Toegepaste Taalwetenschap in Artikelen - Volume 30, Issue 1, 1988

This paper critically examines Pienemann's theory of second-language acquisition orders. The paper raises methodological as well as theoretical issues: the validity of the acquisition order found by Clahsen, Meisel & Pienemann (1983), evidence for this order from longitudinal research, evidence from learners with various mother tongues, the relation between language production and language comprehension, and predictions concerning other linguistic structures than the ones dealt with in the 1983 study.The paper focuses in particular on Pienemann's recent claim (1987b), that the acquisition order of L2 structures is determined by a number of cognitive 'processing prerequisites'. It is observed that Pienemann has shifted his attention from the point at which L2 learners can be said to have acquired, or mastered, grammatical features, towards the onset (first few occurrences) of grammatical features, in order to find evidence for a successive reduction of processing constraints. It is argued that explanations of acquisition orders in terms of onset, or in terms of some other acquisition criterion (e.g. 80% correct use in obligatory context), tend to pay exclusive attention to the rank order of acquisition points, leaving underexposed the time span between first use and final mastery of a rule, and the overlap in acquisition periods of different rules. Often the acquisition of a grammatical feature may not be an instantaneous feat, but a developmental process. Therefore, rank orders of onset and mastery may lose much of their relevance if length and overlap (development) are not taken into account. It is argued that Pienemann's theory, having opened our eyes for the onset of grammatical features, deserves to be deepened and broadened, in order to take the development of grammatical features into account as well.

Within a psycholinguistic approach to second language learning, an attempt is made to investigate the question of how morphology, syntax (word order phenomena), semantics and pragmatics affect the comprehension of Dutch sentences for normative learners of that language. When talking to nonnative language-learners, native spea-kers often tend to dehberately modify their speech -'simplify' it - in an attempt to make the target language more comprehensible. Omitting semantically redundant function words and copulas, or deliberate-ly modifying the word order in a sentence, are but a few characteris-tics of sucn 'simplifications'. In trying to determine whether, and what kinds of, linguistic simplifications promote comprehension, an important theoretical issue arises, namely, the relationship between linguistic (structural) and cognitive (ease of information processing) simplification. That one form of simplification is by no means a guarantee for the other form is an important assumption that forms the backbone to our approach.The results from research on morphological simplifications (omission of redundant function words in utterances) in two parallel experiments - an artificial and a natural language one (Dutch) - are discus-sed. They suggest that the presence of semantically redundant functi-on words is not experienced as bothersome "noise" in the successful inference of the meaning of unfamiliar utterances, as long as supra-segmental cues are present. The suprasegmental structure provides the listener/learner with cues for locating the potentially meaningful elements of such utterances.Research on syntactic simplifications is also discussed. Its aim was to examine the role and effect of syntactic and semantic cues on sen-tence interpretation. Two important questions were: (a) What are the processing strategies and cues responsible for the interpretation of Dutch sentences by native speakers, and how do they compare to those employed by nonnative speakers? (b) Are the processing stra-tegies and cues that are responsible and decisive for first language comprehension also those employed in second language comprehension? The performance of Dutch control subjects on a Dutch sentence interpretation task is presented, and hypotheses are put forward as to the locus and cause of eventual performance differences in a nonnative subject population (English learners of Dutch). Some relevant theoretical implications of our findings are also mentioned.

Among researchers in SLA there is currently a debate on wheter or not L2 learners may have access to Universal Grammar. Clahsen & Muysken (1986), for example, claim that this should not be the case. This is because the interlanguage system of L2 learners should contain rules which do not belong to "a possible rule system". Felix (1987), Flynn (1984, 1985) and White (1987), however, claim that the interlanguage system of L2 learners consists of "instantiations of possible rules". Furthermore, L2 learners should possess linguistic knowledge that cannot be related to general learning principles, nor to linguistic knowledge of L1. In the present article evidence from a study by White (1987) with respect to the so-called "logical problem of language acquisition" and from studies by Flynn (1984, 1985) regarding the so-called "logical problem of language development" is examined critically. With regard to the acquisition of the thai-trace effect in L2 English by native speakers of Dutch it can be shown that it is not necessary to assume that L2 learners should have access to Universal Grammar. Similarly, it can be shown that evidence that is interpreted in favour of the resetting of the head-parameters in the L2 English of native speakers of Japanese and Chinese can also be explained alternatively. Methodologically it appears to be problematical to find empirical evidence for the use of Universal Grammar by adult learners of a second language.

The subset principle, recently formulated by Wexler and Manzini as a theorem in L1 acquisition, can be roughly described as a learning function linking a set of input data to a grammar G which generates the "smallest language" compatible with such a set. This property of G guarantees that the acquisition process can only take place on the basis of positive evidence; negative evidence thus does not have to play a role. This article discusses the question whether the subset principle also plays a role in the process of second language acquisition. Contrary to what is suggested in the literature it has to be concluded that as an L2 acquisition strategy the subset principle is either redundant or incoherent. The reasoning on which this conclusion is based involves certain implications for the research concerning the problem of overgeneralisation in L2 acquisition.

The LARSP procedure by Crystal, Fletcher and Garman (1976) was used as a starting point for GRAMAT, a method by means of which the development of morphosyntax in Dutch children from one to four years of age can be profiled. The developmental scheme was based on 72 spontaneous speech samples of 42 children. The children were selected on the basis of age, sex and socio-economic group. From every sample, recorded at the homes of the children, 100 utterances were analyzed at the clause, phrase and word levels. In the developmental scheme, which was divided in six stages, each covering a period of six months, the morphosyntactic structures were included which a) are used by at least half of the children at a particular stage and b) are used with a median frequency of at least 1.0. The morphosyntactic development which takes place from one to four is describea both in quantitative and qualitative terms.

Increasing use of spontaneous speech data in second language re-search made it necessary to use other measures than tests for getting insight into the lexical richness of (spoken) text corpora of second language learners. In this article several lexical measures (types, TTR(c), Guiraud, logTTR, Uber, theoretical vocabulary) are examined and discussed. On the basis of several studies that the TTR is the worst measure possible, but the better measures (types, Guiraud, theoretical vocabulary) are problematic too. Next, the characteristics of the lexical measures are examined by means of a few simple computer simulations with hypothetical lexica. The outcome is mat neither the reliability, nor the validity of the different measures are satisfactory, especially with respect to the TTR.

In this article a preliminary report is given of a research project concerning a Dutch version of Crystal, Fletcher & Garman's LARSP, Language Assessment Remediation and Screening Procedure, a gram-matical developmental scale for the use of language therapy. The first Part of the project consisted of a longitudinal study of 15 children between 1;3 and 3;6 with six monthly spontaneous language samples per child. The result of this study showed that the most striking and stable characteristic of grammatical development till the age of four was the increasing number of constituents in the sentence (with the Declarative sentence having one more constituent than the Command and the Question in all stages, compare Wells, 1985). Based on this a cross-sectional study was carried out with the number of constituents of the Declarative or the same number minus one in and Q as a language measure. After assigning the 100 children involved to the seven stages of the Dutch TARSP, the lower language level part of the scale was constructed: Wordgroups, Connectives, Pronouns and Word Structure. The Dutch version of TARSP is compared to a smal-ler Dutch study based on LARSP with a different methodological set-up: GRAMAT.

The diversity of finite tense/aspect/modality markers in Turkish means, firstly, that the functional load for each of them is more restricted than in western European languages, and secondly that for certain discourse types specific forms are usually selected. Thus these forms may constitute rather transparant morphological indicators for the semantics of separate utterances. A case in point are descriptions of scenes (such as picture descriptions) which are based on a simple binary tense/aspect-opposition (pres.tense +Iyor for action): (perfect +mIş for states). Stones about happenings in the past on the other hand tend to centre around the opposition (neutral +DI):(imperfective [durative +Iyordu]: [iterative +()rdI]).In the paper the patterns of acquisition of these oppositions in the speech of children growing up in the Netherlands is analyzed on the basis of picture descriptions and interviews.In the picture descriptions it was found that the 4-year-old kinder-garteners frequently used +DI where one would expect + mIs in adult language. This form has almost completely disappeared in the descrip-tions by the 5-year-old children. Frequent marking with +DI reappears in the descriptions by the 7-year-olds; with them, however, the use of the +DI-forms are not a sign for the lack of differentiation from +mIs but rather reflect the fact that the pictures offered for descrip-tion were sequences: by this age the children are aware that the pictures in fact are not merely separate scenes so that they connect them with +DI-forms, mostly together with the conjunction (ondan) sonra "and then". The same acquisitional pattern is reflected in the data from two control groups of children growing up in Turkey. This is interpreted as resulting from the lack of morpho-syntactic difficulties involved: the acquisitional pattern quite directly reflects the development of cognitive skills.

The following questions are taken into account:. what types of word formation principles are used by language learners?. what lexemes are combined in nominal compounds (N+N) and nomi-nal circumscriptions (N+prep+N)?. what semantic relations are expressed in nominal compounds?. what binding principles are taken into account?. what suffixes are used in derivational processes?. what semantic roles do these suffixes refer to?Data analysis is based on the use of L2 Dutch by 2 Turkish and 2 Moroccan adult informants in 2 types of activities (film commenting and conversation), at 3 different moments in the course of language learning (Ntotal=4 informants χ 2 activities χ 3 moments = 24 transcribed texts).In accordance with findings on first language acquisition processes, compounds not only precede derivations, but at the same time they compensate for standard derivations, thus resulting in lexical innova-tions. All informants make a creative and innovative use of a variety of compounding principles. In addition, opposite principles in Arabic and Turkish seem to lead to different preferences of our learners:. the Turkish informants make more use of different types of nominal compounds than do the Moroccan informants;. only the Turkish informants make more use of left oriented com-pounds, based on a combination of more than two lexemes;. only the Moroccan informants make use of circumscriptions.Finally, the following preferences of our learners are in accordance with standard language preferences: nominal compounds, in comparison with other types of compounding;. specifier head compounds, in comparison with other nominal com-pounds;. goal relations in specifier head compounds, in comparison with other types of semantic relations;. zero marking as a binding principle within specifier head com-pounds, in comparison with other binding principles.

Theories of phonological processing and of phonological change stress the importance of the word as the structural context of segments in acquisition and historical development, respectively. After a discussion of these theories with reference to a model of second language acquisition, an experiment is reported on in which the pronunciation of four Dutch learners of English was examined. Specifically, they were required to produce a number of known and unknown words from a written list. Examination of the data shows that indeed a model of second language phonological development may be postulated in which the status of the words in which they are embedded as known or unknown is critical for the form of the target segment produced.

The present paper reports on a study of the continued word associa-tions to eight stimulus words in Dutch and Turkish by Turkish migrant children living in the Netherlands. For three out of the eight stimulus words the meanings in the two languages were not complete-ly overlapping. Two written association tasks were administered with an interval of five months. Seventeen Turkish children (about 12 years old) participated in the study. It was found that:- Interlingual responses hardly occurred (only a few instances of Dutch or Dutch-like words in the Turkish task).- The children gave more responses (number of tokens) in the Tur-kish than in the Dutch version (difference not significant at 5% level; significant at 10% level), suggesting a larger vocabulary in Turkish.- The diversity of the responses was slightly higher in the Dutch version of the task than in the Turkish version.- The level of linguistic-cognitive development as indicated by the proportion of paradigmatic and categorical responses was about the same in the two languages with a slight bias in favour of Turkish.- The number of equivalent responses in the two tasks was 36,8%.- For five of the eight stimulus words ('happiness' and 'city') the responses in the two languages differed strikingly. Probably the response behaviour was influenced by social-cultural factors related to each language.- For one of the three semantically not completely overlapping words (Dutch: dochter, and Turkish kiz, 'daughter'/'girr) the Dutch res-ponses (frequent use of kinship terms) differed strongly from the Turkish responses (few kinship term)

Research into proficiency in the Surinamese languages by Surinamese children in the Netherlands has to contend with a twofold problem with regard to the question of the norm. There are no clear norms for the Surinamese languages. In addition, there is the general problem of norms for the investigation of primary language loss. To characterize the proficiency of children in Sarnami and Sranan, the languages of the Surinamese Hindustani and Creole communities, the method of working with several judges appears to be fraught with problems. This article reports on the results of an estimation and error analysis with one judge for each language, supplemented with some quantitative language measures.The investigation involved 16 Hindustani and 12 Creole children from primary school classes containing a rather high proportion of children from their own ethnic groups in the Hague ana Amsterdam respectively. The children had to recall a taped story in Sarnami or Sranan, give a summary of a Dutch geography text and relate some experience of their own in a conversation with a Hindustani or Creole researcher. The speech was transcribed and presented in written form to the judges.The mean scores for errors turned out to be remarkably low: there were almost no errors in word order, while morphological and lexical errors remained under 5% and transfer of morphology from Dutch was negligible. Depending on the nature of the task, the judges' estimate of the use of Dutch words instead of Surinamese words (in mixed sentences) varied from 3-10%. The summary yielded most of these. The total number of Dutch words in the geography text was a third. The judges' estimate is in line with the result that judges in general are more tolerant in accepting Dutch words concerning education as being part of the Surinamese lexica. The children produce more complete Dutch sentences in the conversation task, probably because this task approximates to spontaneous language most, while recall and summary tasks stimulate monitoring. In general, the use of Dutch words does not affect the basic grammatical structures of the Suri-namese languages. With regard to social variables, proficiency appears to be linked to the home language for the Hindustani children, and to contact with other children of their own ethnic group for the Creole children. Length of residence in the Netherlands appears not to be important for Hindustani children, while a longer stay appears to result in a lesser degree of proficiency for Creole children. The Eroportion of girls and boys was equal for the Hindustani children, oys appear to be slightly more proficient in Sarnami.