Gairaigo and Language Acquisition

Chapter 3

Resolving the Paradox of Cognates

Many languages are linked by the cognate pairs they share sets of words that learners perceive to be similar. Japan’s massive lexical borrowing, for instance, has created many such intersections between Japanese and English. However, loanwords (e.g. apuroochi) and their corresponding borrowed words (e.g. approach) are seldom identical in form or meaning. Therefore, the effect of the mother language (i.e. L1) on the learning of a target language (i.e. L2 or TL) is multifaceted, and negative and positive affects can be observed. This has led to two antithetical views: that cognates are a great pitfall for learners; and that cognates greatly assist acquisition. Particularly in the case of gairaigo, both points of view are abundantly represented in the literature.

This chapter will attempt to resolve the ‘paradox of cognates’ (see Carroll, 1992: 94). First, we will review some basic concepts in language learning: previous knowledge, language transfer and cognate relationships. Next, the semantic and phonological ‘pitfalls of cognates’ will be discussed, and then contrasted with empirical studies. To resolve the contradictions that arise, the theoretical foundations for the misunder- standing of learner errors will be presented, as well as the research flaws that have contributed to an unbalanced view of cognates. These theoretical foundations will enable us to examine the particular case of English loanwords in Japanese in Chapter 4.

Basic Concepts

To understand the effects of cognates (including loanwords), it is necessary to put some basic concepts into perspective: the role of previous knowledge in learning; what lexical transfer is; and what cognates are.

The role of previous knowledge

‘What we know is how we learn’ summarises the fundamental role that previous knowledge plays in all learning; it is a process of finding the best way of mapping new learning onto old (e.g. Baddeley, 1990).

With any learning task, in a process known as transfer, we optimise our 43

effort by making as much use of previous knowledge as possible.

Imagine snowboarding for the first time anxiously gazing down a snowy slope, we instinctively grasp for whatever previous knowledge is available, be it from skiing, skateboarding or slipping on an icy sidewalk.

Before we begin, we already know stopping will be difficult, falling down will hurt and which parts of our body are vulnerable. Such knowledge is precious, and although we may wipe out down the slope, we have done far better than would a typical nomad from the Serengeti.

Our previous knowledge will serve us in many ways, and lead us astray in others. When boiling soba (buckwheat) noodles for the first time, how we prepare pasta can be our mental template. We will already know to preheat the water to a boil, open the package, measure out an appropriate amount of noodles and dump it in. Yet our pasta knowledge may lead us to overcook the soba, and the salt we add to the water is unnecessary.

When I attempted to play theshamisen, I drew heavily from my closest available knowledge guitars. I experienced positive transfer from guitar techniques such as slides and tremolo picking, putting me on a par with advanced shamisen players. However, I also experiencednegative transfer (also known asinterference), and my venerable teacher did not appreciate radical electric guitar techniques such as fret board finger tapping ala Eddie Van Halen.

In general, the process of drawing upon one’s L1 knowledge in learning an L2 is known as language transfer, sometimes referred to as cross-linguistic influence (Kellerman & Sharwood-Smith, 1986), which is the many different ways in which one language may influence (the learning of) another. While infants can learn multiple languages effort- lessly, older learners must largely resort to an analytical approach, and the L1 becomes the language paradigm to be repeatedly referred to.

Learners do not need to build a conceptual framework totally separate from their first language; rather they can add to and integrate their first language system with the second language.

(Hammer & Giauque, 1989: 37)

Thus a learner’s L1 is fundamental in determining the successes, errors and speed of acquisition.

Language transfer affects various aspects of language learning, including: grammar, vocabulary, pronunciation, spelling and meaning (i.e. semantics); for speakers of a Subject Object Verb (SOV) language (e.g.

Japanese) learning a Subject Verb Object (SVO) language (e.g. English), the effect of L1 grammar may be the least facilitative. In contrast to SVO

44 Part 2: Gairaigo and Language Acquisition

languages, SOV languages have a strong tendency to use postpositions rather than prepositions, to place auxiliary verbs after the action verb, to place genitive noun phrases before the possessed noun, to place a name before a title or honorific, and to have subordinators appear at the end of subordinate clauses. Relative clauses preceding the nouns to which they refer usually signal SOV word order. Some SOV languages have special particles to distinguish the subject and the object, such as the Japanesega and o

¯. SOV languages also seem to exhibit a tendency towards using a Time-Manner-Place ordering of prepositional phrases.

Lexical transfer

The several types of L1 knowledge and processing abilities that persistently affect the L2 learning process includelexical transfer, through which the L1 lexicon directly affects L2 vocabulary acquisition. Lexical transfer notably includes the knowledge of L1 phonology, lexical semantics and morphology. Learners tend to employ lexical transfer whenever possible. When faced with the task of comprehending a new language, learners at especially early stages tend to associate new words with primary counterparts (Arabski’s 1979 term).

In comprehension, or reception, lexical transfer affects learners’

perceptions of input over time as they develop their interlanguages the L2 system developing in their brains. Phonetic and orthographic decoding are the key processes involved. Comprehension focuses on decoding of contextual meaning while structural details remain in the background; grammaticality and acceptability are typically nonissues.

Learners’ search for cognates is facilitated by formal similarity; if such similarity can be perceived, transfer is likely.

In production, lexical transfer affects learners’ retrieval of knowledge when they face linguistic knowledge gaps that simple avoidance cannot aid; a learner may attempt to communicate with a word drawn from L1, or find it easier to recall an L2 word that is cognate with a native item.

Grammaticality and acceptability become quite important as learners strive for communicative competence and acceptance into the foreign language culture.Comprehension and production are distinct processes, making use of the store of words in the mind in different ways.

Part of the production process must consist of the selection of appropriate words according to the meaning to be conveyed. The word is then converted into a phonological shape for onward processing into speech. Thus the direction is meaning to sound. In

Resolving the Paradox of Cognates 45

comprehension, the direction of mapping is sound to meaning.

(Channell, 1988: 85)

In both, the logical possibilities are that the mother tongue can support, fail to support, or hinder someone who is learning or using the vocabulary of a second language.

Lexical transfer is particularly important at the lower levels of competence, regardless of language background. We will meet this point again in this chapter.

What is a cognate?

Definitions of what a cognate is vary. In general, there is a diachronic viewpoint and asynchronic viewpoint; in both schools of thought, there is a narrower and a broader approach.

According to the strict diachronic point of view, a cognate has a related form and meaning in two or more languages with a common ancestor (Anthony, 1953). Such a classical definition of cognates with its focus on etymology does not recognise incidental/accidental homographs such as French pain and poisson, and English pain and poison. It may also exclude words that are the result of borrowing, despite that it is not always easy to distinguish historically descended cognates from loanwords. In response, some modern linguists embrace a broader historical definition of cognates that extends the definition to loanwords.

The reluctance of some teachers and researchers to embrace loanwords is influenced by a misapplication of the diachronic view of cognates. While a diachronic view is useful for linguists considering the origins of words, it is not appropriate for the study of cognates, where what matters is not the relationships linguists see but those learners see.

By contrast, a synchronic viewpoint focuses on words with recogni- sable similarities between modern languages, irrespective of etymology.

Different word types including accidental cognates and borrowed words are recognised as potential cognates. Yet there are narrower and broader forms. Under the strict synchronic viewpoint, cognates have to be similar in meaning as well as form. Meanwhile, the broader viewpoint includes words that are similar in form but may have partially or completely different meanings.

The broader synchronic viewpoint is the most appropriate in the study of language transfer. In applied linguistics, research should focus upon perceived cross-linguistic similarities, rather than upon linguistically based analyses. ‘Words do not wear their historical origins on their

46 Part 2: Gairaigo and Language Acquisition

sleeves’ (Carroll, 1992: 102), and ‘accidental cognates’ such aspoisson and poison shows that etymology is irrelevant for learners. It is similarity in form rather than similarity in meaning that triggers learners’ recognition of cognates. We will return to this fundamental point in Chapter 6.

Learners do not seek differences, and they are perceived only when similarities cannot be established. However, when differences exist, they are likely to result in errors.

Pitfalls of Cognates

English words are transformed in various ways when adopted into Japanese (see Chapter 1), and much research has focused on the inhibitory aspect of cognates whose meanings imperfectly overlap. In this context, foreign language instruction in the 20th century can be characterised by among other things teachers’ and researchers’

desire to avoid the L1 in the classroom (Cook, 2002). Sheperd (1996) is among those to use the expression ‘pitfalls of cognates’; other hyperbole includes ‘sure-fire traps’ (Lado, 1957: 84) and ‘a veritable minefield’

(Granger & Swallow, 1988: 108). Errors of various sorts originate from cognates and can be disruptive to communication; native speakers rate lexical errors as more disruptive and more serious than grammatical errors (Johansson, 1978), and indeed they greatly outnumber them (Meara, 1984).

Although semantically identical cognate pairs exist, especially with words firmly grounded in physical reality, imperfect correlations are common, especially for abstract concepts. If learners assume that an L2 word’s meaning is identical to an L1 word applying a semantic equivalence hypothesis the errors that can result range from blatant to subtle, the latter being difficult to notice and correct. Semantic differences are typically of the greatest interest to researchers and teachers, some of whom hold that when meanings do not match, the sum influence of cognates on L2 learning is negative (e.g. Hatch & Brown, 1995; Yip, 1995).

Because the influence of L1 on L2 learning is robust, errors can be quite persistent.

Words that are similar in form but different in meaning are difficult to learn even once the speaker is aware they are different in meaning. (Abberton, 1968: 167168)

The influence of L1 is resistant to teacher intervention (Ijaz, 1986), and it is often the most intellectually gifted of students who are the most frustrated by inappropriate generalisations (Suzuki, 1984).

Resolving the Paradox of Cognates 47

The strong influence of cognates on L2 acquisition is complex and not easily described. Part of the complexity is the myriad ways in which members of cognate pairs can differ semantically. (The particular case of cognates in Japanese will be discussed in Chapter 6.)

Convergence and divergence

Varying ranges of meaning can result in either convergence or divergence. Convergence is when more than one word in L1 is equivalent to a single word in L2; divergence is the opposite case. With errors of convergence, learners who read or hear an L2 word assume it refers to the single meaning of an L1 word they know. These errors are most common in receptive tasks such as reading and listening. Fortunately, there may be contextual clues telling the learner that such an assumption is not correct. Compared to those of convergence, errors of divergence are more serious in production, i.e. speaking and writing, which requires learners to select one word from several options. For instance, as the English words discuss and argue are both expressed by a single Spanish word discutir, Spanish learners are likely to use them interchangeably (Macaulay, 1966). Moreover with errors of divergence, it is not always obvious to a listener what the intended meaning is, making correction less likely.

False friends

Some cognates are known as false friends (or false cognates), originally from the French ‘faux amis’. There are several possible forms, the most common being homography, where the spelling is identical but the meaning is not. The second most common form is homophony, where the pronunciation is similar, but the meaning is somehow different; this form occurs with Japanese loanwords.Close false friends, where meanings are deceptively close, may be the most treacherous for learners, as some occur in much the same context as the appropriate words. Distant false friends, where meanings are clearly different, cause some of the most glaring errors (see Bensoussan & Laufer, 1984; Ringbom, 1982). Yet distant false friends are generally easier than close false friends; in production, learners are less likely to make and continue blatant errors;

and in comprehension they can often spot that the meaning they are trying to apply to a word does not fit the immediate context (Holmes &

Ramos, 1993).

48 Part 2: Gairaigo and Language Acquisition

Frequency and register

One of the more subtle difficulties of cognates involves the proper frequency of their use and register restrictions. There is a natural tendency for learners to rely heavily upon the L2 words with which they are comfortable, their ‘lexical teddy bears’:

. . . Learners even as advanced as Norwegian university students

hold on tightly to words they feel safe with. Sometimes they will be led astray by their lexical teddy bears, but often they will get away with them. The types of teddy bears available are a function of relationships between the L1 and the L2. Some teddies are handed to learners on a plate such as cognates . . . (Hasselgren, 1994: 256) Even advanced learners may rely on cognates in production at the expensive of a more varied vocabulary, or to the exclusion of more appropriate words, such as using gasoline instead of petrol in England as a ‘stylistic false friend’ (Granger & Swallow, 1988: 53). An important aspect of mastering cognates is learning how their frequency and stylistic registers differ from the corresponding L1 words, and that words that may have a harmless or neutral connotation in L1 might be offensive or arrogant in L2, or vice versa.1

On the other hand, frequency may not be so serious a problem when languages borrow lexis mostly from high-frequency vocabulary, and register is not a problem with technical terminology. Moreover, concepts such as frequency and register are much less important in the receptive skills or reading and listening (Ringbom, 1990).

Nativised L2

The most extreme instance of non-correspondence between L1 and L2 is ‘nativised L2’. For instance, French has pseudo-English expressions such as rugbyman meaning rugby player and Japanese has sarariiman for businessman. Their influence on L2 acquisition is not necessarily bad because they are relatively likely to be corrected by listeners.

Experimental Data on the Effect of Transfer on Vocabulary Acquisition

A second language learner is confronted with a formidable task when acquiring the related lexicon, and the L1 can be of great assistance.

This task may be facilitated if the learner is able to apply the system of meaning he already possesses in his first language. More rapid

Resolving the Paradox of Cognates 49

learning of a new vocabulary may be achieved if learning is guided from the known to the unknown. (Hammer & Giauque, 1989: 36) Errors are frequent because cognates encourage learners to produce, and eventually the facilitation of cognates outweighs the early errors they cause. Just as errors from negative transfer have been catalogued at length, the facilitating effects of cognates have been persuasively described. Cognates are the easiest vocabulary to learn (e.g. Laufer, 1990), and learners who have access to cognates do much better than those who do not (e.g. Odlin, 1989). Learners can identify cognates in L2 and guess or inference their meanings even before they have learnt much of that language (e.g. Richman, 1970). Cognates allow rapid expansion of vocabulary for various language students.

For the Spanish speaker learning English, the number of such lexical items that he can add to his vocabulary is enormous. This is important especially for the beginner, for he often feels himself in a linguistic straightjacket, bursting with things to say, but frustrated by his belief that he lacks vocabulary. This must not be taken to mean the phonemic and grammatical (difference) of the language is to be ignored . . . but the psychological impetus gained by the student’s realisation that he knows something about English words simply because his own language contains similar words is tremendous. (Anthony, 1953: 7980)

Learners should approach cognates with appropriate caution (e.g.

Holmes & Ramos, 1993).

The effects of cognates on L2 acquisition are multifarious, yet research has shown a generally positive result. While they may interfere with precision-demanding tasks such as spelling accuracy (Ard & Homburg, 1992; Ringbom, 1987), cognates can: improve learners’ performance in vocabulary tests (e.g. Ard & Homburg, 1992); be retained more easily than other words (e.g. De Groot & Keijzer, 2000); enhance reading fluency for readers who are able to recognise them readily in L2 text (e.g.

Koda, 1989); and play a role in production in being more accessible than other L2 words, although this may lead to over-use (e.g. Hasselgren, 1994).

Cognates’ effect on spelling accuracy

The effect of cognates on spelling accuracy seems to depend on the linguistic environment.

50 Part 2: Gairaigo and Language Acquisition

Cognates can cause spelling errors; Ringbom (1987) found that spelling errors occur more frequently with Swedish Finns than with Finnish Finns because of the superficial similarity between Swedish and English. He posits that this negative effect occurs because ‘. . . Spelling is a rather special task in that accuracy is all important’ (Ringbom, 1987: 91).

If the task is of this rigidly defined nature, previously acquired knowledge and abilities which are similar to those required for the new task very often do not work well, because they lead the learner away from the straight and narrow path in which there is only one correct way of doing things. (Ringbom, 1987: 132)

By contrast, Japanese learners of English abroad (Hashimoto, 1993) and at home in Japan (Daulton, 1998) found English words borrowed into Japanese (i.e. cognates) easier to spell than non-borrowed words. It seems learners whose native language uses a Roman script have more confusion with English spelling than those learners whose L1 used a non-Roman script (Oller & Ziahosseiny, 1970).

Superior performance on vocabulary tests

Many researchers have found that cognates can support learners’

performance in vocabulary tests.

For example, Ard and Homburg (1992) compared 194 Spanish and 100 Arabic learners of English who took the Michigan Test for English Language Proficiency. Spanish possesses many more English cognates than Arabic, and the scores of Spanish speakers were remarkably higher than those of Arabic speakers on not only cognates but also non- cognates. Apparently Spanish speakers have more time to spend on non- cognates, whereas Arabic speakers must expend equal effort on both cognates and non-cognates. That learners with access to cognates do better not only on cognates but overall has been confirmed in a study involving subjects from 20 countries (Laufer-Dvorkin, 1991).

Finland provides an ideal situation for investigating transfer because while the Finnish-speaking and Swedish-speaking populations closely resemble one another culturally and educationally, only Swedish is related to English. Palmberg (1987) found that Swedish-speaking Finns at elementary levels were drawing particularly upon formal similarity between Swedish and English to successfully identify unknown English words in a written text; they managed to assign a correct meaning to most of the unfamiliar words, although the meaning was sometimes approximate rather than exact.

Resolving the Paradox of Cognates 51