RIJKSUNIVERSITEIT GRONINGEN
The Phonological Abilities at Phoneme, Cluster and
Syllable level of a Spanish-Speaking Boy
with Specific Language Impairment: A Case-
Study
J.R.M. Straatman
straatman.jill@gmail.com
Master: Linguistics, specialization Neurolinguistics Course: Master thesis
2
Acknowledgements
There are a couple of people I need to thank for their support and attributions. First of all, I would like to give a big thanks to Gerard Bol, senior lecturer at University of Groningen. Since I have already forgotten him once before, he was first on my list. He may not be aware of it, but he is the one who got me interested and passionate about Specific Language Impairment. I have learned so much from him during the classes in my Linguistics bachelor and Neurolinguistics master. If I ever achieve the level of knowledge he has, I would be a wise woman. Besides that, he made it possible for me to travel twice to Santiago de Compostela, Spain and he has been of great support in those times. Thank you very much, mister Bol!
Secondly, I would like to thank Flor Velo Negreira and Raquel Lago Vázquez. I had the honor of meeting these fantastic speech therapists during my internship in Santiago and Bertamiráns. They have been a great inspiration and source of comfort. Since it is easier for me to tell them this in Spanish: Me invitáis a vuestra clínica como si fuera una de las vuestras, me aceptáis sin problemas como extranjera, estudiante y persona. Me sentí rodeada de amor, conocimiento y confianza. Cuando recuerdo todos los tiempos maravillosos que teníamos, aparece una sonrisa muy grande en mi cara. Nunca os olvidaré y no me puedo imaginar una vida sin vosotras. Muchísimas gracias!
Then, I have to thank AGP and his parents. From the first moment AGP and I met, I felt a great affection for this boy. He accepted me as an intern at the clinic and as a person in general. Luckily, he never felt shy around me and the fact that we do not both speak Spanish as our native language did not stop him at all from chatting and talking to me. He taught me that children with SLI are real human beings with feelings, stories, families, love and affections. This is something you tend to forget when you are working with data instead of people. AGP’s parents also accepted me and they gave me consent to record their son’s sessions. Without their permission, this thesis would never have existed in its current shape. Os doy mis gracias infinitas!
3
I couldn’t have made it through my time at the University of Groningen without my friends. Willem, José, Linda and Marieke, you guys know me, my ambitions and my fears. I can’t imagine my life without your love and friendship. Even though some of us have moved or probably will move to other parts of The Netherlands, all of you will always have a special place in my heart. Thanks for being there for me! Evelien, Nynke-Boudien and Janneke, I am very grateful for our friendship. You made me laugh during our never ending lunches and you gave me advice when I needed it. I am sure you are going to be great speech therapists and clinical linguists.Finally, I would like to thank Ana Isabel Codesido, professor at the Universidade de Santiago de Compostela. You are one of the great inspirations in my life. I will never forget how much you have helped me. Mister Dicky Gilbers, associate professor at University of Groningen, thanks for your support and attributions when it came down to OT. You have really helped me out. Virginia, María and Tamara, speech therapists in Santiago de Compostela, I want to thank you for welcoming me into your clinic and your lives. You have been a great help in getting me started during my internship.
In conclusion, I would like to cite a very funny Dutch woman Brigitte Kaandorp. She once sung: “Ik heb een heel zwaar leven. Nee, nee, echt waar. Het leven is voor mij gewoon óóntzettend zwaar.” This is most definitely true, but I have to counter her statement: Danoontje Powerrrrrr!
4
The Phonological Abilities at Phoneme, Cluster and Syllable
level of a Spanish-Speaking Boy with SLI: A Case-Study
Abstract
The current study concerns the phonological abilities of a Spanish-speaking boy with Specific Language Impairment. His spontaneous speech and Frogstory (Mayer, 1969) have been recorded during three sessions. The main focus is on three different levels, namely phoneme, cluster and syllable level. The results show that the ratios of correctly produced speech are relatively high and that most difficulties can be found at the phoneme level. The mixed results on substitutions and omissions make it hard to conclude in which phase of phonological development the boy is. The results are explained in the light of Leonard’s Surface Hypothesis (1998), the Sonority Principle (Pater and Barlow, 2003) and Optimality Theory (Prince and Smolensky, 1993).
Keywords: SLI, Spanish Language, Phonological Abilities, Spontaneous Speech, Frogstory.
Introduction
Research on language problems of children with Specific Language Impairment (from now on: SLI) has a rich history. The first reports showed up in 1835 (Gall, 1835). Children with SLI show problems in the development of language. There is no obvious reason to why it is delayed or deviant (Bishop, 2006). Despite having normal hearing abilities, intelligence within the normal ranges and no neurological damage, children with SLI present an atypical language development (Leonard, 1998; Leonard, 2003). According to the latest DSM-V (American Psychiatric Association, 2013) there are two groups of children with SLI. First of all, children who only have difficulties in their expressive language production are characterized as SLI-Expressive. Secondly, children who experience difficulties in both comprehension and production of language fall in the SLI-Receptive and Expressive group.
5
also been found in Dutch-speaking children with SLI (Bol and Kuiken, 1990, De Jong, 1999) and in bilingual Catalan-Spanish children with SLI (Sanz-Torrent, Llorenç and Serra, 2008).It is thought that verb inflection is less of a problem for Spanish-speaking children with SLI than for English-speaking children with SLI, since Spanish has a rich morphology system (Bedore and Leonard, 2001; 2005). Spanish-speaking children with SLI show problems in different grammatical areas, mainly in closed class words, such as prepositions and definite determiners (Serra-Raventos, Media-Aguilar and Sanz-Torrent, 2002). Other grammatical difficulties include pronoun and clitic pronoun use (Restrepo and Gutierrez-Clellen, 2001), adjective agreement (Bedore and Leonard, 2001) and argument structure (Sanz-Torrent, 2002). However, a clinical marker for SLI in Spanish-speaking children hasn’t been found yet. Morgan, Restrepo and Auza (2009) set out to find this marker. Due to individual differences in the participants, their results don’t show a specific pattern.
Children with SLI don’t only show problems in grammatical areas. Other language areas, such as phonology, semantics and pragmatics, can be affected as well. The main focus in the current study is on phonology in Spanish-speaking children with SLI. The phonological difficulties of children with SLI can’t be contributed to deviant articulation skills. Investigation has been carried out in a limited amount of languages, such as English (amongst others: Roberts, Rescorla, Giroux en Stevens, 1998), Swedish (Nettelbladt, 1992), Dutch (amongst others: Beers, 1995) and Spanish (amongst other: Aguilar-Mediavilla, Sanz-Torrent and Serra-Raventós, 2002; Aguilar-Mediavilla and Serra-Raventos, 2006).
6
of four. When children reach the age of five, they enter the third and final stage. This is when the phonological development is completed and the amount of phonological errors reduces. Children tend to substitute phonemes instead of omitting them.The acquisition of tauto- or heterosyllabic clusters seems to be difficult for young children. Tautosyllabic clusters are clusters that appear in the onset of a syllable, whereas a heterosyllabic clusters appears across a syllable boundary (Barlow, 2006). The most frequent strategy to avoid producing the consonant cluster is reduction. This means that one segment is omitted. Some children show difficulties in either tauto- or heterosyllabic clusters, while others seem to experience problems in the production of both groups. According to Preisser, Hodson and Paden (1988) children start producing the first clusters around the age of two. In the beginning, about half the clusters is reduced. The amount of reductions decreases with age. When children reach the age of three years and nine months, a segment is omitted in 25% of the clusters. The reduction of consonant clusters remains a frequent phonological process until it disappears around the age of eight years.
7
I II III IV V VI
CV, V CVC, VC CGV, CGVC CCV,CCVC CVCC, VCC CCVCC Figure 1: The general stages in which different syllable structure emerge in Spanish-speaking children. C = consonant, V = vowel and G = glide. Based on: Morales-Font (2006).
Phonological difficulties in children with SLI become clear when the children are very young. According to Rescorla and Bernstein Ratner (1996) English-speaking children with Specific Expressive Language Impairment (SLI-E) have a delay in phonological development at the age of 24 to 31 months. Results of studies in several languages show that young children with a deviant language development babble less than young children with a typical language development (amongst other: Pávez, Schwalm and Maggiolo, 1986). Later on, these children produce more unintelligible utterances, due to difficulties in the correct realization of phonemes (Scarborough and Dobrich, 1990; Rescorla and Schwartz, 1990). Additionally, children with SLI-E seem to have a limited amount consonants and vocals at their disposal, which leads to limited spontaneous speech (Rescorla and Bernstein Ratner, 1996). The liquids /ɾ/ and /l/ seem to be the most difficult phonemes. Frijn and De Haan (1994) claim that children with SLI have great difficulties in the acquisition of these phonemes.
8
Figure 2: According to Beers (1995), most of the substitution processes take place within this triangle.
Spanish-speaking children with SLI show a very similar pattern in phonological difficulties. Aguilar-Mediavilla et al. (2002) study the phonological abilities of three-year-old Spanish speaking children with SLI. At phoneme level, the participants produce only 69% of all consonants and 85% of vocals correctly. The reduction of tauto- and heterosyllabic clusters is a very frequent phonological process amongst Spanish-speaking children with SLI. More than half of the clusters is reduced. At syllable level, children tend to omit the initial and weak syllable, mainly in closed class words (Aguilar-Mediavilla et al., 2002; Aguilar-(Aguilar-Mediavilla and Serra-Raventos, 2006; Aguilar-(Aguilar-Mediavilla, Sanz-Torrent and Serra-Raventos, 2007). Strikingly, children with SLI have great difficulties with the simple CV syllable structure. Aguilar-Mediavilla et al. (2002) conclude that three-year-old Spanish-speaking children with SLI show a plateau in the development of early elements.
The results of a follow-up study (Aguilar-Mediavilla and Serra-Raventós, 2006) show that the phonological profile of the children with SLI has changed. The percentage of correct produced phonemes increases to 81% and the problems with the initial syllables slowly disappear (Aguilar-Mediavilla and Serra-Raventós, 2006; Aguilar-Mediavilla et al., 2007). However, phonological difficulties still exist at the age of four years and ten months. Children with SLI show significantly more simplification processes on segmental and syllabic level than the control groups. The most frequent phonological processes are cluster reduction and the language specific deletion or reduction of the trill /r/. About 44% of the clusters is reduced. The participants also omit the final consonant around the age of five. On syllable level, the children with SLI show difficulties in the more complex syllables, namely CCV and CVC syllables.
9
difficulties in elements which are short and non-salient. These elements often are not stressed. Leonard (1998) predicts that English-speaking children with SLI will omit morphological elements, due to the limited amount of morphological elements in the English language. On the contrary, in a language that has a rich morphology system, such as Spanish, these elements will be preserved. In these languages, children with SLI will omit auxiliaries, determiners and prepositions.Some results of Spanish studies support the Surface Hypothesis (Leonard, 1998), while others contradict it. The three-year-old children with SLI of Aguilar-Mediavilla et al. (2002) don’t omit stressed and evident morphological elements. Instead, the children reduce the initial and unstressed syllable. Additionally, auxiliaries and prepositions, which are unstressed in Spanish, are omitted. These results support the Surface Hypothesis (Leonard, 1998) The four-year-old children with SLI, on the contrary, frequently omit the final consonant and the trill /r/ (Aguilar-Mediavilla and Serra-Raventós, 2006). When the consonant in the final position is deleted, this can lead to an effect at the morphological level. For example, the plural of nouns in Spanish is formed by adding the morpheme –s at the end of the word. Thus, when this consonant is omitted, a morphological element is omitted. These results don’t support the Surface Hypothesis (Leonard, 1998). Aguilar-Mediavilla and Serra-Raventós (2006) conclude that more research about the connection between the phonological and morphological level is required to explain these data.
10
Table 1: A table with all possible tautosyllabic consonant clusters in Spanish, based on Hualde et al. (2010). pɾ primero pl pluma tɾ trampa (tl) (Tlaloc) kɾ crimen kl claúsula fɾ francés fl flor bɾ broma bl blusa dɾ drama *dl * gɾ gris gl gloria
The reduction of tautosyllabic clusters is mostly explained based on the Sonority Principle (Pater and Barlow, 2003). It is stated that sonority is a characteristic of phonemes. Sonority increases in the optimal syllable, until it reaches its highest point at the vowel (Clements, 1990). Every phoneme receives a value of sonority, showing how much the phoneme is similar to a vowel. These values are ranked and projected on a sonority scale (see figure 3). Several studies show that the most sonorant element is omitted in cluster reduction. The most sonorant segment is the most vowel-like element. Spanish tautosyllabic clusters consist of a fricative or occlusive and a liquid. Since the liquids /ɾ/ and /l/ have a high sonorant value (3) compared to the fricatives (1) and stops (0), these elements are omitted.
Figure 3: The sonority scale, including the sonority values, according to Barlow (1997).
11
consonants include nasals /m, n, ŋ/, liquids /l, ɾ/ and glides /w, j/ (Shepherd, 2003; Kooij and Van Oostendorp, 2003). In a more complex syllable, a consonant can be added before or after the nucleus. The pre-vocalic consonant is positioned in the onset. If a cluster is part of the onset, the least sonorant element is placed in the satellite position. In Spanish, the cluster in the onset has a maximum of two segments. This means that the position of pre-margin is an extrasyllabic position is the Spanish language. On the contrary, the pre-margin is used in Dutch, in cases like streep. The postvocalic consonant is placed in the coda. When the syllable ends in a cluster, the second element stays in the appendix.f l o r
d j e z
c o n s
Figure 4: The representation of a complex syllable in a tree. The thin lines reflect positions that exist in every language, while the thick lines reflect extrasyllabic positions. Based on Kooij and Van Oostendorp (2003).
12
Table 2: An example of the checking of possible output, according to the ranking of several constraints, based on Gilbers and De Hoop (1998).
constraints Input candidates C1 >> C2 >> C3 >> C4 ... Output 1 *! * Output 2 * Output 3 * *!
In table 2, three different possible outputs are shown horizontally and the applicable constraints are shown vertically. The most dominant constraint is placed on the left side and the least dominant one is on the right. Output 1 violates the first and highest ranked constraint (*!), so this option is no longer relevant. This is indicated by the grey squares. Both output 2 and output 3 violate the second constraint (*). The checking process continues, since the optimal output hasn’t been selected yet. Only output 3 violates the third constraint. This leads to the conclusion that output 2 is the optimal output. The optimal candidate is indicated by the symbol.
OT has different sets of constraints. Two of the most important groups are markedness
constraints and correspondence constraints (Gilbers and De Hoop, 1998). Markedness
constraints make sure that the output with an unmarked structure is preferred. Examples are Ons (every syllable has an onset) and *Coda (no syllable has a coda). In correspondence constraints the focus is on the relationship between the input and the output. Examples are Max-IO, which prevents deletion (every segment in the input has a correspondent in the output) and Dep-IO, which avoids insertion (every segment in the output has a correspondent in the input).
13
differences, Barlow (2006) states that there is a preference for the unmarked syllable structure. This means that the *Complex-Ons and *Complex-Coda constraints are dominant over correspondence constraints. This ranking leads to simplification processes, such as vowel intrusion and cluster reduction.The current study focuses on the phonological abilities of a Spanish-speaking boy with SLI, AGP. By studying his semi-spontaneous in a narrative and spontaneous speech at segment, cluster and syllable level, the current goal is to create a phonological profile of his abilities. Then, a closer look is taken at the difficulties at segment, cluster and syllable level. In the discussion section, AGP’s results will be compared to the results of Aguilar-Mediavilla et al. (2002), Aguilar-Mediavilla and Serra-Raventós (2006). Secondly, the results will be compared to those of a Spanish-speaking boy named Edgar. His data have been collected by Miranda, Martínez and Diez-Itza (1998) and they are published in the CHILDES database (MacWhinney, 2000). The analysis at phonological level of Edgar’s data has been carried out by the current author for her unpublished bachelor thesis. Furthermore, the results will be discussed in the light of the three theories - Leonard’s Surface Hypothesis (1998), the Sonority Principle by Pater and Barlow (2003) and Optimality Theory (Prince and Smolensky, 1993).
Method
1. ParticipantThe child AGP is a five-year-old boy with SLI. His native tongue is Castellano (Spanish). He lives in Bertamiráns, Galicia. He has had limited contact with the official language of Galicia, Gallego. All of the recorded data are in his first language Castellano. AGP attends primary school in Bertamiráns. At the age of four he was diagnosed with Specific Language Impairment. Since then, he attends a speech therapist once a week and he receives additional support at school from his teacher.
14
The data stem from sessions that were recorded during the internship of the current author. AGP’s speech was recorded three times. Every session includes at least 60 minutes of spontaneous and semi-spontaneous speech. The first recording is on the 6th of June 2013. During this first session AGP performed the Frog Story (Mayer, 1969). The second half of the transcript consists of free play. The second session was recorded a week later, on the 13th of June. AGP, the speech therapist and the intern played a guessing game and AGP enjoyed free play with another boy (M) that attends the speech therapist. The final recording is of the 20th of June. During this session AGP, the speech therapist and the intern conversed spontaneously.2. Transcription
The recorded speech is transcribed according to the CHAT-program of CHILDES (MacWhinney, 2000). The phonological errors have been coded. These codes are based on Miranda, Martínez and Diez-Itza (1998), who present a case study of a Spanish-speaking boy with SLI, Edgar. At the age of seven, Edgar shows severe phonological and grammatical difficulties. Miranda et al. (1998) designed a system to code the errors at phonological level. An example of the transcription and the phonological coding of an error of AGP is presented in figure 5. Appendix 1 presents a general view of the used phonological codes.
Figure 5: An example of a phonological error in the transcript and its phonological coding. 3. Analysis
15
On the phoneme level, the ratio of omission, substitution, insertion and metathesis is calculated by dividing the amount of errors by the total number of phonemes. On the cluster level, the errors are specified in two groups, clusters with the tap /ɾ/ and clusters with the liquid /l/. Again, the ratios are calculated by dividing the errors by the total amount of clusters. Finally, on the syllable level, the focus is on omission of entire syllables and the complexity of the syllable structure. The ratio of both errors is calculated by dividing the amount of errors by the total number of syllables.Results
Before turning to the most frequent phonological processes, the less frequent processes of insertion and metathesis will be discussed. First of all, insertions are present in every transcript. The first and second transcript both contain 14 insertions. The most frequent insertion of the first transcript is that of the nasal phoneme –n to the end of a word. In the second transcript, AGP most often adds a vowel. The third and final transcript contains eight insertions. The majority of these insertions consists of adding –n at the end of the word.
The second uncommon phonological process is that of switching of phonemes or syllables, metathesis. Metathesis at phoneme level is more frequent than metathesis at syllable level. In the first transcript, AGP shows the process six times at phoneme level, while it only occurs twice at syllable level. During the second session seven processes of metathesis at phoneme level and four processes of metathesis at syllable level appear. Metathesis occurs twice at phoneme level and only once at syllable level in the final transcript.
16
Table 3: In the table the total number of words, syllables and cluster is presented for every transcript.
Words Syllables Clusters
6-6-2013 1551 2635 63
13-6-2013 1545 2540 120
20-6-2013 1601 2642 87
Table 4: The table presents the ratios of correct realizations of phonemes, syllables and clusters for every transcript.
Phonemes Syllables Clusters
6-6-2013 88,7% 97% 90,4%
13-6-2013 86,3% 95,2% 83,3%
20-6-2013 89,3% 96,5% 87,3%
AGP has the lowest ratio of correct realizations at phoneme level. In the first transcript, 88,7% of his speech is produced correctly. For the second session the ratio of correctly produced phonemes is 86,3%. In the final transcript, 89,3% of the words isn’t affected by a phonological error. At syllable level, AGP’s production contains few errors, such as omission of an entire syllable or a simplification process. During the first session, 97% of the syllable structures is produced correctly. This percentage decreases to 95,2% during the second session. 96,5% of the syllables are produced correctly in the third transcript. The ratio of correctly produced clusters is 90,4% for the first transcript. 83,3% of tautosyllabic cluster is produced correctly in the second transcript. The ratio of correctly produced cluster is 87,3% for the last session.
17
4,3% of the words a phoneme is replaced by a different phoneme. The most frequent substitutions concern a vowel. When the speech production of all transcripts is taken together, 120 vowels are replaced by a different vowel. The second most frequent substitution process is rhotacism. This simplification process occurs when the trill /r/ is replaced by the tap /ɾ/. It occurs eleven times in all transcripts together. The third high frequent substitution process is a process known as seseo in Spanish. This happens when the interdental /θ/ is pronounced as the dental /s/. It occurs nine times.Figure 6: The figure shows the ratio of omission and substitution of phonemes in AGP’s spontaneous speech. The ratios are shown separately for the three transcripts.
The total amount of clusters produced by AGP can be found in table 3. Words containing a tautosyllabic cluster with the tap /ɾ/ are more frequent than clusters with the liquid /l/ in AGP’s spontaneous speech. In the first transcript, AGP produces 47 clusters with the tap /ɾ/ and 16 clusters with /l/. In the second transcript, the child produces 66 words with a /ɾ/ cluster and 54 words with a /l/ cluster. During the final session, the amount of clusters with /ɾ/ is 73 and the number of cluster with /l/ is 14.
18
Table 5: The table shows the errors in the production of all possible clusters with /ɾ / and /l/.
Cluster Errors Cluster Errors
pɾ 1 pl 2 tɾ 12 (tl) 0 kɾ 0 kl 1 fɾ 0 fl 3 bɾ 0 bl 8 dɾ 3 *dl 0 gɾ 6 gl 1 Total 22 Total 15
At syllable level, AGP seems to experience few phonological difficulties, since the percentage of correctly produced syllable structures is high (see table 3). The errors in syllable structure are divided into two groups, namely the phonological processes leading to omission or insertion of an entire syllable (errors in number) and those resulting in a reduced syllable complexity (complexity errors). The ratios these errors are presented in figure 7. Complexity errors are more frequent than errors in number in every transcript. During the first session, 1,7% of the syllables is affected by a phonological processes that influences the complexity of the syllable structure. In the second transcript, the ratio of complexity errors increases to 2,7%. In the third and final transcript the ratio is 2,2%. The most difficult structure seems to be the CVC-syllable. In 71 CVC-syllables a process that leads to reduced complexity occurs. The second most difficult syllable is the simple CV-syllable. A total of 34 CV-syllables are reduced in complexity.
19
occurs 61 times. Secondly, words of three syllables are reduced to a two-syllables word 21 times in AGP’s speech.Figure 7: The ratios of errors in syllable structure, divided into number of syllables and complexity.
Discussion
Even though AGP experiences some phonological difficulties, especially at phoneme level, AGP’s spontaneous speech production contains few errors. The percentages of correctly produced speech are relatively high, namely an average of 88,3% correctly produced phonemes, 87% correctly produced clusters and 96,2% correctly produced syllables. These results are line with those of Aguilar-Mediavilla et al. (2002) and Aguilar-Mediavilla and Serra-Raventós (2006). They indicate that children with SLI progress in their phonological development in a period of one year. The three-year-old Spanish-speaking children with SLI of Aguilar-Mediavilla et al. (2002) produce 69% of the consonants and 85% of the vowels correctly. At the age of four, the children have improved. The ratio of correctly produced phonemes increases towards 81% (Aguilar-Mediavilla and Serra-Raventós, 2006). The five-year-old AGP in the current study produces 88,3% of all phonemes correctly. The study of Edgar ‘s spontaneous speech (Miranda et al., 1998; Straatman, 2012) shows that this seven-year-old boy produces 62,5% of the phonemes correctly. This percentage differs from that of the earlier studies and the current study. However, Edgar has been selected as a participant for the study of Miranda et al. (1998), based on his striking phonological difficulties.
The results at phoneme level are partially consistent with results of earlier studies on Spanish and other languages. Those studies show that children with SLI from different
20
language backgrounds present several phonological difficulties. Some of their compensation processes include omission of phonemes and syllables and substitution (Rescorla and Bernstein, 1996; Nettelbladt, 1992; Bortolini and Leonard, 2000). The most frequent phonological processes at phoneme level in AGP’s spontaneous speech are omission and substitution. In two transcripts omission occurs more often than substitution, while in the first transcript substitution is the most frequent phonological process. Diez-Itza and Martínez (2004) claim that children enter the third and final stage of phonological development at the age of five. This is when substitution becomes more frequent than omission. Since AGP doesn’t present a clear pattern in omission and substitution errors, it isn’t very easy to place AGP in the different stages in the final phase of phonological development.A possible explanation for the mixed results may come from the different tasks performed during the sessions. During the first session, AGP tells the Frog Story (Mayer, 1969). This isn’t completely free and spontaneous speech. The child has to rely on his narrative abilities. Even more, when telling a story one has to meet several demands. During the other sessions the speech produced by AGP is spontaneous and free. There is a possibility that the different settings affected the results.
When a closer look is taken at the substitution processes, the results aren’t consistent with earlier studies on substitutions. Beers (1992) proposes that most frequent substitutions take place within a triangle (see figure 2 on page seven). The most frequent substitution processes in AGP’s speech production do not take place within this triangle. The results of the analysis of the language production of the child described by Straatman (2012), Edgar, do support the idea that the most frequent substitutions include the phonemes /l/, /d/ and /ɾ/. Apparently, there are differences in the pattern of substitution errors in children with SLI speaking the same language. This could be a striking result, but there might be several explanations.
21
AGP and children with a typical language development show well-known phonological processes in Spanish, such as rhotacism and seseo. The substitution of vowels in AGP’s production could be attributed to difficulties at morphology level. We will return to this point later on. AGP does present substitutions that take place within the triangle, but these aren’t the most frequent processes. Secondly, Edgar and AGP are not matched on age nor language age. AGP is five years old at time of testing, while Edgar has already reached the age of seven. Moreover, AGP has an average MLU of 3.471. The MLU of Edgar is unknown. This lack of matching makes it hard to directly compare the two children.According to Beers (1995), children with SLI don’t only present omission and substitutions to compensate for their limited phonological skills. She states that children with SLI also show insertion and metatheses. These processes also occur in AGP’s spontaneous speech. A segment is inserted in 36 words, which is most frequently placed at the end of a word. A total of 22 metatheses processes appear. Switching phonemes is more frequent than switching syllables.
It is impossible to state that all errors at phoneme level are purely phonological. In most phonological studies there is a close relationship between phonology and morphology. This is most obvious in words that show morphological concordance. For example, the article las is frequently produced as los. This is a phonological error, since a vowel /a/ is substituted by the vowel /o/. However, this is also a morphological error if the noun is a masculine noun. Whether these types of errors should be analyzed as a phonological error, a morphological error or as both, remains uncertain.
22
findings in Edgar’s study (Straatman, 2012). Edgar reduces the sonorant element of the cluster in over 80% of all produced clusters.The reduction of the most sonorant element in clusters is consistent with existing theories on cluster reduction. According to Menn and Stoel-Gammon (1995), the occlusive will be produced in cluster reduction. Several explanations for this finding can be found in the literature on cluster reduction. First of all, Frijn and De Haan (1994) state that the liquids are acquired in a late stage of phonological development. Since children with SLI show a delayed language development, it is possible that they haven’t fully acquired the liquids. The second possible explanation stems from the theory on syllable trees, which is based on the Sonority Principle (Pater and Barlow, 2003). Elements with a high sonorant value are placed in a satellite position in the syllable tree. This position is extrasyllabic and weak. In the Surface Hypothesis (Leonard, 1998), it is stated that children with SLI have great difficulties with elements which are short and non-salient, due to limited processing capacity. Since the satellite position is extrasyllabic, it makes sense that this position is eliminated.
Reduction isn’t the only phonological process present in cluster production. Especially in the clusters with /l/, the entire cluster is replaced by a different phoneme. The latter less frequent phonological process is covered by Optimality Theory (Prince and Smolensky, 1993). Table 6 shows an example of how OT can deal with a phonological error in AGP’s spontaneous speech. In the second transcript the word ancla is produced as
ancha. This error has an effect on two levels. First of all, the error takes place at the
23
Table 6: The table shows an example of the checking process in OT (Prince and Smolensky, 1993) for the production of the word ancha instead of ancla by AGP.
/aŋ.kla/ *Complex-Ons Max-IO
aŋ.kla *!
an. t ʃa *
AGP also replaces the liquid /l/ of the cluster by the other liquid /ɾ/. This occurs twice in his speech, namely sabre /’sa.bɾe/ instead of sable /’sa.ble/ and igrú /i.’ɣɾu/ instead of
iglú /i.’ɣlu/. Based on the previous OT analysis in table 6, it could be concluded that
unmarkedness constraints are ranked higher than faithfulness constraints. However, both unmarkedness constraints MSD (Maximum Sonority Distance) and *Complex-Ons don’t apply here. The two outputs contain a liquid, so there is no difference in sonority. Secondly, both outputs have a complex cluster. The explanation for the /l/ substitution process can be found at acoustic level, since /ɾ/ and /l/ are very similar at this level. The difference between the two can be found their formant pattern in locus frequencies (Gilbers, 2002). As can be seen in figure 8, /ɾ/ has a middle F2 frequency and a high F3 frequency. The liquid /l/, on the contrary, has a middle F2 frequency and a low F3 frequency.
Figure 8: The figure shows the acoustic value in hertz (Hz) of the second and third formants of the liquids /r, l/ and glides /w, j/. Based on Ainsworth and Paliwal, 1984 (simplified).
24
Max-IOF2 checks the value of the second formant. This constraint ensures that the elements in the output have the same value for F2 as the elements in the input. Since the glides /j, w/ have a different F2 value than /l/, these outputs violate the constraint Max-IOF2. The possible outputs /’sa.bje/ and /’sa.bwe/ aren’t relevant any longer. This leaves /’sa.ble/ or /’sa.bɾe/. Since the difference between /ɾ/ and /l/ lies in the third formant, the constraint Max-IOF3 is an important constraint. However, AGP doesn’t produce /’sa.ble/, but /’sa.bɾe/. This means that there is a different constraint ranked higher than Max-IOF3. This constraint is SonCon*lat, which states that a lateral of the sonorant consonants isn’t produced.Table 7: The table shows the checking process of the applicable constraint for the realization of
/sa.ble/ as /sa bɾe/.
/sa.ble/ Max-IOroot Max-IOF1 Max-IOF2 SonCon*lat Max-IOF3 *Compl-Ons
sa.be *!
sa.bwe *!
sa.bje *!
sa.bɾe * *
sa.ble *!
According to Gilbers (2002), the liquid /l/ is a difficult phoneme to acquire. Its production requires the tip of the tongue to be placed against the teeth and the air has to pass the sides of the tongue. In the early stages of liquid acquisition /ɾ/ and /l/ are allophones for the same underlying form. This statement is supported by the results of several studies in Dutch (Gilbers, 2002; Hinskens, 2012). Even more, the liquids can be difficult for second language learners. A well-known example is Japanese. According to McClelland, Fiez and McCandliss (2002), Japanese adults experience difficulties in the acquisition of the English /l/ and /r/, both in production and perception. These difficulties persist, even though the Japanese speaker may have been in touch with the English language for a longer period of time and may have had a lot of training.
25
CV-syllables and the more complex CCV-structure. According to Morales-Font (2006), syllable acquisition passes through six general stages (see figure 1 on page six). The most difficult structures for AGP are syllables which are acquired during the early stages. The CVC-syllable is acquired in the second stage, CV-syllable in the first stage and CCV in the fourth phase. These results are consistent with the studies on three- and four-year-old Spanish speaking children with SLI. The results of Aguilar-Mediavilla, et al. (2002) show that the three-year-old children have great difficulties with the simple CV-structure. They conclude that there is a plateau in the development of syllable structures. Aguilar-Mediavilla and Serra-Raventós (2006) state that the abilities at syllable level of the children have improved. The most difficult structures are CCV and CVC. The exact same structures are most frequently simplified by AGP.The second simplification process at syllable level is the reduction or insertion of an entire syllable, or an error in number. The most frequent number error is that of reduction of a syllable. The most frequently affected words are short words. They consist of two or three syllables. These results support the idea that children with SLI show a plateau in the development of elements which are acquired early in the phonological development. In conclusion, AGP does show some phonological difficulties in his spontaneous speech. The phoneme level is affected most, since the percentage of correctly produced phonemes is the lowest compared to the percentages at cluster and syllable level. The most frequent error is the omission of a phoneme. For future research it is interesting to focus on these omissions in order to test the Surface Hypothesis (Leonard, 1998). In this theory it’s assumed that in a morphologically rich language no elements that attribute to the meaning of a word are omitted. It isn’t within the scope of the current study to test this assumption.
26
Finally, the current study focuses on AGP’s spontaneous speech. It is possible that the results might change when a more controlled setting is chosen. Maybe AGP avoids the production of several phonemes, clusters or complex words in his spontaneous speech, due to limited phonological skills or the fear of making an error. In a more controlled setting, for example a repetition task, AGP could show a different error pattern.References
Aguilar-Mediavilla, E., Sanz-Torrent, M. and Serra-Raventós, M. (2002). A Comparative study of the Phonology of Pre-school Children with Specific Language Impairment (SLI), Language Delay (LD) and Normal Acquisition. Clinical Linguistics and Phonetics, 16, 573 - 596.
Aguilar-Mediavilla, E. and Serra-Raventós, M. (2006). Phonological Profile of Spanish-Catalan Children with Specific Language Impairment at Age 4: Are There any Changes over Time?. Folia Phoniatrica et Logopaedica, 58, 400 - 414.
Aguilar-Mediavilla, E., Sanz-Torrent, M. and Serra-Raventós, M. (2007). Influence of Phonology on Morpho-Syntax in Children with Specific Language Impairment (SLI).
International Journal of Language and Communication Disorders, 42, 325 – 347.
American Psychiatric Association (2013). Diagnostic and Statistical Manual of Mental
Disorders (5th Ed.). Arlington, VA: American Psychiatric Publishing.
Ainsworth, W.A. and Paliwal K.K. (1984). Correlation between the Production and Perception of the English Glides /w,r,l,j/’. Journal of Phonetics, 12, 237-243.
Barlow, J.A. (2006). “Constraint Conflict in the Acquisition of Clusters in Spanish”. In: Martínez-Gil, F. and Colina, S. (eds.), Optimality-Theoretic Studies in Spanish
Phonology, Amsterdam/Philadelphia: John Benjamins Publishing Company, 525 – 548.
Bedore, L.M. and Leonard, L. (2001). Grammatical Morphology Deficits in Spanish-speaking Children with Specific Language Impairment. Journal of Speech, Language,
and Hearing Research, 44, 905 – 924.
27
Impairment. Applied Psycholinguistics: Special Issue: Psychological and LinguisticStudies across Languages and Learners, 26, 195 – 225.
Beers, M. (1992). Phonological Processes in Dutch Language-Impaired Children.
Journal of Logopedics and Phoniatrics, 17, 9 - 16.
Beers, M. (1995). The Phonology of Normally Developing and Language-Impaired Children. Studies on Language and Language Use, 20 (IFOTT Amsterdam).
Bishop, D.V.M. (2006). What Causes Specific Language Impairment in Children?
Current Directions in Psychological Science, 25, 217 – 221.
Bol, G. W., and Kuiken, F. (1990). Grammatical Analysis of Developmental Language Disorders: A Study of the Morphosyntax of Children with Specific Language Disorders, Children with Hearing Impairment and Children with Down’s syndrome. Clinical
Linguistics and Phonetics, 4, 77–86.
Bortolini, U. and Leonard, L. (2000). Phonology and Specific Language Impairment: Status of Structural Constraints in Two Languages. Journal of Language and
Communication Disorders, 30, 131 – 150.
Clements, N.G. (1990): "The Role of the Sonority Cycle in Core Syllabification". In: Kingston, J. and Beckmann, M. (eds): Papers in Laboratory Phonology I, Cambridge: Cambridge University Press, 283 - 333.
Craig, K. and Evans, J.L. (1993). Pragmatics and SLI: Within-Group Variations in Discourse Behaviors. Journal of Speech and Hearing Research, 36, 777 – 789.
Diez-Itza, E. and Martínez, V. (2004): Las Etapas Tardías de la Adquisición Fonológica: Procesos de Reducción de Grupos Consonánticos. Anuarios de Psicología, 35, 177 - 202.
Fikkert, P. (1994). On the Acquisition of Prosodic Structure. PhD Dissertation, University of Leiden. The Hague: HAG.
28
Frijn, J. and De Haan G. (1994): Het Taallerende Kind. Dordrecht: ICG Publications. Gall, F. (1835). The Function of the Brain and Each of its Parts. 5, Organology. Boston: Marsh, Capen & Lyon.Gilbers, D. (2002). “Conflicting Phonologically Based and Phonetically Based Constraints in the Analysis of /l/-Substitutions”. In: Beers, M., Jongmans, P. and Wijnands, A. (eds), Netwerk Eerste Taalverwerving, Net-Bulletin 2001, Leiden, 22 – 40. Gilbers, D. and De Hoop, H (1998). Conflicting Constraints: An Introduction to Optimality Theory. Lingua, 104, 1 – 12.
Hiskens, F. (2012). “Liquids in a Case of Unfolding Early L1 Dutch: From Null Realizations through Free Variation through Probabilistic Bound Variation to Lexical Contrast”. In: Botma, B. and Noske, R. (eds), Phonological Explorations Empirical,
Theoretical and Diachronic Issues, Berlin / Boston: De Gruyter, 313 – 331.
Hualde, J.I. (1991). “On Spanish Syllabification”. In: Campos, H. and Martínez-Gil, F. (eds), Current Studies in Spanish Linguistics, Washington DC: Georgetown University Press, 475 – 493.
Hualde, J.I., Olarrea, A., Escobar, A.M. & Travis, C.E. (2010). Introducción a la
Lingüística Hispánica. Cambridge: Cambridge University Press.
Ingram, D. (1976). Phonological Disability in Children. New York: Elsevier.
Jong, J. de (1999). Specific Language Impairment in Dutch: Inflectional Morphology and
Argument Structure. Dissertation, University of Groningen, Groningen, The Netherlands.
Kooij, J. and Oostendorp van, M. (2003). Fonologie: Uitnodiging tot de Klankleer van het
Nederlands. Amsterdam: Amsterdam University Press.
Leonard, L.B. (1998). Children with Specific Language Impairment. Cambridge, MA: MIT Press.
Leonard, L.B. (2003). “Specific Language Impairment: Characterizing the Deficit”. In: Levy, J. and Schaeffer, J. (eds), Language Competence across Populations: Toward a
Definition of Specific Language Impairment, London: Lawrence Erlbaum Associates, Inc,
29
Levelt, C., Schiller, N.O. and Levelt, W.J. (2000). The Acquisition of Syllable Types.Language Acquisition, 8, 237-264.
MacWhinney, B. (2000). The CHILDES Project: Tools for Analyzing Talk (3rd Ed.). Mahwah, NJ: Lawrence Erlbaum Associates.
Mayer, M. (1969). Frog, Where Are You? New York: Dial Press.
McClelland, J.L., Fiez, J.A. and McCandliss, B.D. (2002). Teaching the /r/-/l/ Discrimination to Japanese Adults: Behavioral and Neural Aspects. Physiology and
Behavior, 77, 657 – 662.
Miranda M., Martínez V. and Diez-Itza, E. (1998). "Procesos Fonológicos en la Disfasia
Infantil." Paper presented at: The Second Meeting on Language Acquisition in Barcelona
(1998).
Morales-Font, A. (2006). “Acquisition of Syllable Structure in Spanish”. In: Martínez-Gil, F. and Colina, S. (eds), Optimality-Theoretic Studies in Spanish Phonology, Amsterdam/Philadelphia: John Benjamins Publishing Company, 497 – 524.
Morgan, G., Restrepo, M.A. and Auza, A. (2009). “Variability in the Grammatical Profiles of Spanish-Speaking Children.” In: Grinstead, J. (ed), Hispanic Child Languages: Typical
and Atypical Development, Amsterdam / Philadelphia: John Benjamins Publishing
Company, 283 – 302.
Nettelbladt, U. (1992). "Language Impairment in Swedish Children: A Multidisciplinary Approach". In: Fletcher, P. and Hall, D. (eds), Specific Speech and Language Disorders
in Children: Correlates, Characteristics and Outcomes, London: Whurr, 138 – 151. .
Pater, J. and Barlow, J.A. (2003). Constraint Conflict in Cluster Reduction. Journal of
Child Language, 30, 487-526.
Pávez, M.M., Schwalm, E. and Maggiolo, M. (1986). Trastornos Fonológicos en Niños con Retraso Simple de Lenguaje: El Uso de Fonémas Consonánticos. Revista
Logopédica de Fonoaudiología y Audiología, 6, 205 - 214.
30
Prieto, P. and Bosch-Baliarda, M. (2005). The Development of Codas in Catalan.Catalan Journal of Linguistics, 5.
Prince, A. and Smolensky, P. (1993). Optimality Theory: Constraint Interaction in
Generative Grammar. Ms., Rutgers University and University of Colorado.
Rescorla, L. and Bernstein Ratner, N. (1996). Phonetic Profiles of Toddlers with Specific Expressive Language Impairment (SLI-E). Journal of Speech and Language Research, 39, 153 - 165.
Rescorla, L. and Schwartz, E. (1990). Outcome of Toddlers with Expressive Language Delay. Applied Psycholinguistics, 11, 393 – 407.
Restrepo, M.A. and Gutierrez-Clellen, V.F. (2001). Article Use in Spanish-Speaking Children with Specific Language Impairment. Journal of Child Language, 28, 433 – 452. Rice, M. and Wexler, K. (1996). Towards Tense as a Clinical Marker of Specific Language Impairment in English-Speaking Children. Journal of Speech and Hearing
Research, 39, 239-257.
Roberts, J., Rescorla, L., Giroux, J. and Stevens, L. (1998). Phonological Skills of Children with Specific Expressive Language Impairment (SLI-E): Outcome at Age 3.
Journal of Speech, Language and Hearing Research, 41, 374 - 384.
Sanz-Torrent, M. (2002). Los Verbos en Niños con Trastornos del Lenguaje. Revista de
Logopedia, Foniatría y Audiología, 22, 100 – 110.
Sanz-Torrent, M., Serrat, E., Llorenc, A and Serra, M. (2008). Verb Morphology in Catalan and Spanish in Children with Specific Language Impairment: A Developmental Study. Clinical Linguistics and Phonetics, 22, 459-474.
Scarborough, H. and Dobrich, W. (1990). Development of Children with Early Language Delay. Journal of Speech and Hearing Research, 33, 70 - 83.
Serra-Raventos, M., Aguilar-Mediavilla, E. and Sanz-Torrent, M. (2002). Evolución del Perfil Productivo en el Trastorno del Lenguaje. Revista de Logopedia, Foniatría y
31
Shepherd, M.A. (2003). Constraint Interactions in Spanish Phonotactics: An OptimalityTheory Analysis of Syllable-Level Phenomena in the Spanish Language. Northridge:
California State University.
Southwood, F. and Russell, A. (2004). Comparison of Conversation, Free Play and Story Generation as Methods of Language Sample Elicitation. Journal of Speech, Language
and Hearing Research, 47, 377 - 388.
Straatman, J. (2012). De Fonologische Ontwikkeling van een Spaanse Jongen met SLI:
Een Case Study. Unpublished Manuscript, Dept. Of Linguistics, University of Groningen.
Wethering, D., Botting, N. and Conti-Ramsden, G. (2007). Narrative in Adolescent SLI: A Comparison with Peers across two Different Narrative Genres. International Journal of
32
Appendix 1: An overview of the phonological codes used in the analysis.
ES (relativos a la estructura de la sílaba) – Related to syllable structureRD (reduplicación) - Reduplication
CF (supresión de consonantes finales) - Omission of the final consonant SA (supresión de sílabas átonas) – Omission of a unstressed syllable MT (metátesis) - Metatheses:
FO (fonémica) - Phonematic SI (silábica) - Syllabic
CT (consonánticos) – Concerns a consonant VO (vocálicos) – Concerns a vowel
OM (omisión) - Omission: LQ (líquidas) - Liquids NS (nasales) - Nasals
ON (oclusivas sonoras) – Voiced occlusive OR (oclusivas sordas) – Unvoiced occlusive FC (fricativas) - Fricatives
ST (sustitución) - Substitution: LQ (líquidas) - Liquids
LV (lateralización de vibrantes) – Lateralisation of trill AM (ausencia de vibrante múltiple) – Absence of trill /r/ AV (ausencia de vibrante simple) – Absence of tap /ɾ/ LR (sustitución de l por vibrante) – Substitution of /l/ by /ɾ/ SV (simplificación de vibrante) – Simplification of /r/ NS (nasales) - Nasals:
LT (lateralización) - Lateralisation
00 (procesos no analizables) – Processes that can’t be analyzed NI (no inteligible) – Incomprehensible utterance
