First Pan-American/Iberian Meeting on Acoustics, Cancun

[ Lay Language Paper Index | Press Room ]

Learning the Sounds of a New Language:
Adults Can Do It Too!

(Click on the flag for a Spanish version)

Paola Escudero -
Institute of Linguistics, OTS
Utrecht University
Trans 10
3512 JK Utrecht
The Netherlands

Popular version of paper 4pSC2
Presented Thursday afternoon, December 5, 2002
First Pan-American/Iberian Meeting on Acoustics, Cancun, Mexico

        Children normally manage to learn the language they hear around them. This task never proves too difficult for them. Adults, on the other hand, struggle to do the same thing. They often have an 'accent' when they speak a foreign language. Why should that be?

        The problem seems to lie in the fact that the sound system of our native language acts as a filter that prevents us from hearing or producing foreign language sounds faithfully. Given this "handicap" of an entrenched native language system, is there any hope for adult learners? This paper shows that adults can be successful in solving problems encountered while learning to perceive second language sounds. More specifically, it deals with the problems that emerge as a result of conflicting vowel systems. In other words, it tries to account for what happens when someones native language has "too few" or "too many" vowels in comparison to the language being learned.

        Whereas Spanish has only five vowels (si, que, la, no, tu), English has more than twice as many (heed, hid, hayed, head, had, who'd, hood, hoed, Hud, hawed, hod, heard). As a consequence, Spanish learners of English generally hear the English vowels in sheep and ship as instances of a single Spanish vowel, namely i. The reason is that the two English vowels differ in quality and length - the first one (ee) is both higher in terms of tongue position and longer in terms of duration than the second one (i) - and Spanish speakers are not attuned to making this distinction since it does not exist in their language.

        The question that was addressed in this study, then, was whether Spanish learners of English could perceive the two English vowels if the difference in quality and length were sufficiently large. To this end, we tested Spanish learners of English who had been exposed to two varieties of English, each of which emphasizes either vowel quality or vowel length in its differentiation of forms like sheep and ship.

        Thus, on the one hand, given the fact that Scottish English speakers pronounce these vowels with a prominent quality difference and a relatively small length difference (listen to Scottish English), it was hypothesized that Spanish listeners could potentially hear the Scottish vowels as two distinct sounds on the basis of quality alone. On the other hand, Southern British English speakers produce these vowels with a relatively large length difference and a rather small quality difference (listen to Southern British English) so it was surmised that even though Spanish listeners could potentially hear the quality difference, they would focus mainly on the length difference.

        The Spanish learners that were tested listened to ee (sheep) and i (ship) vowels with different combinations of quality and length values. These vowels were distributed along a matrix, which is represented by the square in the figure. The most sheep-like vowel (top right in the figure) was very high and very long, while the most ship-like vowel (bottom left in the figure) was much lower and shorter. The Spanish learners' performance was then compared to that of native Scottish and Southern British English listeners.

        The results of the vowel identification task are given below. In the figures, the line across the squares represents how the listeners identified the vowels: a horizontal line means a high vs. low distinction (quality alone), a vertical line means a short vs. long distinction (length alone), and a diagonal line means a quality and length distinction.

Scottish English

  • Native listeners focus on quality differences almost exclusively when identifying the vowels, which is characteristic of their variety of English.
  • Inexperienced Spanish learners of this variety can identify the two vowels by their quality difference.
  • Experienced Spanish learners, just like the natives, identify the quality difference and ignore length.

Southern British English

  • Native listeners identify the vowels by focusing on both quality and length, despite the greater difference in length in their variety.
  • Inexperienced Spanish learners of this variety identify the two vowels by their quality difference.
  • Most experienced Spanish learners identify the vowels by focusing exclusively on the length difference and ignoring quality; other experienced Spanish learners, however, identify the vowels by focusing on both quality and length, just as the natives do.

        In sum, then, the quality difference in Scottish English is large enough for Spanish learners: they have no problem identifying the sheep and ship vowels. The learners of Southern British English also identify two vowels but, unlike the natives who combine quality and length, they only focus on the difference in length. The reason seems to be that to Spanish ears, the length difference between the two vowels is more salient than the quality difference. As a result, they start out by relying on that distinction alone. Eventually, however, these listeners can reach the stage where the small quality difference can be perceived, as shown by the performance of some experienced learners. In other words, some experienced learners resemble native performance by combining quality and length.

        Perhaps surprisingly, having "too many" vowels can also be a problem. For instance, Dutch, just like English, has twice as many vowels as Spanish, so that one might be led to believe that English or Dutch speakers should have no trouble mastering them. However, vowel identification tasks have showed that Dutch learners have problems identifying Spanish i (si) and e (que). They listened to the vowels of Spanish syllables produced in a phrase (listen to a Spanish phrase) and were asked to select which of the 12 vowels in their native language they thought they heard. In the particular case of the two Spanish vowels i and e, inexperienced learners heard three vowels. In other words, some instances of both Spanish i and e were identified as the Dutch equivalent of the English ship vowel (see A in the figure). This would not be a problem, if the listeners could manage to assign them to the correct Spanish i and e categories in the way native listeners do. However, inexperienced learners of Spanish cannot do that, as was shown by the results of a Spanish-to-Spanish vowel identification task in which the learners had to identify the vowels in the syllables by selecting from the 5 Spanish vowels. These learners misidentified half of the Spanish e vowels, as shown in the B part of the figure.

        Can this problem be resolved? The experienced learners in the figure below can hardly be differentiated from the Spanish listeners. They have learned to disregard their ship-like vowel when they listen to Spanish and, as a result, they can correctly assign the Spanish vowels.

        Summarily, then, adult language learners can run into conflicts when they start with "too few" sounds as well as when they start with "too many" sounds. Crucially, they solve both types of problems as they gain experience in the second language. The strategies that adult learners use could be similar to those that children use when learning their first language. For example, research has shown that infants use only one dimension when identifying visual or language categories, as most experienced adult learners of Southern British English have been shown to do with the sheep and ship vowels. Further research should allow us to determine in what specific ways adult strategies differ from or resemble the strategies used by first language learners.


[Lay Language Paper Index | Press Room ]