In general it is true that the younger people start learning a language, the higher their level of proficiency in that language will ultimately be. People who start acquiring a new language after the age of twelve, therefore, usually do not reach the same level of proficiency in that language as native speakers do. According to the critical period hypothesis this is due to maturational changes in the brain before puberty, which make people less and less sensitive to language input. Because of this reduced sensitivity, a native-like level of proficiency should not be attainable after puberty. This hypothesis, applied to the domain of syntax, is the basis of the first research question in this dissertation:
- Are there any late second language learners who fall within the native speaker range in their command of grammatical constructions that are known to be very difficult for second language learners and which can only be acquired on the basis of the input?
This dissertation also contains an investigation into the relationship between a native-like level of proficiency (if attainable at all) on the one hand and the typological distance between the language pairs involved and the background characteristics of the participants on the other hand. This is expressed in the following research questions:
- How is the level attained in L2 grammar after the age of twelve related to the typological distance between the L1 and the L2?
- What are the input and background characteristics of late learners who perform within the native speaker range (if they exist)?
For pronunciation, there are a number of previous studies that have identified second language learners who could not be distinguished from native speakers (see e.g. Bongaerts, 1999). For morphosyntax, results have been less clear and more controversial. (compare for example Coppieters, 1987; Birdsong, 1992; Hyltenstam, 1992; Ioup e.a. 1994; White & Genesee, 1996). Moreover, there were methodological problems with many of these studies and little attention had been paid to the role of the mother tongue. In the study presented in this dissertation, these problems were addressed and the relation between proficiency level at the end state and differences between the L2 and the first languages involved was systematically investigated.
In this dissertation, a study is presented in which 43 native speakers of German, French and Turkish participated, who arrived in the Netherlands after the age of twelve and who were highly proficient in Dutch. Their performance on two grammar tests was compared to the performance on the same tests of (highly educated) native speakers of Dutch.
To assess the (implicit) grammatical knowledge of these participants, their command of dummy subject constructions in Dutch was tested. In these constructions the logical subject is not in its normal syntactic position for semantic or pragmatic reasons. Instead, this position is occupied by het, er or 0.
In our study, we distinguish three types of dummy subject constructions:
- (active) sentences with er or 0 in which the (logical) subject is a noun phrase (DP) (DP-type)
- active sentences with er, het or 0 and a sentential (logical) subject (AStype)
- passive sentences with er, het or 0 and a sentential (logical) subject (PStype)
The native speakers of Dutch in this study revealed a general preference pattern for each type (a preference for het, er and/or 0). This pattern is disturbed by certain factors. Therefore, each type consists of two or three categories with different judgement patterns. Examples are presented in (1)(6):
Men beseft niet altijd dat 0 een pinguïn een vogel is. (DP-type, general pattern) “One does not always realise that a penguin is a bird.”
Ik vind het vervelend dat er boven een raam open staat. (DP-type, non-specific subject in intransitive sentence) “It bothers me that there is a window open upstairs.”
Meestal valt het niet mee om kaartjes voor een concert te krijgen. (AS-type, general pattern) “Usually it is not easy to get tickets for a concert.”
Nu schiet 0 mij ineens te binnen dat ik nog boodschappen moet doen. (AS-type, change of state) “Now it suddenly occurs to me that I still have to go out shopping.”
In de krant wordt 0 beweerd dat hij dronken achter het stuur gezeten heeft. (PS-type, general pattern) “It is claimed in the newspaper that he was drunk while he was driving.”
(6) Door haar vrienden wordt het bewonderd dat ze ook in moeilijke tijden vrolijk blijft. (PS-type, dummy object in equivalent active sentence) “Her friends admire her for remaining cheerful, even in difficult times.”
There were two important reasons for choosing dummy subject constructions. First, they are known to be very difficult to acquire for second language learners. Second, they are hardly covered in Dutch grammars and L2 text books. This means that learners, having no access to explicitly formulated rules, can only acquire these constructions on the basis of processing language input.
Tot test the participants' command of dummy subject constructions in Dutch, two tasks were used in this study: a sentence imitation task and a sentence preference task. In the sentence imitation task participants had to repeat orally presented sentences literally. It has turned out that participants often unconsciously change elements that are phonologically non-salient and ungrammatical (from the point of view of the participant). In the sentence preference task participants had to indicate on a scale which sentence of a minimal pair they preferred. We also gave all participants a questionnaire with questions about background characteristics, such as age of arrival in the Netherlands, self-reported proficiency in various languages, level of education and questions about usage of Dutch and the L1.
The results on the tasks described above show that there are second language learners in each L1 group who have reached a native level in L2 grammar after the age of twelve. On the sentence preference task there were eight second language learners who performed within the native speaker range: three native speakers of German, four native speakers of French and one native speaker of Turkish. For the sentence imitation task there were eleven second language learners who performed within the native speaker range: seven native speakers of German, three native speakers of French and one native speaker of Turkish. As can be seen from these results, the role of the typological distance between the L1 and Dutch seemed to be greater for the sentence imitation task than for the sentence preference task.
A comparison of the learners who fell within the native speakers range on the sentence preference task according to our (strict) criteria with the other second language learners suggested that the role of factors such as input, attending Dutch classes and age of arrival (after the age of twelve) were rather limited. At the same time, there did seem to be a meaningful relation with level of education, proficiency in some other language and pleasure in learning languages. In addition, it appeared that many participants within the native speaker range had a linguistic background.
On the basis of these results it was concluded that reaching a native level after the age of twelve is possible for constructions that are difficult to learn and for which no explicit knowledge is available. The results thus falsify the critical period hypothesis. We also established that reaching this level is even possible for second language learners with an L1 which is very different from the L2 (both typologically and with respect to the constructions investigated). It should be noted, though, that the people who reach a native level constitute only a small percentage of second language learners. One should, therefore, exercise caution and not have unrealistic expectations for the majority of second language learners.
Finally, it was argued that the results with respect to the background characteristics of the second language learners suggest that factors on which learners can exert most influence seem to play a rather limited role, while something like language aptitude or language awareness seems to play a more important role. It seems plausible that people with higher aptitude or language awareness should be better able to notice and process details in the form of the L2 input than average L2 learners. This might have contributed to their greater success in acquiring difficult constructions that are phonologically non-salient and do not contribute much to the meaning of a sentence, as is the case for dummy subject constructions.