AS VARIÁVEIS INTERVENIENTES NA PRODUÇÃO DO ONSET COMPLEXO MEDIANTE UMA ANÁLISE SILÁBICA The intervening variables in the production of consonant clusters by syllabic analysis

Purpose : to verify and to compare the repair strategies and the influence of linguistic (syllabic and prosodic) and extralinguistic variables in the production of consonant clusters by children with typical and atypical phonological development. Method : it was analyzed the speech of 48 children, 24 with typical phonological development and 24 with atypical phonological development, similar in relation to sex, and age between 2:6 to 5:5;29 (typical group) and 5:0 to 7:11;29 (atypical group). The samples were collected transversely, based on the instrument Avaliação Fonológica da Criança. It was analyzed the words presented as target consonant clusters, with a corpus of 278 words of typical development and 460 of atypical development. The correct production, C 2 deletion, C 1 deletion, syllable deletion, epenthetic, metathesis and idiosyncrasies were considered as variants of the dependent variable. The extralinguistic factors such as age, sex and development type, and linguistic variables the number of syllables, next precedent syllabic context, the position in the word, the complexity of the onset in the syllable and metrical foot were considered as independent intervening variables. The speech data were statistically analyzed through VARBRUL. Results : the statistical program selected as significant for the correct production and other types of repair strategies in the consonant clusters the variables sex, age, development type, position in the word, metrical foot and precedent syllabic context. Conclusion : it was found that the linguistic and extralinguistic variables significantly influence the production of consonant clusters in children with both developments. The most widely used repair strategy was the deletion of C 2 .


INTRODUCTION
The phonological acquisition, along with the other modules of language development occurs throughout childhood, with the overall maturity of the child, including acquisitions and simultaneous developments 1,2 .
In most children, the domain of the standard phonological system is reached spontaneously in a continuously and gradually way from four or five years old.Although this domain occurs in a very similar way among children, it is found that acquisition is nonlinear and there are individual variations -such as the age of acquisition and the repair strategies used 2,3 .children do not know or whose production does not dominate.In the case of CO as aforesaid, simplifying the feature is most often adopted.Besides this, it is mentioned as an example of repair strategies on target CO syllable omissions (e.g.plástico -[´tSiku]), epentheses (e.g.bruxa -[bu'4uSa]), coalescence/ fusion (e.g.cravo -[´davu]), compensatory lengthening (e.g.planta -[´pã: n ta]) and metatheses (e.g.prato -['pa4tu]) 3 .
Studies have shown that extralinguistic and linguistic variables can influence the segmental phonological acquisition.Linguistic variables constitute up to factors such as previous and following context, tone and number of syllables.Regarding the extralinguistic variables, research on the subject refers to factors such as age or gender, verifying that these may or may not influence the phonological domain of BP 12,13 .
There were found studies in the literature mentioning the influence in the field of segmental phonology, however, there are few studies that verify the influence of syllabic or prosodic variables in the acquisition of a particular template (template syllabic), independent segments that occupy it 10,14 .
Therefore, this study aims to check and compare the repair strategies and the influence of linguistic variables (syllabic and prosodic) and extralinguistic in the production of the syllable with CO in children with typical and atypical phonological development.

METHOD
This research presents a transversal, exploratory and quantitative character.The sample was comprised of data from 48 children, divided into two groups: typical and atypical acquisition, each group consisting of 12 girls and 12 boys, monolingual speakers of BP.
To make up the database used in this study, after parental consent by signing the free and informed consent and oral assent of the children, these were submitted to speech screening, receiving a diagnosis of phonological disorder or typical phonological acquisition.
To be part of the sample, participants should not be receiving or have undergone any type of speech therapy and should not present hearing loss and changes evident in the neurological, cognitive and psychological aspects.
Speech samples were collected based on the instrument "Phonological Assessment of Child -Avaliação Fonológica da Criança: AFC" 15 , in which it uses the spontaneous nomination of 125 words represented through five themed designs, or toys that reproduce the list of AFC (in the case of the younger children).A recording of the speech sample The domain of the phonological system of a language is when a child can establish it in a consistent way with the target-adult of the community in which the child is.The maturation of phonological knowledge results in the establishment of a system consistent with this input 3,4 .
When there is difficulty in the eligibility of the received input, occurs what is called phonological disorder, i.e., it is difficult to properly use the language sounds of contrastive form 5 .
The main difference between the typical and atypical acquisition, or normal deviant is in the timing and time of development.In typical acquisition there are variations that occur within a pattern of rapid and dynamic change to the phonological system to stabilize.In the atypical acquisition, however, its changes remain the same for a longer period of time, causing a delay in the stability of the phonological system which consequently prevents the production of correct speech 6 .
The Brazilian Portuguese phonological system (BP) consists of 19 consonants (/p, b, t, d, k, g, s, z, S, Z, m, n, J, l, L, r, R/) and 7 vowels (/a, e, E, i, o, O, u/)), and when combined they form syllables and words.The syllable may be comprised of three components: onset, nucleus, and coda, the core being a mandatory element.The possible syllabic structures in Portuguese are V, CV, CCV, CVC, CCVC, VCC, VGC, CVCC, CVGC, and the maximum possibility is CCVCC 6,7 .
The CV structure is the first to be acquired during the process of phonological acquisition.Studies 4,8 refer to the following order of acquisition syllable: CV, V> CVV> CVC> CCV.
The complex onset (CO) has the possibility of 8 obstruents consonants (/p, b, t, d, k, g, f, v/) occupy the position C 1 and two liquid consonants (/l, r/) the C 2 position.The possibilities of initial onset with liquid /l/ can be /pl, bl, gl, fl, kl/, and the liquid /r/ can be /pr, br, tr, dr, kr, gr, fr/.In medial onset, with /l/ the possibilities are /pl, bl, tl, kl, gl, fl/, and with /r/ is /pr, br, tr, dr, kr, gr, fr, vr/.This structure, for the highest degree of complexity, is the last to be acquired in the BP, but on the acquisition way there is no intermediate stages, i.e., the child passes through the simplification of OC to the correct prodution 6,9,10 .The CO domain occurs around the age of five, independent from the segments that compose it.Furthermore, in the presence of phonological structure it appears as the most affected structure 6,9 .
In the process of stabilization of the phonological system, the child makes use of repair strategies in an attempt to get closer to the sound or syllable structure target, being an occurrence in normal or deviant children 3,11 .These strategies are used instead of the production segment and/or syllable structure that locked with complex coda and simple onset (e.g.monstro), syllable locked with complex coda and without onset (e.g.instrumento), syllable locked and CO (e.g.prostrado), syllables without onset (e.g.atriz) and syllables without onset and simple coda (e.g.astronauta).The variable following syllabic context was analyzed for variations zero/null (e.g.livro), open syllable and simple onset (e.g.trigo), syllable locked with simple coda and simple onset (e.g.pretos), syllables without onset (e.g.triagem), syllables without onset and simple coda (e.g.triângulo), syllable locked with simple coda and CO (e.g.imprópria).
In the condition of position in the word, the words are categorized as at the initial position (e.g.preto), medial (e.g.cabrito) and final (e.g., compra).The complexity of the onset in the very syllable had its classification as a complex onset with no coda (e.g.prato), CO with simple coda (e.g.branco), CO with locked coda (e.g.transporte).
For the position as the foot metric of the accent it was assigned variants such as extrametric syllable (e.g.(pál.pe).bra), head of the metrical foot (e.g.(fral.da), the weak part of the metrical foot (e.g.: (li.vro), outside the metrical foot (e.g., trans.(por.te)).It was also considered the extralinguistic variable type of development with typical and atypical variations.
Speech samples which were collected and analyzed in this study are part of the database of two research projects of the institution, both approved by the Ethics Committee in Research of the institution under numbers 064/2004 and 046/02, respectively.
The statistical program used was Computational Package VARBRUL 17 for Windows known as Varbwin 18 .This program has been widely used, with data from language acquisition because it provides frequencies and probabilities of the studied phenomena, in addition to select the relevant variables in the process of phonological acquisition 3 .The program makes the probabilistic analysis in binary form, assigning relative weights to variants of the independent variables, to both variations (producing correct and incorrect) the linguistic phenomenon in question, represented by the dependent variable.We emphasize that the Varbwin assigns values of significance to linguistic and extralinguistic variables through the interaction between them (e.g.gender versus age; tonicity versus number of syllables).Thus, this software does not assign p-value to variants contained within a variable.For example, the Varbwin does not generate a significant value in the comparison between male and female.For these variants, it was assigned relative weights.The relative weights or probabilities of occurrence of the studied phenomenon (CO acquisition) will be removed from the statistical interaction that will from each child was transcribed by the researcher through the restrict phonetic transcription, and then reviewed by two judges, scholars of the last year of Speech Language Pathology.
For this study, only words that had targeted CO were gathered, being as criterion, a correct performance of CO by two children, of two consecutive age groups.The criterion to consider the acquired phoneme was 80% correct performance of CO 16 .In the group of children with typical development, the corpus consisted of 278 words, and in the group of children with atypical development the corpus was composed of 460 words, totaling 738 words.
After gathering the CO words, the words were coded according to their production.This coding was done by typing in the form of Microsoft Access program and served as input for the statistical program.It was considered intervening variables for the realization of the statistical analysis both the linguistic variables: syllabic repair strategies (dependent variable), tonicity, number of syllables, syllabic contexts preceding and following, position in the word, complexity of the syllable itself and position on the metrical foot accent.The extralinguistic variables considered were gender, age and type of development.
For the analysis of variable age of both groups it was considered six age groups.The typical acquisition group comprises children aged 2:6 to 5:5, 29, and atypical acquisition comprises children aged 5:0 to 7:11; 29.In both groups there is the counting of age groups every 6 months, two boys and two girls in each age group.As to the variable gender, the analysis included 12 girls and 12 boys in each group, as previously mentioned.
The variable precedent syllabic context had as variants zero/null (e.g.: broto), open syllable and simple onset (e.g.: livro), syllable locked with simple coda and simple onset (e.g.compra), syllable of correct production in the age group 7:0 -7:5 29.As for the variable type of development, the results showed a higher probability of correct production in typical development.In regards to gender, it was found that the female variant has a greater likelihood of correct production (Figure 1).
In the statistical analysis involving typical and atypical data isolated, it turns out that both for the normal development and for the atypical development, the age and gender extralinguistic variables proved to be significant.In relation to gender in typical development, it was found that girls are more likely to correct production.In the atypical development, the more likely correct production variation has occurred in males.
contain, together, all variables selected as significant by the program.Thus, probabilistic values between .50 and .59are considered neutral, the acquisition of the phonemes in the CO.Values which were equal or higher than .60 will be considered favorable and lower values below .50 are considered unfavorable to the studied phenomenon.

RESULTS
The statistical program, analyzing typical and atypical data together, revealed as statistically significant variables for the correct production of the CO, the variables age, gender and type of development.When analyzing age, there was a higher probability Legend: Statistical program: VARBRUL; Significance: 5% (p <0.05)

Figure 1 -Relevant variables to the correct production of the complex onset in single group
When analyzing the age variable in typical development it was observed a higher probability of correct production gradually with age, being more favoring the age period 5:0 -5:5 29.In atypical development, the probability of correct production was not increasing age, and the age range of 7:0 -7:5; 29, was most likely to correct production (Figure 2).
29.As for the variable development type, the results demonstrated a greater probability of deletion of the C 2 in thenatypical development.As for the gender variable, it was found that the male variant has a higher probability of deletion of C 2 (Figure 3).
By analyzing the repair strategy of deletion of C 2 , the results of the statistical analysis, considering one group (typical and atypical), showed that the age, type of development and gender were significant.In the analysis of the age variable, there was a higher probability of deletion of C 2 aged 6:0 -6:5 The linguistic variable position in the word, selected as favoring the deletion of C 2 in typical development, was most likely to occur when the CO occupies the medial position (e.g.quebrado).In the development of atypical syllable position in relation to the foot metric showed favoring the C 2 deletion occurring when it occupies the "head" of the metric foot (Figure 4).
As statistically significant variables for the deletion of the C 2 of the CO in typical development were selected extralinguistic variables gender and age, and linguistic variable position in the word.In the atypical development selected variables were age, sex and metrical foot.
In regard to gender in typical development, it was found that the variant has a higher probability male deletion of C 2 unlike what was observed in atypical development in which most likely occurs in the C 2 female deletion variant.
In the analysis of the age variable, the typical development showed higher probability of deletion Regarding the dependent variable metathesis, the joint data analysis of typical and atypical development, as a single group, there was the influence of the variables following syllabic context, age and type of development in the production of this resource.
In variable following syllabic context it was observed that the complex onset is more likely to be targeted of metathesis when followed by syllable locked with simple coda and simple onset.
Referring to age, the age of 5:6 -5:11; 29 presented favoring the realization of metathesis, i.e., in the ages which understand the atypical data.The typical development stood out as one that has a higher occurrence of metathesis during the process of acquisition of CO (Table 1).In round individual statistic, considering the typical development, the selected variables were significant as following syllabic context, gender and age, in order of statistical significance, respectively.
Regarding variable following syllabic context, it was found that the metathesis is most likely to occur when the CO is followed by syllable locked with simple coda and simple onset.It was also observed a higher probability to utilize the metathesis in the male variant and age group between 4:0 and 4:5;

29.
It was not possible to separate analysis of the metathesis resource of atypical data, due to limited number of occurrences.
In round involving idiosyncratic substitutions in atypical development it was selected variables age and gender as statistically significant.The age range of 5:6 -5:11, 29 was the most favoring to the realization of idiosyncrasies, as well as males (Table 2).
It was not possible to make the analysis of idiosyncratic substitutions in the data of the single group (typical and atypical), due to the non-occurrence of these substitutions in typical development.
In figure 5, it is seen the distribution of the percentages of production of the dependent variables (correct production, deletion of C 2 , deletion of the syllable, epentheses, metatheses, idiosyncrasies) for typical and atypical development.It was observed that in typical development there is a higher percentage of correct production, while the atypical development corresponded to the highest percentage of C 2 deletion.

Figure 5 -Comparison of the percentages of correct production and repair strategies in typical and atypical phonological development
It was also observed that the atypical development uses a greater number of repair strategies (n = 5) compared to typical group (n = 3), being the idiosyncrasies and deletion of the syllable of exclusive use from the atypical group.The other strategies were common to both types of development, with only differences in frequency of occurrence.
Due to the restricted number of occurrences of the variable epentheses, the same has not been selected by the VARBRUL computer program as statistically significant.Regarding the variable deletion of C 1 , it is stressed that there was no occurrence in the studied groups.

DISCUSSION
The discussion will be held in accordance with the repair strategies used by children with typical and atypical phonological development of speech, checking the similarities and differences found from the analysis of the selected variables and considered statistically significant at VARBRUL program.
Analyzing typical and atypical data together as a single group, the statistically significant variables for the correct production of the CO was age, gender and type of development.The variable age had a higher probability of correct production at the age of 7:0 to 7:5, 29, belonging to atypical data, demonstrating that in the phonological disorder there is a significant delay in phonemic acquisition when compared to the normal development 6 .This fact was also verified when the program selected the variable typical development.
In regards to gender in both developments, it was found that the female variant is more likely to correct production and is compatible with a study 10 that talks about the difficulty in the phonological acquisition in males.
Both for normal development and for atypical development, the extralinguistic variables gender and age proved to be significant for the correct production of the CO.In typical development, girls were more likely to correct production, agreeing with the findings of a study on phonological acquisition 19 .In another study 10 , mentioned earlier, the author said that speech disorders are more prevalent in male children, which agrees with the analysis of atypical development.
As the child's age advances, it tends to stabilize the correct production; it means that the phonological system gets closer to the adult 20.21 .It was proved correct when analyzing the variable age in typical development, as there was a greater benefit for correct production gradually with increasing age.In the atypical development, the probability of correct production did not occur gradually with age, showing that children with phonological disorders exhibit instability in correct production CO 9 .The medial position was verified as favoring the deletion of C2 in typical group.In a survey 27 about simple onset and coda, it was found that this same position was identified as favoring the strategy of semivocalization of liquid and the omission of the coda, respectively.
Regarding the dependent variable metathesis, in the joint data analysis of typical and atypical development, as a single group, there was the influence of the variables following syllabic context, age and type of development in the production of this resource.In the group with typical development it was selected in addition to the variables following syllabic context and age, the gender variable.This extralinguistic variable showed that boys use more this feature.This gender difference is consistent with a study that says that there are differences in verbal abilities between boys and girls.It is observed in general that the girls speak earlier and with greater grammatical accuracy, being more precise in language skills 28 .
Regarding to the variable following syllabic context, it was found that the metathesis is most likely to occur when the CO is followed by syllable lock ((C) VC) with simple coda and simple onset (e.g.pretos).This variant was not exclusive of typical group, suggesting that actually sets the variable favoring the production of metathesis in different types of development is the variable age.
The age in the typical was the most relevant age group before the single group, confirming that the use of metathesis takes place in separate periods of developments, namely the use of metathesis typically developing in the group can be considered a stage prior to the complete acquisition of the complex structure and atypical development group may be regarded as a repair strategy performed in the inability of the target.
The variable type of development was always selected when the typical and atypical data ran together, showing no difference between the types of development.
Moreover, it was not possible to analyze the strategy metathesis in the group of children with atypical development for the low number of occurrences.This fact disagrees with a comparative study between children with typical and atypical phonological development in which only the atypical group performed the repair strategy metathesis 26 .
The idiosyncrasies strategy occurred only in the group with atypical development, with low frequency for the selected extralinguistic variables (gender and age).The variant male and the group age 5:6 -5:11, 29 (first track with atypical data) are more favorable to the realization of idiosyncrasies.This way, atypical is characterized by the presence In phonological disorder, the most affected structure as to the correct production is usually the CO 6,9 , agreeing with the findings that the highest probability of correct production occurred in typical development.
By analyzing the repair strategy deleting of C 2 of the OC in single group (typical and atypical), the variables age, type of development and gender were significant for the realization of this strategy.In the analysis of the age variable, the highest probability of C 2 deletion occurred between the ages of 6:0 to 6:5, 29, belonging to the atypical development, and this development was also selected as the most favorable to use the strategy.It is expected that as the age increases, there is less occurrence in the use of strategies 22 and it is noted that in the studied corpus the selected age belongs to the atypical group.The occurrence of this strategy is similar in both groups, confirming what the literature says that in the atypical development there is higher frequency in the use of this resource when trying to produce the target syllable, with no continuous pattern acquisition 6.21 .
In regards to the gender variable, the literature is not specific in addressing the influence of this variable on deletion of C 2 strategy, reporting higher rates of phonological disorders in children of the male gender 23,24 , was found in this study that the male variant has a higher probability of deletion of C 2 .As influential variables for the deletion of C 2 in typical development, selected variables were: the extralinguistic variables gender and age, and linguistic variable position in the word.In the atypical development, there is the influence of the variables age, sex and metrical foot.Age proved to be favorable to the deletion of C 2 whereas in earlier ages there is a high incidence of this strategy, reducing the aging years to complete acquisition of CO 21 .However, in the case of phonological disorder, this linearity did not occur in the acquisition, as peaks were observed in the use of this strategy.In relation to gender, in typical development a higher occurrence of the studied strategy occurred in boys, and in contrast, the most atypical occurrence was observed in girls.As previously mentioned, the literature states that in males occur more cases of phonological disorders, but it is also observed in another study 25 , there may be a higher prevalence of speech disorders in female children.
The position of the syllable in relation to metrical foot showed favorable to the C 2 deletion, occurring when it occupies the "head" of the metrical foot.In a recent study, it was found that when the liquids of CO were in the position of "head" of the metrical foot, there was a deletion of the same strategy 26 .phonological disorders could not properly do it until the expected age, producing the same repair strategies that younger children do in normal phonological acquisition.
The repair strategies, syllable deletion and idiosyncrasies were of exclusive use of the atypical group.This data of idiosyncrasies corroborates with other studies 29,30 .

CONCLUSION
It was found, from the results of this study, that the selected extralinguistic variables gender, age and type of development, and the linguistic variables position in the word, syllable metrical foot and following syllabic context influenced the production of OC in the analyzed corpus.It is noteworthy that as the data were obtained with a small sample; it was not possible to complete their generalization.
It was observed that there are differences in speech development in children typical and atypical in regard to the production of OC, as the variable type of development was selected in every round statistics on which the data were analyzed as a single group.However, considering the separate analysis of the two corpus, the extralinguistic variables (age and gender) were similar to the correct production and other types of strategies used in the two types of phonological development -typical and atypical, influencing both in production and in the correct use of repair strategies.
The intention of this study is to contribute to the existing bases for the elaboration and implementation of a more effective therapy in children with atypical development of speech.At the moment prosodic variables and syllabic favoring the correct production of the OC are known, they aid in the choice of target words, putting the best OC in the context of production, seeking to shorten the therapeutic process. of distinct features in the phonological system and can use unique processes or idiosyncratic, such as slurred speech beyond five years of old 29,30 which corroborates the findings of this research.
Still, in this study, it was possible to analyze the percentages of correct production and repair strategies in typical and atypical phonological development.In typical development there is a higher percentage of correct production, whereas atypical development in the deletion of C 2 showed a higher percentage.It was found that, as mentioned above, given the impossibility of performing the CO children prefer to produce only the obstruent of this onset, or use the strategy of deletion of C 2 13 .
Thus, it was found that in the course of acquiring the complex onset, deletion of C 2 was the repair strategy more frequent in children with typical and atypical development 26 .
In a study 31 on the repair strategy of epentheses it was observed both in typical phonological development as in the speech of children with phonological evolutive disorder, this type of resource.The author also raises the hypothesis that an attempt would be that the epentheses would be a tentative of organization of the phonological system, mainly in typical phonological development, as this acquisition is governed by syllable structure.The results of this research are in the mentioned study 31 , in which both groups used this feature in equal percentages.
Although with a reduced number of occurrences, it was observed that in the atypical development, there was also, as the typical group, the occurrence of metathesis.This finding agrees with the one obtained in another study that observed a very low productivity of metathesis strategy for children with disorder 6 .

Figure 2 -Figure 3 -
Figure 2 -Comparison of the significant variables to the correct production of the complex onset in typical and atypical development

Figure 4 -
Figure 4 -Comparison of statistically significant variables in the deletion of C 2 in typical and atypical development C o r r e c t p r o d u c t i o n c 2 D e l e t i o n E p e n t e s e m e t a t e s e I d i o s y n c r a s i e s S y l l a b l e d e l e t i o n TYPICAL ATYPICAL

Table 2 -Variables statistically significant in the production of idiosyncrasies on target with selected complex onset for the group with atypical development
Legend: Statistical program: VARBRUL; Significance: 5% (p <0.05)