THE ACQUISITION OF CLICKS BY NON-MOTHER-TONGUE SPEAKERS

It is generally accepted that non-mother-tongue speakers acquiring Xhosa often encounter difficulties with the acquisition of clicks. This paper addresses this issue by focussing on some of the results obtained from 18 Englishand Afrikaans-speaking children aged between 10;0 and 13;0, acquiring Xhosa as a third language in the primary school. The results firstly indicate that non-mother-tongue speakers find certain click types easier to acquire than other click types. Secondly, based on the manner in which clicks are acquired by non-mother tongue-speakers, this paper indicates numerous phonological processes that are operative in the acquisition of these sounds. Lastly, tentative suggestions are provided to ease the acquisition of clicks by non-mothertongue speakers.


INTRODUCTION
The clicks, once regarded as "clacks" by the early missionaries of the Cape in the 1820's (Shepherd, 1940: 57), are exceptional as well as unusual sounds in that they not only occur as interjections or non-linguistic gestures in some languages, but they also occur in the consonant inventory of a limited number of languages spoken in Africa (Roux, 1989b).Such sounds also occur in the Xhosa language!There is a total of fifteen clicks occuring in the Xhosa language2.Table 1 presents the clicks occuring in Xhosa as well as the phonetic transcription for each of these sounds, together with the test items in which these sounds are found, as used in this study.List of clicks occurring in the Xhosa language Apart from being extremely rare sounds, clicks are also generally regarded as being very difficult sounds to acquire, even by mother tongue speakers (Lewis, 1994).This claim is based on research which indicates that clicks are acquired relatively late by Xhosa-speaking children (see Lewis, 1994;Roux, 1989a), as well as being among the last sounds to be acquired by Zulu-speaking children (Lanham, 1969).
Although these sounds occur in numerous languages in South Africa, they do not occur in the consonant inventory of English or Afrikaans, two other languages spoken in South Africa3.One can imagine, based on the relative difficulty mother tongue speakers have in acquiring clicks, that these sounds would also be very difficult for non-mother-tongue speakers to acquire.Unfortunately, very little research, apart from Doke (1926), MeskEer =Eihmiger.44494,Lewis (1994) and possibly a few others, has been recorded or even undertaken to substantiate this point.For instance, very little has been documented about which click types are more easily acquired than others by non-mother-tongue speakers.Furthermore, very little has been documented on the phonological processes that are operative in the acquisition of these sounds by this specific group.This paper will focus on some of these issues.
The aim of this paper is to (1) indicate those click types that non-mother-tongue speakers acquire more easily than other click types; (2) indicate phonological processes that are applied by non-mother-tongue speakers to acquire clicks; and lastly (3) mention tentative suggestions that can lead to easier acquisition of clicks.
Although this paper will discuss primarily the acquisition of clicks by non-mother-tongue speakers, it will also compare these results to the acquisition of clicks by mother tongue speakers4.

Subjects
Eighteen children acquiring Xhosa as a third language in the primary school, six from each of the standards 3, 4 and 5, ranging between the ages 10;0 and 13;0, were identified to participate in the rescarch5.In each of the age groups there were three English-and three Afrikaans-speaking children.All eighteen subjects were regarded by the Xhosa teacher as "average achievers".None of the subjects had much or even any informal exposure to the Xhosa language outside the classroom.

Material
The fifteen clicks occuring in Xhosa were tested.Although articulation tests are in most cases elicited in the initial, medial and final positions in words (Grunwell, 1983), all the test sounds-(i.e.clicks) for this investigation were tested in the intervocalic positions, because Xhosa consonants rarely occur in initial or final positions.After consulting two Xhosa dictionaries, McLaren (1981) and Nabe, Dreyer & Kakana (1985), 17 words that contain the test sounds were chosen.The words containing the test sounds had to fulfil the following criteria: they had to be depicted by line drawings; and had to be regarded in general as "familiar" to the subject group.
In Table 1 a list of the test sounds as well as the words in which these test sounds are used can be found.Of the 17 pictures depicting these words, 15 represented objects named by the test item, while the other two pictures were pretest items6.

Elicitation
Three types of elicitation tests were applied.The three types were imitation, sentence completion and sentence recall (see Lewis, 1994: 12 -13 on the three methods of elicitation that were applied to test subjects' articulation of clicks).

Procedure
Each subject was tested individually.The recording equipment used in the investigation consisted of a SONY Cassette Corder TCM-2 and an Electret condensor microphone.After rapport was established, each subject was asked to look at the file containing pictures of the test items.Imitation, sentence completion and sentence recall were used with the two pretest pictures and then testing proceeded in the manner described above7.

Trascription of subjects' responses
All responses made by the subjects were transcribed by four independent transcribers.A broad transcription was made of all the responses.All responses were transcribed according to the International Phonetic Alphabet.Transcriptions of each subject's productions were accepted as final when at least two of the four transcribers agreed on them.

RESULTS AND DISCUSSION
1.
Ease of acquisition of clicks School teachers and linguists have often been asked by non-mother-tongue speakers, which click type is commonly regarded as being the easier or more difficult to acquire.In order to determine this, percentages of correct production for each click by each of the .nonmother-tonguespeaking subjects was calculated8.Subjects had to attempt a sound at least three times to have performace on that sound included in this analysis.66,6% was set as the criterion of correct production for a subject to have "acquired" each click.Table 2 presents the percentage of subject across all ages to have acquired each of the 15 clielcs in Xhosa.For example the voiceless dental click is 67%, whereas for the voiced alveolar lateral click it is 6%.The mean pooled values of the percentage of non-mother-tongue speakers to have produced each of the click types correctly The results in Figure 1 indicate that the dental click type (45,6%) tends to be correctly produced more frequently than the alveolar lateral click type (36,8%) and the alveopalatal click type (26,6%), by non-mother-tongue speakers.Therefore, based on these results, one can assume that the dental click type is the easiest of the three click types to be acquired by non-mother-tongue speakers, followed by the alveolar lateral click type and then the alveopalatal click type.These results, indicating that the alveopalatal click type is the more difficult click type to acquire, also corresponds with Doke's (1926: 129) analysis on the acquisition of Zulu clicks by non-mother-tongue speakers.He mention that the alveopalatal click -it is uncertain whether this refers to all the alveopalatal clicks, or only the voiceless alveopalatal click "generally proves to be the most difficult for a European to acquire".
Based on the results indicating that alveopalatal clicks are more difficult to acquire, it is no wonder that Xhosa speaking people often enjoy asking non-mother-tongue speakers, who are acquiring or studying Xhosa as a third language, to say as well-known Xhosa phrase in order to see if they have mastered the clicks.The phrase, which consists primarily of the alveopalatal click type, is as follows in (1). (1) "Iqaqa lizigikaqike kuqaqaqa laqhawuka ughoqhoqho" (The pole-cat rolled itself in the couchgrass and broke its trachea") 2.

Phonological processes operative in the acquisition of clicks by non-mothertongue speakers
There is to date very little research available to indicate the various phonological processes applied by non-mother-tongue speakers when acquiring clicks.Research by Lewis (1994) sheds some light on this issue.Numerous phonological processes are proposed by him to describe the phonological tendencies occuring among non-mother-tongue speakers.This paper will briefly focus on the application of phonological processes by non-mother-tongue speakers.It will also compare the percentage frequency of occurrence of such processes to that of the phonological processes that are likewise applied by mother tongue speakers.The criteria, both qualitative and quantitative, that were applied to validate the presence of such phonological processes can be found in Lewis (1994: 37-38).Table 3 provides a list of the major phonological processes and sub-processes that are involved in the acquisition of clicks by non-mother-tongue as well as mother tongue speakers.The percentage frequency of occurrence of each of the major processes is applied in this paper only to indicate those processes that are more favoured by the specific subject groups.More detail concerning the percentage frequency of occurrence of each of the phonological processes, as well as a more detailed account of each of the processes describing the phonological tendencies, occurs in Lewis (1994).

Processes
Mother  Major phonological processes and sub-processes involved in the acquisition of clicks by both non-mother-tongue and mother tongue speakers.
Figure 2 provides a graphic representation of each of the major phonological processes by non-mother-tongue speakers as well as mother tongue speakers.Graph representing frequency of occurrence of each of the major phonological processes The content of these processes will now be discussed.

Simplification processes
Non-mother-tongue speakers often showed the tendency to produce clicks as voiceless and aspirated alveolar and velar stops.Two types of simplification processes were identified, based on these tendencies.In the first of the two processes, i.e. alveolar simplification, subjects would simplify clicks to voiceless and aspirated alveolar stops.Examples illustrating this process occur in (2). (2) In velar simplification, subjects would simplify clicks to voiceless and aspirated velar stops.This process is indicated by the following tendiencies in (3). (3) In this phonological process subjects often applied the incorrect click type when producing the target sound.Three types of substitution processes were identified based on subjects' production tendencies.The first process, regarded as delateralisation, occurred when subjects produced dental or alveopalatal clicks in place of alveolar lateral clicks.The following tendencies in (4) illustrate this process. (4 A second substitution process, regarded as depalatalisation, occued when subjects produced alveopalatal clicks as dental or alveolar lateral clicks.In (5) the following tendencies occurred that illustrate this process. ( In dedentalisation subjects would produce dental clicks as alveopalatal or alveolar lateral clicks.The tendencies in (6) illustrate this process. (6) /iT wadi/ > /i i_ph/ > It is interesting to note in Table 3 and Figure 2 than non-mother-tongue speakers applied the above-mentioned substitition processes more frequently, 19% in total, in comparison to mother tongue speakers (8%).Results also indicate that non-mother-tongue speakers resort more to substitution processes (19%) than simplification processes (14%) when acquiring clicks.One can assume, based on these results in Table 3, that non-mother-tongue speakers are aware of the clicks and will attempt to produce the click even though it is not the correct click type.

Reduction processes
In this phonological process, subjects would often reduce an element of the click, be it the click itself or the accompaniment of the click.There was even the tendency by subjects to reduce one or more of the consonants in a cluster.It is based on these tendencies that three types of reduction processes are identified.In the data numerous tendencies indicated that a click was often reduced with only its accompaniment remaining.The tendencies in ( 7) provide examples of this process, regarded as primary element reduction. ( The second reduction process to be identified involves accompaniment reduction.In this process the click is reduced of its accompaniment(s) with only the click remaining.The following tendencies illustrating this process occur in (8).Lastly, in the process regarded as cluster reduction process speakers would reduce clusters of consonants by deleting one of the consonants in that cluster.The following tendencies in (9) illustrate this process.3), that a click is often produced with a nasal element attached to it.Based on the tendencies of non-mother-tongue speakers to nasalize clicks, three nasalisation processes are identified.
The first process, regarded as nasalisation, is a tendency among subjects, as indicated in (10), to nasalize non-nasal clicks.
(10) /ilephi..1 A second nasalisation process, regarded as prenasalisation, is a tendency by subjects to prenasalize clicks as well as alveolar and velar stops.The following tendencies in (11) illustrate this process. ( In postnasalisation, two types of nasal release processes occured, i.e. a normal postnasalisation, as indicated in ( 12).

3.
Tentative sugestions that can ease the acquisition of clicks As it has been generally accepted, and indicated in this paper, that clicks are rather difficult sounds to acquire, as well as being generally feared by non-mother-tongue speakers, this paper provides tentative suggestions that can possibly ease the acquisition of these sounds.These suggestions are based on tendencies that were identified by Lewis (1994) in both nonmother-tongue and mother tongue speakers' acquisition of clicks.Two approaches are mentioned.
The first approach that one can use to possibly ease the acquisition of clicks, is to apply similar stages of acquisition to those that mother tongue speakers apply when they acquire clicks.Lewis (1994), based on the data in his investigation, indicates that four stages of click acquisition occur.The four stages of acquisition are as follows: avoidance of words containing clicks; avoidance of words as well as the production of clicks as alveolar and velar stops; occurrence of click production, even though they do not reach the criteria of Therefore, by applying such stages in acquiring the clicks, non-mother-tongue speakers could gradually acquire clicks in a similar manner to that in which mother tongue speakers acquire clicks.
A second approach that could possibly ease the acquisition of clicks is based on the tendency by non-mother-tongue speakers to nasalize click types.In Table 3 and Figure 2, results indicate that non-mother-tongue speakers (16%) nasalize clicks more than mother tongue speakers (3%) do.Therefore, based on this specific occurrence mentioned above, non-mother-tongue speakers who are acquiring Xhosa and find difficulty with the production of clicks could apply a strategy of simply nasalising all clicks.Once a person can produce a click type that is assisted by the nasal element, the nasal element could be deleted leaving only the click element behind.

CONCLUSION
This paper has focused on three issues which generally speaking have not been formally addressed before in research on the acquisition of clicks by non-mother-tongue speakers.
The results indicate that dental click types show the tendency to be acquired more easily than alveopalatal click types.Secondly, the results also indicate that there are four major phonological processes operative in the acquisition of clicks by this subject group.Lastly in view of the degree of difficulty with which these sounds are acquired, two suggestions are provided that may at least make the acquisition of clicks more "user friendly" for non- monther-tongue speakers.

FOOTNOTES t
Xhosa is a Bantu language which is spoken in South Africa and belongs to the Nguni group of Bantu languages of Southern Africa.The language is predominantly the vernacular of the Transkei and Ciskei as well as of a large portion of the Xhosaspeaking people living in the eastern Cape and large industrial areas of South Africa.
2 Xhosa has three click types, i.e. the dental click, the alveo-palatal click and the alveolar lateral click.Each of the three click types are able to occur with one of five different accompaniments, these being either voiceless, nasalized, aspirated, voiced or voiced nasalized (for further details on the click types in Xhosa see Lewis, 1994).
3 Roux (1989b) states that the main language groups of Southern Africa that utilize clicks are Research on the manner in which mother tongue Xhosa-speaking children and aphasics acquire and lose clicks can be found in Lewis (1994).The age of the Xhosa-speaking children participating in the investigation ranged from 1;6 5;5.

5
The school that the eighteen pupils attended and where research was undertaken in 1992 is the Somer3et West Primary School in Somerset West.The researcher is grateful to the headmaster, staff and pupils of the school for their assistance and cooperation during the course of this investigation. 6 The pictures depicting the words can be found in Lewis (1994). 7 All recordings of the subject groups' production of clicks are obtainable at the Phonetics Laboratory of the Department of African Languages at the University of Stellenbosch. 8 Figure 1The mean pooled values of the percentage of non-mother-tongue speakers to have produced each of the click types correctly Figure 2 observed, especially among non-mother-tongue speakers (16%) (see Table the percentage frequency of occurence of the nasalisation processes in Table In East Africa the languages that ulitlize clicks are Sandawe, Hadza and Dahalo.