THE ACQUISITION OF SPANISH PRONUNCIATION BY WELSH LEARNERS : TRANSFER FROM A REGIONAL VARIETY OF ENGLISH INTO SPANISH

Language teachers agree that the phonetic/phonological distance between a learner’s L1 and L2 is of vital importance in mastering the sounds of the L2; however, no attention is given to the phonetic/phonological distance between the regional variety of the speaker’s L1 and the L2. After comparing linguistic peculiarities of English, and of Welsh English in particular, with Castilian Spanish, the author proceeds to study the interlanguage of advanced students of Spanish from Wales. This helps to explain positive and negative transfer from this variety of English into Spanish and assists in producing a catalogue of the interferences to be corrected. Finally, using coarticulatory processes, the author presents a list of exercises to eliminate some of the interferences found in Welsh learners of Spanish. A descriptive analysis of the learners’ interlanguage shows that the herrero: spanish pronunciation 170 phonetic/phonological distance between the regional variety of the speaker’s L1 and the L2 is what will influence this learning process the most.


Introduction
The study of L2 phonetic acquisition is complex, and this becomes apparent once we realise the great deal of existing disagreement on issues related to the acquisition of the L2 phonetic system.As we will see, negative transfer or interference (those features of the L1 that influence negatively the pronunciation of a learner in the L2), and positive transfer (those features of the L1 that are similar to the L2 and which make it easier for a speaker to pronounce the L2) have been widely studied.Virtually all scholars agree on the fact that mispronunciation of the L2 is due to negative transfer from the L1; however, there is no consensus on what causes this.There are some authors, such as Major (2008: 76), who believe that interferences are mainly caused by universals, while others believe that this is mainly due to an influence from the L1, such as Strange and Shafer (2008: 153); the debate is ongoing.However, we cannot ignore some obvious facts.As Strange and Shafer (2008: 153) explain: "The fact that native listeners can readily identify a) that a speaker is a late learner of their language and b) the native language of the speakers, (i.e., Spanishaccented English speakers; American-accented Japanese speakers) justifies the characterisation of the accented phonological patterns as being due, to a considerable extent, to interference from the native language phonology." There are several features of L1 speech which offer numerous data about a speaker, such as social class, and region of origin.The analysis of the linguistic details present in a person's L1 speech, especially if it is in a relaxed and spontaneous conversation (Silva-Corvalán & Enrique-Arias, 2001: 52), will enable us to create his/her sociolinguistic profile and identify his/her regional accent.Furthermore, as explained in Herrero de Haro and Andión Herrero (2011, 2012a, 2012b), regional identification is usually possible in both, the L1 and the L2.This supports the belief that L1 factors are more important than universals in L2 phonetic acquisition.Flege (1995) and Best (1995) also agree on the fact that the L1 shapes how the learner perceives the L2, giving more importance to L1 factors than to universals.
As explained in Herrero de Haro and Andión Herrero (2011), the interferences that speakers of L2 present in their speech can tell us not only the L1 of a speaker, but also which regional variety of L1 they speak.This way of pronouncing a given language clac 65/2016,  herrero: spanish pronunciation 172 with certain characteristics of a person's regional accent is what makes Welsh and English speakers of English have a different accent in Spanish.This is because learners access Spanish sounds through a filter which is particular to their first language, and more specifically, to their geolect.The Spanish accents of these speakers will present phonetic/phonological characteristics of their regional variety of English1 , showing a good example of the effects of language contact.Speakers of Welsh English present language transfer from their regional accent into Spanish and these are caused by the instability of plurilingualism and diglossia (Appel & Muysken 1987).Moreno, (2002, p. 25), had this in mind when he said that: "In the production of the L2, for example, we can observe an infinity of transferences from the L1, that will also depend on the language of origin and on its linguistic distance with the target language."Furthermore, the L2 can also influence on the L1, proving that the phenomenon of transfer is not unidirectional (Flege, 1987).The two ideas presented by Moreno and considered also in Conxita (1997) have great importance in the learning process; however, in this paper I will not deal with the linguistic distance between English and Spanish, but with the linguistic distance between a geolect of English (Welsh English) and Castilian Spanish.This prompts the question: how important are the linguistic differences between different regional varieties of English in the mastering of Spanish pronunciation?As Hansen Edwards and Zampini (2008) explained: "Major research findings have shown that predicting areas of difficulty and explaining L2 phonological acquisition is much more complex than a straightforward contrastive analysis of the first Language (L1) and the second (Lado 1957)." Regarding this, Ioup (2008) said that: "some studies established that learners do not substitute the exact L1 values for L2 sounds, but use a value somewhere between the two." All this means that we cannot anticipate what type of transfers will appear in the interlanguage of our speakers.Therefore, if we want to find out how important the linguistic differences between different regional varieties of English are in the mastering of the L2 we cannot simply compare the phonetic-phonological system of Welsh English and Spanish, on the contrary, we will need to analyse samples to identify areas of conflict and areas of transfer.
Taking this issue to the classroom, how is dialectical distance tackled in the teaching of pronunciation to L2 learners?We believe that the reference point has always been a learner with a general and standard form of L1, which represents an unrealistic and hypothetical point of view.If we study any materials aimed at teaching Spanish pronunciation to British students we will realise that they are targeted at RP speakers.
That sociolectal assumption is far from reality as only 3% of the United Kingdom speak RP (Hughes &Trudgill, 1988: 3) andTrudgill (2002).Why do all these materials assume that all British learners have a neutral and standard accent?What happens to the other 97% of the British students?Can all of them work with the same materials?It is obvious that attending this dialectical diversity is -to a greater or lesser degree-the teacher's responsibility.Some authors, such as Ioup (2008), andMajor (2008), have already illustrated how speakers of different geolects will present different positive and negative transfer; however, I believe that there is not sufficient literature to guide teachers on how to tackle these differences, with Herrero de Haro and Andión Herrero (2011, 2012a, 2012b) being the first attempts at addressing transfer between varieties of British and Irish English into Spanish.It is with all this in mind that I have carried out this study; firstly, to show how the dialectical region of the L1 (henceforth DR1), is of paramount importance in the acquisition of foreign language pronunciation, and secondly, to support teachers who teach Spanish as a foreign language (henceforth SFL) to speakers of Welsh English 2 , providing them with a catalogue of pronunciation errors that speakers of Welsh English usually present in Spanish, and suggesting some exercises to tackle these pronunciation problems.
2 I will follow the dialectical division of Great Britain proposed by Hughes and Trudgill (1988), where it is posited that the limits of Welsh English coincides, to an enormous degree, with the political boundaries of Wales.For the present study I have analysed 43 minutes of recordings in English and Spanish by three speakers from Wales.The recordings were transcribed by me, and checked by two Spanish teachers with English as their L1 and by a lecturer in Phonetics from a Spanish university.Given the amount of transfer found, I considered a perceptual analysis a better way of identifying areas of transfer for teachers to work with.However, I am hopeful that other researchers in the future will complement the data of this modest study with acoustic measurements for transfer on each Spanish vowel and consonant.
One of the samples was recorded using an Olympus voice recorder during an informal conversation, while the other two, due to logistic problems, were recorded by the participants at home talking about their interests and holidays; the purpose was to obtain continuous speech data in an informal register, which tends to be rare in language research (Zampini, 2008).The speakers were chosen due to their level of Spanish, and I only selected those with a level of at least C1 in the Common European Framework for Languages who were able to speak fluently and that showed a stable interlanguage.
Regarding the stability of their interlanguage, I analysed the speech of the participants and found that the positive and negative interferences were constant, that is, that these speakers maintained the same pronunciation of phonemes and allophones throughout the conversation.This requirement was important as they needed to have a settled and consistent pronunciation in Spanish and they also had to be able to hold a natural conversation during the recording.Furthermore, it was important for them to have a typical and recognisable Welsh accent in English to guarantee their suitability as Welsh participants.Two experienced language teachers with British English as their L1 and I listened to the recordings of the speakers, and after a detailed perceptive analysis we concluded that the accents of these participants were very clearly identifiable as Welsh.
It could be argued that a greater sample of participants should be needed for this study; however, the analysis of these recordings present consistent positive and negative phonic transfer from Welsh English into Spanish, and I believe that analysing the interlanguage of three Welsh English speakers with the same level of proficiency provides a starting point to show the importance of the RD1 in L2 learning.As Major (2008: 67)

Spanish as a foreign language in Wales
As explained in Herrero de Haro and Andión Herrero (2011), changes in education policies in the UK and various social factors have caused a rapid increase in students of Spanish at different educational levels.Various languages have been affected negatively by these measures, but it is German the one that has suffered the most, as many schools have either reduced the courses they offer or closed their German departments.
In Wales, we can see a similar trend to that in the rest of the UK.This tendency, as explained in Herrero de Haro andAndión Herrero (2012a, 2012b), is similar to the one found in other parts of the UK and in the Republic of Ireland.
However, in spite of similar language policies in Wales and in Ireland, the requirement in the latter for teachers to have a minimum level of Irish makes it harder for native Spanish speakers to enter its education system, hence the slower increase of students of Spanish in Irish schools.

The English language in Wales
According to the last census of 2001, Wales has a population of 2958 600 inhabitants, which is 4.9% of the UK population.The data obtained in the last UK census say that 24% of the Welsh population can understand Welsh; 21% can speak it; 20% can read it; and 18% can write it.However, only 16% declared themselves proficient in the four language skills.
The following graph shows the evolution of Welsh speakers from 18914 until 2001, where a change in language policy improved the figures of 1981 and 1991, where the historic minimum of 19% was reached4   Both English and Welsh have official status in Wales and both languages are used in education.Studying Welsh is compulsory until GCSE: over a quarter of youngsters go to schools where they are taught through the medium of Welsh and the rest go to schools where they study it as a second language in schools where the first language is English.The situation is different at University level: all Welsh universities teach some subjects in Welsh; however, the main language of instruction is English.clac 65/2016, 169-198 Fig. 3. Percentage of the Welsh population, divided by age groups over the age of 3, who are proficient in Welsh.
Welsh English 5 is characterised by a series of particular features, and being familiar with these will help teachers who work with Welsh students explain some of the interferences that these students present.I have summarised the main ones characteristics of Welsh English below; however, a thorough description can be found in Wells (1982): -Welsh English accents tend to be non-rhotic, meaning that the postvocalic continuant consonant r is only pronounced when it is followed by a vowel in speech.If this happens, the consonant marks the beginning of the next syllable, phenomenon known as linking r.This feature does not apply to some native speakers of Welsh or to speakers of Welsh English who are from the regions next to the rhotic regions of South West England.The trill consonant pronounced in Wales tends to be an alveolar tap.
-In the Southern cities, [t], [d] and [n] are alveolar, but these are dental in the rest of Wales.The lateral consonant tends to be alveolar in the South and velar in the North 6 Another characteristic of this regional accent is its tendency to introduce aspirations after voiceless stops, which also happens in many varieties of English.
5 Authors like Mees and Collins (1999) differentiate between three types of linguistic areas in Wales, depending on when Anglicisation took place, however, for our purposes, I will only consider general Welsh English features. 6Cardiff English is similar to English accents and has the allophonic distribution of alveolar/velar lateral consonant found in Southern English accents (Mees & Collins, 1999: 185).clac 65/2016, 169-198 herrero: spanish pronunciation 179 -The Welsh language does not have /tʃ/ or /dƷ/, which is why some speakers, especially in the North, tend to replace them with [ts].The voiceless velar fricative and the voiceless alveolar lateral fricative can be heard in words borrowed from Welsh.The phoneme /h/ can be omitted in careless pronunciation, although this does not occur in Welsh speaking areas.
-Welsh English accents tend to have an intonation characterised by a peculiar musicality caused, partly, by the lengthening of stressed vowels.
-In some Welsh English accents consonants are geminated when they appear between a stressed and an unstressed vowel, therefore, the word city can be pronounced [' sɪtti].
2.4.Similarities and differences between the phonic systems of Spanish, RP and Welsh

English
It is important to have a good understanding of the superposition and substitution of sounds between English and Spanish on one side, and between Spanish and Welsh English on the other, if we want to tackle the teaching of Spanish pronunciation to Welsh students in an effective way.Therefore, I will present a table with the sounds of Castilian Spanish and RP, and then the sounds of Welsh English.
Regarding the variety of Spanish taken as target model, I will maintain the 19 consonant phonemes and the allophones [β], [ð], [ɣ], and [ŋ] which, even though they will not deter the intelligibility of learners, they will help students reach a more natural Spanish pronunciation, free from the interferences of its own native language7 .Wells (1982) and Mees and Collins (1999). 10Following IPA conventions, in this paper the vibrant simple will be represented by /ɾ/, and the multiple by /r/.The most common continuant in Welsh English is [ɾ], however, [ r] and [ɹ] can also be heard in this geolect. 11 Mees and Collins (1999: 192), explain how [ʔ] is a prestigious rather than stigmatised feature in Cardiff English.Unlike other varieties of English, RP English does not use [ʔ] as an allophone of /t/. 12 Many features of Welsh English do not apply in Cardiff English (Mees & Collins, 1999: 187). 13According to Mees and Collins (1999: 191) Wells (1982) and Mees and Collins (1999).
Having compared the similarities and differences of RP and Welsh English with those of Castilian Spanish, we notice that the vocalic system of Spanish has a higher level of resemblance with Welsh English than with RP.Furthermore, we can also see the existence of a single flap continuant, a voiceless velar fricative, and an alveolar lateral consonant, which may cause positive transfer between Welsh English and Spanish.
In spite of having now a table illustrating the phonetic/phonological differences between RP English, Welsh English, and Castilian Spanish, we cannot guess what transfers will appear in the interlanguage of Welsh learners of Spanish, given the conflicting views expressed by the theories presented in the introduction and below.
The phonetic/phonological debate continues as different researchers disagree on whether some L1 sounds will cause positive or negative transfer depending on their proximity with a sound from the L2.Ioup (2008) and Zampini (2008) believe that proximity between features of the L1 and L2 phonic systems will result in positive transfer; however, other authors, such as Fledge and Hillenbrand (1987), believe the opposite.Some of the authors who believe that proximity between L1 and L2 phonic 14 Some authors, such as Mees and Collins (1999), consider that /ə/ is the vowel in ago and in bus, but I have decided to use different symbols, as Hughes and Trudgill (1987: 56), for greater clarity.
clac 65/2016, 169-198 herrero: spanish pronunciation 183 systems will create negative transfer are Strange and Shafer (2008); Best (1994); Major (2008); Ohala (2008); and Flege and Hillenbrand (1984).Flege and Hillenbrand (1984) explained this with the term equivalence classification, which Major (2008) explained as: "Equivalent or similar sounds are difficult to acquire because a speaker perceives them as equivalent to those in the L1; however, new (dissimilar or different) sounds are easier to acquire because there are salient differences.What his model implies (although not explicitly stated) is that transfer persists more for similar sounds than for dissimilar sounds." The tables above are useful to compare phonemes and allophones between Welsh English and Spanish; however, given the ongoing debate on transfer, the recordings will have to be analysed to see exactly what positive and negative transfer can be found in Welsh learners of Spanish.

Transferences between Welsh English and Spanish
As previously explained, there is no consensus on how transfer works or whether the proximity between L1 and L2 sounds will be an advantage or a disadvantage for the language learner, which is why I decided to analyse recordings and work with real data.
Following a perceptual analysis, I transcribed the recordings, and their accuracy was checked by two Spanish teachers with English as the L1 and by a lecturer in Phonetics from a Spanish university.
The analysis of the recordings 15 shows how speakers who share a common DR1 present the same positive and negative phonetic/phonological transfer when speaking in a foreign language, which illustrates the importance of considering the students' DR1 in language learning.Furthermore, the series of characteristic features of Welsh English which continue being present when the participants speak Spanish also enable us to 15 The transcriptions of the recordings show the actual pronunciation of the participants; G1, G2 and G3 refer to each participant and the number in brackets is the second is the recording where the word is pronounced.
clac 65/2016, 169-198 herrero: spanish pronunciation 184 place the DR1 of the participants, even when they speak in a foreign language.These examples of positive and negative transfer can be compared to the ones presented in Herrero de Haro and Andión Herrero (2011, 2012a, 2012) for other regions of the British Isles to see how transfer into Spanish varies depending on the speaker's DR1.. G1 is from the Welsh coast but she has been living in Norfolk for a few years.She is 34 years old and has a high sociocultural level, having completed a PhD in French literature.G1 works as a teacher of Spanish and French in an independent boarding school and has visited Spain on various occasions.
She speaks French and Spanish and her level of proficiency in them, according to the European framework, would be C2 and C1, respectively.G1's first language is English and she has an intermediate level in Welsh, although she has never used the latter at home with her family.

Second participant [G2]
G2 is from a town near Swansea, in the Southeast of Wales, a region where more than 60% of the population speak Welsh as their L1.He has been living in London for a few years, is 38 years old, and has a high sociocultural background.
G2 is bilingual English Welsh, although he considers Welsh his L1 and English his L2.
He goes to Mexico once a year and his level in Spanish (his L3) would correspond with a C2 in the European Framework.

Third participant [G3]
G3 is 42 years old and works as a lecturer at a Welsh University.She manages the department of Communication Sciences, Languages and Culture, and one of her research areas is the role of different languages on television in bilingual communities.
G3 has been to Spain on various occasions and has studied Basque, Galician and Catalan, although her level in these languages is not as high as it is in Spanish, where her level is C1.G3 is bilingual English Welsh and she uses the latter at home with her family.
clac 65/2016, 169-198 As previously explained, an analysis of these speaker's accent in English shows that, despite differences in their backgrounds, they all have a very clearly recognisable Welsh accent in English, thus, being a good sample for our study.The analysis of their accent in Spanish will highlight some of the interferences that Welsh students present in Spanish; however, we are aware that studies with larger numbers of participants will be beneficial in order to offer a more in-depth analysis of interferences in the interlanguage of Welsh students of Spanish.

Positive transferences of Welsh English into Spanish
The Spanish alveolar lateral is not reinterpreted.Most accents of Welsh English do not We also notice underdifferenciation of Spanish /r/ and /ɾ/, as even though we can find taps and trills in Welsh English, these are allophones of the same phoneme, unlike in Spanish, where both sounds have different phonemic values.This explains the fact that Some authors, such as Weinreich (1968), Major (2008), andIoup (2008), believe that underdifferentiation will be the hardest interference to master.Regarding this, Ioup (2008) pointed out that, "if the L2 contains a phonemic contrast in which both members are perceived as a single native language sounds, establishing different categories for the L2 will be extremely difficult", meaning that underdifferentiation will be the hardest interference to master.As Munro (2008)  As we have seen, there is no consensus on whether it is markedness or L1 phonology that is more relevant when learning the L2.Regarding markedness, Major (2008: 78) explains it by saying that "x is more marked than y if the presence of x implies the presence of y but not vice versa."Hansen Edwards and Zampini (2008: 4) say that "in the process of reranking the constraints from the L1 to the L2, the least marked structures' rerankings will emerge before those that are more marked."Regarding markedness, Eckman (2008) says that "A structure X is typologically marked relative to another structure, Y, (and Y is typologically unmarked relative to X) if every language that has X also has Y, but every language that has Y does not necessarily have X." Zampini ( 2008) also believes that "those areas of the L2 that are different from the L1 will be difficult to acquire if they are also more marked than the L1.Aspects of the L2 that are different but less marked will not be difficult to acquire."However, /b/ is less marked than /v/, as the former is much more common amongst world languages, in spite this, pronouncing /b/ in Spanish seems to be difficult for speakers of Welsh English when the phoneme is represented by the grapheme v, which is a negative transfer caused by the English spelling system.

Reinterpretation
Stressed vowels tend to be lengthened, /a/ and /o/ are also lengthened at the end of words, and /i/ is backed and pronounced When analysing the reasons for the aspirated pronunciations of the Spanish phonemes /t/, /p/, /k/ and /ʧ/, it will be worth remembering the concept of equivalence classification, which Flege would undoubtedly point out as the cause of these negative transfers.

Substitution
The phoneme /o/ is substituted by [əʊ] at the end of a word in several occasions, following the pattern of English: Novio We see how speakers of Welsh English struggle to pronounce [ ʝ], [ʎ], and [ɲ].These phonemes are substituted by the closest ones available in Welsh English, which would be an interference caused by the phonetic system of the students' L1.However, we could also explain these interferences using the concepts of markedness and universals: The Spanish consonants [ ʝ ], [ʎ], and [ ɲ] are more marked than [j], and [nj], which are the sounds that the Welsh English speakers use instead.Major (2008: 78) explains that "less marked phenomena are acquired before more marked phenomena".This would explain why these speakers have acquired other less marked sounds of Spanish (for example, most vowels) but not these consonants.Speaking about L1 acquisition, Ohala (2008) says that "in every given language, more frequent sounds are mastered earlier", which could be why these fairly infrequent Spanish sounds are hard to master for Spanish students from Wales.We could say that these substitutions are due to L1 transfer and to universals/level of markedness.Regarding this, Hansen Edwards and Zampini (2008) say that "when an interference is due to universals and L1 transfer the substitution will persist longer", therefore, we can anticipate that the problems with [ʝ], [ʎ], and [ɲ] will be very hard to overcome for students of Spanish who have Welsh as their DR1.

Discussion
4.1.Summary of interferences I will now proceed to mention the main interferences found consistently in Welsh learners of Spanish, as these will provide us with guidance about which sounds to focus on.The analysis of the recordings shows how the characteristics of the DR1 of the learners can be traced when they speak in a foreign language and how this has a direct effect on the Spanish accent of the learners.However, this has been ignored or simply not dealt with in detail in book sections aimed at teaching Spanish phonetics to learners of SFL.
A study of many of these materials has made obvious the fact that the general tendency is explaining the relationship between phonemes and spelling, without giving due attention to teaching how to pronounce the sounds of the language correctly.Some examples of this are Alonso and Everett (2008), Borobio (2006) andO'Connor (2002).
Equally, we can notice in these materials an exclusive concern with the Spanish sounds considered typically difficult for foreign students, ignoring not only the phoneticlinguistic distance between Spanish and the L1 of the speakers, but also the distance between the variety of Spanish used as target model and the regional variety of the L1 of the students.Therefore, I present here some exercises aimed at Welsh leaners of Spanish designed to correct the negative phonetic/phonological interferences that Welsh students present in Spanish, as shown in the analysis of the participant's accent in Spanish.
I have focused on the specific characteristics that speakers of Welsh English present and on the phonetic-linguistic distance between this variety of English and Castilian Spanish, although the principles used here can be used with speakers of other regional varieties of English.I have used these exercises in lesson and I invite teachers who work with Welsh students of Spanish to try these in class.
These exercises could be used as an independent lesson focused on pronunciation, or as part of a lesson dealing with Spanish phonetics.Students can also use these as individual exercises to practice and improve those aspects of Spanish pronunciation that they find most difficult.
I have included a list of Spanish words and phrases to practice Spanish This will make students aware of this phenomenon of Welsh English, but it will be necessary to remind students that in Spanish vibrant consonants are pronounced in all contexts.
The students could now proceed to read a list of Spanish words in which they will tend to drop the vibrant consonants, but they will need to make sure that they do not do this in Spanish.It would be convenient to read these words in pairs, so that students can tell their partners whether they are dropping the continuants or not:

Color por estar
The learners should start dropping the vibrant consonants less frequently now, but most likely they will still continue doing it at times.To avoid this, we could add a vowel after r, so that this now appears in a context where that consonant is always pronounced in their DR1.As Hansen Edwards (2008) explains, this use of a following vowel to avoid deletion is in line with the findings of Gatbonton (1978) and Major (1996).This will create sequences, not real Spanish words, such as *paroque, *amoro, *baroco, *coloro, *poro and *estaro.Once that the students are used to pronouncing the vibrant consonants in those words, they should start shortening the inserted vowel until they 17 The continuant will be pronounced in the first four words.In the next four, it will be omitted as it is not followed by a vowel.Finally, we have examples of linking r in the third group.
clac 65/2016, 169-198 herrero: spanish pronunciation 193 stop pronouncing it, making sure that they do not drop the vibrant consonant during this process.After that, the students can continue practising with more complex sequences.
It will be important to be aware of the phenomenon of linking r, which might still appear.The teacher must make sure that the vibrant consonant is linked to the previous vowel, not to the next one, preventing it from marking the beginning of the new word.
To do this, the teacher could ask students to insert a brief pause between the continuant and the vowel.

Gemination of consonants
The gemination of consonants between a stressed and an unstressed vowel is one of the most characteristic phenomena of Welsh English.To avoid this, we could show students how this works in their DR1 by comparing the consonants [t], [s] and [t ] in the following pairs of words:

Attack, city
Assess, passing

Church, switching
After showing the learners how gemination appears in the second word of each pair, we can present a list of Spanish words which will promote this phenomenon:

Casa coche mesa leche Poco antiguo noche folleto
To avoid this gemination of consonants, our students could start to pronounce the words separating the syllables (/ka sa/, /kotʃe/), and then proceed to shorten the pauses between syllables until they pronounce the words correctly without geminating the consonants.

Insertion of aspirations after voiceless stops
Voiceless stops are aspirated at the beginning of words, although [t] is also aspirated in other prevocalic contexts.
This interference appears in Spanish because of an allophonic rule of which students might not be aware.In English, intensity tends to be higher in voiceless stops before vowels than after vowels, and when these appear at the beginning of words their clac 65/2016, 169-198 herrero: spanish pronunciation 194 intensity is even higher, inserting an aspiration between the consonant and the vowel.
The students could pronounce the following English words and compare the pronunciation of [p], [t] and [k] in each case:

Park stop
To dirty

Car uncle
Voiceless stops are stronger when they appear at the beginning of a word, so we could introduce a syllable before this to control this tendency.It would be better to pronounce the following list of words in pairs so that the students can comment on their partners' pronunciation, but this can also be done independently:

Casa queso corazón
After that, we could start inserting the word mi 18 before each of those words, what will soften the pronunciation of [p], [t] and [k].This softer pronunciation of the consonants is the one that needs to be maintained in Spanish.Finally, I recommend continue pronouncing these words but removing mi and making sure that the soft pronunciation of [p], [t] and [k] is maintained.

Conclusions
This study has focused on teaching Spanish pronunciation to Welsh students in order to support teachers who work with these students.
An analysis of the interlanguage of our Welsh participants has shown that the students of SFL not only share positive and negative transfer with other speakers of their DR1, but also that their regional variety of English can be identified thanks to phoneticphonological transfer from their local accent into Spanish.
clac 65/2016, 169-198 herrero: spanish pronunciation 195 We have also noticed how the phonetic-linguistic distance between a regional variety of the L1 and the variety of L2 that we use as target model is a key factor which has a direct influence on the level of phonetic proficiency that students reach.
Furthermore, the analysis of the interlanguage of Welsh students speaking Spanish has highlighted some of the positive and negative transfer that the speakers of this regional variety of English present in Spanish.This has allowed us to identify the main phoneticphonological areas of difficulty for these speakers, thus, enabling us to develop some phonetic exercises that meet the specific needs of this particular group of students.
The results presented in this article prove the need to consider not only the L1 of the learners of a FL, but also the variety of the L1 that he/she speaks.I believe that the learning of SFL pronunciation, and of foreign languages in general, is a difficult task; however, approaches like the one presented in our study can make teachers and students aware of the importance of considering regional varieties of languages in the FL classroom and it will, hopefully, provide teachers with some ideas to offer further support to the FL student.
Finally, as previously explained, the perceptual analysis of Welsh students of Spanish is only a limited, but effective I believe, way to start analysing the interlanguage of these learners.However, I hope that other researchers will complement the modest analysis of this study with acoustic measurements for the pronunciation of each Spanish vowel and consonant by Welsh students.
the conditions on transfer remain a mystery", thus, the study of what sounds are transferred from Welsh English into Spanish might help us understand more about how transfer works, and offer some explanations until this matter is studied in more detail by other researchers.The recordings were transcribed phonetically using IPA symbols and I focused on those words containing interesting examples of language transfer from Welsh English into Castilian Spanish, (the variety taught as target model in the UK).I describe, mainly, negative transfer caused by the phonetic/phonological distance between the DR1 and the L2, although I will also talk about positive transfer caused by the affinity between Welsh English and Castilian Spanish.All the examples of transfer included in this article appear extremely frequently, with most of these examples being the only pronunciation registered for a specific Spanish sound by the participants.Likewise, all the examples are consistent throughout the recordings, as I have refrained from incorporating isolated examples.These interferences are likely to hinder the communication of the learners 3 .

Fig. 1 .
Fig. 1.Percentage of Welsh population over the age of 3 able to speak Welsh.

Fig. 2 .
Fig. 2. Percentage of Welsh speakers in different parts of the country.

Table 1 .
Number of total GCSE entries in Wales.Table adapted from the Welsh

Table 2 .
Number of total GCSE entries in Wales.Figures obtained from http://www.jcq.org.uk/national-results/gcses/Those figures show a decrease in student numbers of French and German, but an increase in the number of students taking GCSE Spanish.This means that between 1999 and 2012 student numbers doing GCSE in Wales decreased by 51.9% in French, and by 39.3% in German, but it increased by 48.4% in Spanish.

Table 3 .
Consonant phonemes and allophones, and glides in Castilian Spanish, RP and Welsh English 12 .Table based on the data presented in

Table 4 .
, /e:/ and /o:/ are pronounced as /ei/ /ou/, respectively, in Cardiff.Vowels in Castilian Spanish, RP and Welsh English.Table based on the data presented in Continuant consonants are dropped in postvocalic position when they are followed by a consonant.If they are followed by a vowel the continuant consonant is pronounced and linked to the next vowel, and it marks the beginning of the new syllable.It would be convenient to start by showing the student how this phenomenon works in his/her own variety of English, so we could start by providing a list of English words and by comparing how the grapheme r is treated in each case: