Results and discussion

We will begin the presentation of the results of our experiment with an evaluation of the stress perception and production tests, and the musicality test results will be presented afterwards.

In the stress perception test, the participants scored a weaker overall result (especially when compared to the results of the stress production test – see below). The average point they

scored is 9.4 points out of the maximum of 16, which is a result below 60% (see the distribution of the results in Figure 5.14).

Figure 5.14: The distribution of the results of the stress perception test (max. 16 points) What will require some explanation is that there is no consistency whatsoever in the errors the learners made: there were not any participants who marked word-initial stresses in the majority of the words (which could have been expected in at least some learners’ case, given Hungarians’

high chance of suffering from stress deafness), nor was it found that they would mark non-word-initial stresses in heavy syllables in otherwise initial-stressed words. It is a further observation that the participants generally did not make use of the last answer option (the one saying “I hear all syllables equally stressed”), which suggests that even if they did not hear a difference in the stress degrees in the words, they resorted to guessing the answer rather than choosing the last option, which they may have suspected that was not the “correct answer”. The lack of consistency in their answers confirms the assumption that in many cases the learners did guess the answers.

On the other hand, learners did exceptionally well in the stress production test: in 78% of cases the learners pronounced the stress patterns of the nonsense words correctly (cf. Figure 5.15).

Figure 5.15: The proportion of correct and incorrect answers given in the stress production test

If we take a closer look at the 22% of words that were pronounced incorrectly, we may observe that contrary to our expectations, the majority of the errors (76% of them to be precise, cf.

Figure 5.16) were hypercorrections: with the exception of one single participant (whom we will discuss later), incorrect pronunciations occurred in words in which stress happened to fall on the first syllable, but the learners placed stress somewhere else in their pronunciations.

Figure 5.16: The proportion of the sources of incorrect answers given in the stress production test

Such occurrences happened in heavy syllables, especially in syllables containing long vowels:

e.g., many participants pronounced the words ínnace and íffating with stress falling on the second syllable (even though these two words were stressed on the first syllable in the sound

bank), most probably because of the diphthong /eɪ/ in the second syllable. In contrast, the stress pattern of words where a stressed syllable did contain a long vowel (e.g., enígn and innícing) were copied correctly by almost all participants.

Though to a smaller extent, the tendency to pronounce non-initial unstressed syllables as stressed was also observed in closed syllables (e.g., in words like ímmincing, which half of the informants pronounced incorrectly with stress on the 2nd syllable) and even word-final (C)VC syllables (e.g., in ínnick), which – in English – do not happen to count as closed (and thus heavy).

To gain a deeper understanding of such hypercorrect forms, let us look more closely at and compare the pronunciations of the example words ínnace (cf. Figure 5.17) and ínnick (cf.

Figure 5.18).

Figure 5.17: The proportion of correct and incorrect placements of stress in the word ínnace

Figure 5.18: The proportion of correct and incorrect placements of stress in the word ínnick

As the figures show, the word ínnace (stressed on the first syllable, but having a diphthong in the second) was only pronounced correctly by 54% of the participants – the remaining 46% of them stressed the word on the second syllable, mostly due to the fact that they mistook stress for vowel length. In contrast, the word enígn (which was stressed on the second syllable) was pronounced correctly by all of the participants, which further supports the claim that the learners associated stress with vowel length: they could easily assign stress to syllables containing a long vowel, and many of them had difficulty pronouncing initial stress if there was a long vowel in a syllable not carrying major stress.

Interestingly, the other word in question (ínnick, stressed on the first syllable) was also mispronounced by some of the participants. This suggests that the learners may have subconsciously acquired some parts of the weight-sensitive stress rules of English, namely that (C)VC syllables often count as heavy and attract stress, but they overgeneralised the rule and applied it in cases where a syllable-final consonant is extrametrical and does not make the syllable heavy. However, more data would be necessary to support this claim.

As for the participant who systematically pronounced the nonsense words with stress falling on the first syllable, this informant was in the group of beginners and he was the only one in whose pronunciation the effect of L1 transfer manifested itself. The other participants (even the beginners) performed so well that the case requires some explanation. The reason behind the exceptionally good results may lie in the use of nonsense words: in the case of non-existent (or unknown) words one is expected to pay more careful attention to their pronunciation and has a better chance of repeating the word correctly. This phenomenon is known as the

“lexical bias effect” (often abbreviated to LBE), which is mostly discussed in the field of sound substitutions (i.e., when target language sound segments not found in the inventory of the L1 are substituted by an L1 sound): what has been observed (cf., e.g., Costa et al. 2006) is that in non-existent words speakers produce fewer sound substitutions and more sounds that are target-like or at least close to the target. This is the effect that apparently operates in other fields similarly (in our case in the field of non-native word stress production), and this is what is likely to have caused the unexpectedly high results.

In terms of musicality, let us first see how our participants can be assigned into the categories determined by the creator of the musicality tests. In the three tables to be presented (Tables 5.11–13), the three groups of learners are considered separately in three different columns. Let us see the results of the tone deafness test:

1. 2. 3.

91–100% exceptional – – –

81–90% very good performance 1 1 1

71–80% normal performance 1 1 4

61–70% low-normal performance 6 7 2

55–60% low performance(?) 1 – –

0–54% possible pitch perception or memory deficit 1 2 – Table 5.11: The results of the tone deafness test

The results of the rhythm test are as follows:

1. 2. 3.

91–100% world-class performance – – –

81–90% outstanding performance 1 1 1

71–80% very good performance 5 1 2

61–70% low-normal performance 3 8 3

56–60% low performance(?) – – –

0–55% possible rhythm perception or memory deficit 1 1 1 Table 5.12: The results of the rhythm test

Finally, the results of the pitch test:

1. 2. 3.

0–0.74 Hz exceptional ear – – –

0.75–1.4 Hz very good – – –

1.5–5.9 Hz normal 2 6 2

6–11.9 Hz low-normal 6 1 3

12–15.9 Hz low(?) – – –

above 16 Hz possible pitch perception deficit 2 4 2 Table 5.13: The results of the pitch test

As for the three components of musicality examined, the results show that none of the participants produced a result belonging to the topmost category in any of the three tests; the majority delivered a “low-normal” performance, although there were a few good or even

outstanding results especially in the tone deafness test and in the rhythm test. Relatively few learners scored a result belonging to the lowest category, and it is worth noting that the few instances of the lowest scores in the three test components do not belong to the same participants – if one of them was categorised as potentially having amusia in the first test, the same person did reasonably well in the other two tests, and this is true for the lowest scorers in the other two tests, too. This observation clearly supports the idea that the different components of musical talent do need to be differentiated and separately considered.

As no connection was found between the results of the stress perception test and the musicality tests, these results will not be presented. It is a lot more intriguing to compare the results of the stress production tests and the musicality tests. We calculated the degree of correlation between the results of the stress perception test and each of the three musicality tests; the results are summarised in Table 5.14:

correlation coefficient (Pearson’s r)

stress production – tone deafness 0.03

stress production – sense of rhythm 0.41 stress production – pitch perception 0.09⁷⁴

Table 5.14: Correlation between stress production and the components of musicality As the coefficients show, tone deafness and pitch perception are not correlated with stress production, however, there is moderate correlation between the rhythm component of musicality and stress production. If we look at the correlation between sense of rhythm and stress production separately in the three groups of learners, we can notice that correlation is higher in the case of the elementary group, and it even counts as strong correlation.

74 While in the pitch and tone tests higher numbers indicate a better performance, in the rhythm test a lower number is better (i.e., the lower the result of the test is in Hz, the more sensitive is the participant’s ear to pitch). Therefore, results scored in the pitch test were multiplied with –1 when calculating the correlation coefficients so that positive or negative correlation will be displayed consistently in all rows of the table summarising correlation coefficients.

the degree of correlation between stress production and sense of rhythm

(Pearson’s r)

Group 1 (elementary) 0.61

Group 2 (intermediate) 0.48

Group 3 (advanced) 0.33

Table 5.15: Correlation between stress production and sense of rhythm in the three groups What this means is that although in English there are significant differences in pitch between different stress degrees, it is not pitch by which learners (at least the ones participating in this study) identify the degrees of stress in English, but they rather perceive the characteristic rhythm of the language.