior” with a different sense from the SPE [anterior] feature, and this term is adopted by Güldemann (2001). For the second covering feature, the term [guttural] would be most appropriate, although the same term is used by Miller-Ockhuizen (2003) in a different sense. The final covering feature is necessary for stating the Back Vowel Constraint in G|ui, as discussed in Section 4.6.2 in Chapter 4. A phonetic basis for this covering feature is unclear at this stage, and it is a topic for future research.
Table 5.27 The features for click and non-click integration
labial coronal velar uvular glottal POA [–apical] [+apical] [-palatal] [+palatal] [affricated] – + – + + +/– – – + – – + – [lateral] – – – – – – – – + – – – – [grave] – + – – – – – + + + + + – [enhanced] – – – – + – + + + – – – – Examples p k t ts k| t/c k k! k k q q’
The set of features presented in Table 5.27 is based on the MCA interpretation that I have discussed throughout this thesis. The proposal of this set of features is not intended for a final resolution of the issue of the click and non-click integration. It
290 should rather be understood as a starting point of a further exploration of the issue, which will be conducted based on other Khoisan languages.
291