Phonological Analysis of a Second Language

Introduction

The purpose of this project is to analyze the English spoken by a non-native. My consultant for this project was a thirty year old man from Ecuador. His first language is Spanish, and in addition to knowing English, reads Koiné Greek and knows a little Quechua. His first exposure to learning English was being taught in high school when he was about fifteen. He moved to the United States when he was twenty three and has lived in the South for the nine years since then.

I met with my consultant on November 22nd, 2003. During the meeting he gave me the preceding information and also read a one page story that I wrote specifically for the interview. In the story, I tried to use as many words and phrases with which I predicted he would have the most difficulty pronouncing. While he read the story in a quiet room, I recorded his speech using an iRiver iFP-380T which encoded the speech sample in MP3 with a sampling rate of 44.1 kHz and a bitrate of 96 kbps. With this recording, I was able to make a narrow transcription of his speech. I wrote the transcription using solely symbols from the standard International Phonetic Alphabet (as found on the chart). This document contains many of these IPA Unicode characters and, in order to be viewed properly, you must have a Unicode font that contains IPA symbols. I recommend the SIL Unicode IPA font.

In the following section, I will make some predictions about his speech based upon my knowledge and research of Spanish, his native language.

Research and Prediction

From my personal knowledge of and research on the Spanish language, I found that American English has a larger phoneme inventory than Latin American Spanish. Both have roughly the same inventory of consonants, but there are some differences. English has a few more such as [v θ ð z ʃ ʒ ɹ], and Spanish has [ɲ]. Although there is no [v] in Spanish, its orthography seems to indicate otherwise; nevertheless, the Spanish letters <b v> are pronounced exactly the same. The remaining consonants in these languages sound alike; these are the corresponding sounds [h] in English and [x] in Spanish, [dʒ] in English and [ɟ] in Spanish, and the slightly less similar sounds [ɹ] in English and [ɾ r] in Spanish.  The alveolar stops in English are dental in Spanish. Stops in Spanish are “softer” than those in English; there are no aspirated stops, and voiced plosives become fricatives when between vowels (Conde).

Spanish has a simple five vowel system [a e i o u]; diphthongs are formed by combining the weak vowels [i u] with the strong vowels [a e o] or another weak vowel; triphthongs are formed by combining a weak vowel, a strong vowel, and another weak vowel (Conde).  The vowel system of American English is much more diverse than that of Spanish. Whereas the Spanish system has five vowel anchors, the English system has more than twice that number. The five English vowels that roughly correspond to the Spanish vowels are the tense vowels [ɑː eɪ iː oʊ uː]; the vowels that do not correspond to the Spanish system include the tense vowels [ɔː aɪ aʊ ɔɪ ju], the lax vowels [ɪ ɛ æ ʊ ʌ], and the schwa [ə] (Ladefoged). Needless to say, this more complicated system presents difficulties for Spanish speakers because they make no distinction between tense and lax vowels in their language, and the vowels that they do produce are clear, never turned into schwas or diphthongs.

Syllables in English are also more complex than those found in Spanish. One of the most complex syllables in English, “strengths”, has a structure of CCCVCCC (Spencer). This is impossible in Spanish for it allows a maximum syllable size of CCVCC, for example in the syllable “trans-”. The maximum sized onset may contain a plosive followed by a fricative or a glide. Therefore, the common English consonant cluster of [s] with a following consonant is not permitted in Spanish (Conde). Because of this, the English words special, stress, and ski are not possible and thus [s] must become part of a separate syllable; these words in Spanish are especial, estrés, and esquí respectively. In the coda, the only consonant cluster allowed in Spanish is [ns].

With these comparisons between the vowels, consonants, and syllable structures of English and Spanish, I make the following predictions of how a native Spanish speaker will pronounce English words:

1. [h] may be pronounced as [x], and [ɹ] may be pronounced as [ɾ r]. Alveolar stops and [θ ð] may become dental stops.
   
2. Between vowels, [b d g] may become [β ð ɣ] respectively.
3. It may be difficult for the speaker to distinguish between tense and lax vowels; he may also skip reducing vowels into schwas.
   
4. Complex consonant clusters may be simplified, especially in the coda.
   
5. Onsets that have clusters beginning with [s] followed by a stop may be broken up by epenthesis.
   

Speech Sample

The audio recording of this sample is available.

Analysis

I was impressed with my consultant's command of the English language. His fifteen years of study have paid off and given him excellent English pronunciation. However, he did have a notable Spanish accent with peculiarities that I expected and predicted, for the most part. I will now go through my predictions and give examples for each from the speech sample to verify them.

1. My consultant did not have any difficulty with [ɹ], which surprised me. He did, however, substitute [h] for [x] in a few places. On line 9 he pronounced the word “happy” as [ˈxæpiː]; on line 19 he pronounced “huge” as [xjudʒ]; and on line 27 he pronounced “have” as [xæv]. This substitution was sporadic. For the majority of instances of /h/, he did not make the substitution (e.g. line 26 “hard” as [hɑːɹd]). In addition to this substitution, my consultant's alveolar stops did frequently become dental (as far as I could determine). Examples of this are on line 16 where he pronounced “during” as [ˈd̪uːɹɪ̃ŋ], on line 21 “dad” as [d̪æd̚], and on line 31 “day” as [d̪eɪ]. Nevertheless, this was difficult to perceive so I am uncertain of how wide spread it was. However, I am not uncertain about my consultant's substitution of [d̪] for [θ] and [ð]; this happened very frequently and can be observed throughout his speech. For a couple examples, on line 16 “that's” was pronounced as [d̪æts] and on line 21 “the” as [d̪ə]. This substitution was not universal, for he did pronounce [θ ð] often as well (e.g. on line 1 “things” and line 7 “that”). Here are a few rules that he applied sporadically.

/h/ → [x]
/d/ → [d̪]
/θ/ → [d̪]

2. My prediction that [b d g] would become [β ð ɣ] respectively between vowels, did not happen as I thought. The lenition happened just three times and only the lenition of [b] into [β]. In the three cases, /b/ was preceded by a vowel and followed, although not by a vowel, by an approximant or a lateral. On line 13, he pronounced “able” as [aɪβl] and on lines 17 and 21 he pronounced “brother” as [ˈβɹʌðɚ].  He evidently has almost completely removed the following Spanish rule from his phonology of English:

/b/ → [β] / V_V

3. I was impressed with my consultant's correct usage of English tense and lax vowels. I purposefully included the words “sheep”, “ship”, and “sheet” because they depend on a tense/lax vowel to distinguish meaning. For “sheep”, on line 14, he pronounced it incorrectly as [ʃɪp̚]. But for “ship” and “sheet” on lines 18 and 19 he pronounced them correctly as [ʃɪp] and [ʃiːt̚]. Additionally, as I predicted, he seemed to skip reducing unstressed vowels into schwas, especially in the word “to” which he usually pronounced [tu]. Examples are found on lines 1, 9, and 31. Examples where he did apply the rule are found on lines 19, 20, and 33. Note that, again, he performs the reduction sporadically.

4. My prediction of the deletion of consonants in complex clusters did not happen. The most difficult consonant clusters were in the word “strengths” on line 5 which he pronounced [əsˈtɹɛ̃n̪θs]. No deletion occurred, but only the epenthetic [ə] and the substitution of [ŋ] for [n̪] since it is homorganic with the following consonant [̪θ].

5. He did use epenthesis to help him pronounce complex consonant clusters, as I predicted. On line 3 he pronounced “skiing” as [əsˈkiː̃ŋ], on line 4 “stresses” as [əsˈtɹɛs], and on line 7 “spite” as [ĕsˈpʰaɪd]; in each of these cases he generated an epenthetic vowel to move [s] into the coda of the newly prepended syllable. As the other rules were not universal, this rule was not either. In other instances he did not use epenthesis, but was able to pronounce the consonant clusters perfectly. On line 2 he correctly pronounced “skier” as [skiːɹ], on line 24 “stop” as [stɑːp], and on line 13 “spring” as [spɹɪ̃ŋ]. The following rule was applied sporadically:

Conclusion

Overall my consultant's speech was superior. Although his native language did influence his pronunciation of English, this influence was minimal and not consistent. He had perfect pronunciation most of the time, and seldom spoke with a thick Spanish accent. Less proficient speakers tend to speak with very heavy accents and constantly apply the rules that I presented in my analysis. He, however, applied the rules sporadically. This leads me to conclude that the influence of the phonological rules from the L1 diminish over time as they are replaced with the rules from the L2.

Bibliography