The basics: Interlanguage and its components

Non-native speakers of languages (especially if their first exposure to the target language happens beyond what is referred to as a sensitive or critical period¹⁴ in the literature of language acquisition), will inevitably produce errors of various types (i.e., at all levels of grammar) in their L2/FL speech. The intermediate variety spoken by non-native speakers of a given language (i.e., non-native speakers’ idiolects that deviate both from the L1 and the L2/FL, and are thus somewhere in between) has been called interlanguage since Selinker (1972) introduced the term. In what follows, we will be concerned with the peculiarities of the pronunciation aspect of interlanguage by casting light on some of those language-independent features of foreign accent that will be the most relevant to our discussion later.

Figure 2.1: Major’s (2001: 6) model of the components of interlanguage

According to Major’s (2001) famous model (see Figure 2.1), interlanguage (abbreviated to IL in the figure) has the following three¹⁵ components:

14 See more on the age factor in Section 4.1.

15 Major’s model (and thus the discussion here too) is not concerned with the possibility of further languages (other FL’s or an L3, L4, etc.) potentially affecting a particular interlanguage variety. This issue will not be taken into account throughout the thesis either, mainly because an average Hungarian learner rarely speaks multiple foreign

1. “Parts of L1” can be considered the most important of all the three components: the sum of the features belonging to this category is what makes a particular foreign accent recognisable: for example, a speaker who mixes up /l/ and /r/ and applies extensive vowel epenthesis in their English speech is recognised as being Japanese; pronouncing an epenthetic vowel before word-initial sC clusters is typical of Spanish speakers; gemination of word-final single consonants in monosyllabic English words (accompanied by schwa epenthesis) is a characteristic feature of Italian-accented English; and so on. In other words, this category constitutes those pronunciation errors that are rooted in the phonetic and phonological differences between the L1 and the L2/FL, that is, when a non-native form is pronounced because a particular sound or a phonological rule is transferred from the L1 onto the L2/FL (this is called negative transfer or interference).

In the contact of a Hungarian L1 and English L2/FL, word pairs such as vet and wet or bed and bad pronounced the same, or finished ending in /ʒd/ are examples of a typical Hungarian learner’s interlanguage: the former example illustrates when a target language sound that does not exist in Hungarian is replaced by one that exists in the L1¹⁶ (resulting in the target language minimal pair becoming Hunglish homophones), and the second one exemplifies the case where all segments in question exist in both languages, but it is due to a phonological rule (assimilation) as well as the role of spelling that is responsible for the pronunciation error.¹⁷

The phenomenon of L1 transfer was in the centre of attention when extensive work on Contrastive Analysis (a study field focussing on comparing and contrasting languages, often abbreviated as CA) was carried out during the 1960s. Advocates of a CA approach (Lado 1957 in particular, but cf. also Weinreich 1953, Haugen 1956, Moulton 1962, Lado 1964, Stockwell

& Bowen 1965, Brière 1966, Brière 1968) held the view that all potential learner errors are due to L1 transfer, and thus any error can be predicted as well as explained based on the differences in the sound systems of the languages involved. The view received widespread criticism on

languages (those learners in whose case this aspect is worth considering are a privileged minority), therefore the issue is too special to be worth discussing. (The participants of the case studies presented in Chapter 5 do not speak any other foreign languages – apart from English – which may have influenced their pronunciations of English.)

16 This claim is not as true of bed and bad as of vet and wet, because it is not obvious that the vowel of bad does not exist in Hungarian (it is not entirely clear what the vowel of bad is in English, or what the vowel of E is in Hungarian), but what is important (and relevant) here is the absence of the contrast between the two sounds in question (and thus word pairs like the ones in question) in Hunglish.

17 See Chapter 3 for a systematic and thorough overview of the potential Hunglish pronunciation errors stemming from L1 transfer.

various grounds and got redefined as a consequence of the concerns voiced, but it became evident after a while that there exist learner errors that cannot be attributed to L1 transfer (cf.

the discussion on the third component of interlanguage below). With this realisation the fundamental claim of CA was eventually dismissed.

2. “Parts of L2” refer to native-like pronunciation features, which may have two different sources: First, they may be the result of positive transfer, i.e., when a target language form is not different from its equivalent in the L1. In the case of Hungarian learners of English, for example, the pronunciation of the sound [ʃ] is highly unlikely to cause any problems, as neither the articulation nor the distribution of this consonant differs significantly in the two languages (the sound exists in the inventory of both languages; there is no phonetic difference between an English and a Hungarian [ʃ]; the sound has no allophonic variants in either language; and there are no major differences in its distribution¹⁸ in the two languages either that would cause difficulties for a Hungarian speaker of English).

Second, native-like elements may also occur in an individual’s accent as a result of successful acquisition of the pronunciation feature: for instance, a Hungarian learner’s English accent which features the correct pronunciation of the dental fricatives is only possible if the learner has learnt how to pronounce these sounds¹⁹, as they do not exist in Hungarian.

3. “Parts of U” stand for universals and they refer to examples when a learner’s pronunciation error is not an example of L1 transfer, but it is a result of linguistic universals, that is, a part of an innate Universal Grammar (Chomsky 1986). Although it is certainly possible that an error can be considered totally idiosyncratic,²⁰ many errors unattributable to L1 transfer are explicable (and may even be predictable to some extent) on the grounds that learners with a variety of different L1’s make the same types of errors in the same L2/FL and that the same errors are made by children in the course of L1 acquisition (Major 2001: 3). This supports the innateness hypothesis, that is, the idea that children are born with an innate Universal Grammar.

18 There are differences in the distribution of [ʃ] in English and in Hungarian in terms of its occurrence in consonant clusters: for example, word-initial clusters like [ʃp-], [ʃt-], [ʃk-], [ʃn-], [ʃl-], etc., which are well-formed in Hungarian, are not attested in English (except in a few foreign words such as spiel and schnapps). Also, English [ʃ] is less common before plosives in general than Hungarian [ʃ]: for instance, a word ending in [ʃt] in English is a past tense verb form, since this cluster is not available morpheme internally, but it is in Hungarian (e.g., füst [fyʃt]).

However, these differences do not cause difficulties with English pronunciation for a Hungarian learner.

19 Or if the learner lisps, but this is not strictly relevant here.

20 Some of Altenberg & Vago’s (1983: 437–438) observations of Hunglish are grouped into such a category.

A notion closely related to universals is markedness – often what is meant by universals is equated solely with markedness-related phenomena (cf. the following section, viz. Section 2.2).²¹ In a broader sense of the term (cf., e.g., Major 2001: 41), universals subsume a number of further (mostly non-phonological) issues, such as overgeneralisations and hypercorrections (the two are not unrelated), as well as sociolinguistic issues like intra-speaker variation depending on text category (i.e., the fact that the proportion of pronunciation errors is significantly higher in free conversation than when reading out a word list), to mention just a few examples that will be mentioned later, especially in the discussion of the empirical data presented in Chapter 5.

It is the combination of interlanguage components belonging to these three categories that creates the various idiolects of interlanguage, and it is the extent to which certain error types and native-like forms are represented in a variety that will be different in each individual speaker’s foreign accent. The extent depends mostly on language-external factors, which are to be discussed in Chapter 4.

A final remark relevant to the components of interlanguage is that there is a significant overlap between L1 and U as well as L2 and U (as an error or a native-like feature might be universal as well), but the most peculiar case is when a seemingly inexplicable pronunciation feature is found in a learner’s accent, that is, one that is part of neither the L1 nor the L2/FL.

The next section delves more deeply into how markedness is able to account for such examples.