• Nie Znaleziono Wyników

6. Development of Accuracy

6.1. The dynamics of error distribution

6.1.2. Development of morphological errors

Errors in the inflection or morphological structure of words are in gen-eral the least common mistakes made by the majority of participants.

The mean curve decreases steadily in an almost linear fashion. One of the reasons for the low frequency of such errors may be the language typology. Compared with Polish or German Swedish is less inflected, so students have to learn only a  few inflectional paradigms. Nouns have two cases: nominative and genitive (German has four and Polish seven), verbs are inflected for tense and voice and not for person (German and Polish), number (German and Polish), aspect (Polish) or gender (Polish).

Swedish as one of a few languages that admittedly has a far more differ-entiated category of definiteness, a “double definiteness,” where in some constellations both prenominal and enclitic articles occur. The question of double definiteness tends to be a syntactic and semantic phenome-non and from a structural point of view the opposition between an un-inflected bare noun and an un-inflected noun with a definite suffix is not sophisticated.

Another reason for the low frequency of morphological errors lies in the fact that this particular group of students are accustomed to learn-ing languages. They study Swedish philology and have prior knowledge of at least two other Germanic languages (German and English) that they had studied in classroom settings at primary and secondary school where grammar is one of the main topics. So we can assume that learning inflectional paradigms can proceed faster than in the case of e.g. untu-tored learners. The lower frequency of morphological errors can thus be interconnected both with language typology and learning habits – these students have mostly mastered the main inflectional patterns during the first months of learning Swedish. On the other hand, it does not auto-matically mean that they do not make any errors at all.

The following errors could be recognized in the analysed data:

1) errors in the morphological structure of words, 3) errors in the inflection of nouns,

2) errors in the inflection of verbs, 4) errors in the inflection of participles, 5) errors in the inflection of pronouns, 6) errors in the inflection of adjectives.

Errors in the morphological structure of words occurred when learn-ers tried to build new words using assumed derivational or lexical mor-phemes, i.e. sjukhuspersonell* – correct form: sjukhuspersonal (hospital staff), väntning* – correct form: väntan (anticipation), färig* – correct form: färjerik (colourful). Inaccurate inflection of nouns concerning above all plural forms: dator* – correct form: datum (dates), männer* – correct form: män (mens), handskor* – correct form: handskar (gloves). In few cases incorrect building of the definite form of nouns occurred: moren* – correct form: modern (the mother), nyckelen* – correct form: nyckeln (the key), parten* – correct form: partyn (the party). In the inflection of verbs errors occurred in the building of the past tense, the present tense or the supine: glömte*, glömade* – correct form: glömdepast (forgot), sittade*

– correct form: sattpast (sat), beto* – correct form: beteddepast (behaved),

134

minnas – correct form: minnspres. (remember), tillbringer* – correct form:

tillbringarpres., sagit* – correct form: sagtsupine (from: say), bett* – correct form: betettsupine (from: behave). Samples for errors in the inflection of participle were: fascineradde* – correct form: fascinerandepres.part. (fasci-nating), överrasked* – correct form: överraskadpast.part. (surprized), hyrde*

– correct form: hyrdapast.part.pl. (hired). Errors in the inflection of pronouns concerned personal pronouns and their inflection: de – correct form:

demobject form (them), sig – correct form: migrefl.pron.1.pers (myself), eras*

– correct form: ossobject form.1.pers.plur (us). And errors in the inflection of adjectives were connected with comparison and/or congruence: värste*

– värstasuperl.def.plur. (worst), svärsta* – correct form: svårastesuperl.def. (most difficult), främmandet* – correct form: främmandenot inflectable (foreign).

As Figure 6.23 shows, errors in the inflection of verbs are the most common category and are the only type of error that every learner made. Also very common were errors in the morphological structure of words. In this case, however, the problem is clearly interconnected with individual characteristics. Almost one half of all the errors of this type were made by one student (S9), who has been characterized as a “risk taker:” with a high level of lexical diversity and a high number of lexical-semantic errors. Errors in the morphological structure of words are without doubt a consequence of lexical creativity, where the learner tries to find expressions that help to build utterances in his/her next language, and this type of error is therefore connected with lexical-se-mantic errors.

Figure 6.23. Development of morphological errors by type 0

EXP1 EXP2 EXP3 EXP4 EXP5 EXP6 N

The mean developmental trajectory for all morphological errors shows a steady decrease, with the first value being 0.17 morphological errors per T-unit. Compared with a value of 0.57 E/T, which was the correspond-ing result for lexical-semantic errors, we clearly see a huge difference.

However, the almost linear drop-off in morphological errors could not be confirmed in individuals. No single participant in the present study could be labelled as “average.” However, the ratio of this type of error decreased or remained at the same level in all the students. On the other hand, the frequency of morphological errors only remained unchanged in students who did not make any errors in the first experiment or made very few errors. In the distribution dynamics for this error we can recognize the pattern observed in the development of AccInd: the big-gest decrease occurred at the beginning of the learning period (Figure 6.24). In 60% of the students the growth rate between the first and the second experiment was at least –38%. And this pattern was observed in both learners with low and high morphological error ratios in the first experiment. Two other learners (S6, S15) made very few morphological errors throughout the entire experimental period, so in their case they had already mastered the morphological systems of Swedish during the first semester. Four students needed more time to order in their minds the morphological pattern in Swedish. In their case such a substantial decrease in errors of this type occurred one or two semesters later.

The development of morphological errors corresponds to the devel-opment of general accuracy: the biggest improvement in Swedish mor-phology occurred early on in the study period. After 300 hours (two se-mesters) of foreign language instruction almost all the learners properly identified and used inflection patterns and morphological structures of words. This developmental path also corresponds with the assumptions of processability theory, where the processing of lexical morphemes lies at the next lowest level, which means that their acquisition occurred at a relatively early stage. In the context of Swedish at this level plural, definiteness on nouns, and past or present tense on verbs have been listed (Pienemann, 1998; Pienemann & Håkansson, 1999). And these errors were the most common mistakes made by the participants of the present study.

Even if the morphological level in general develops early, most stu-dents develop dynamically even in later periods. Their trajectories re-main a sinusoid, which is a sign that the morphological system has still not been fully stabilized. On the other hand, there are some learners

136

(consisting exclusively of students with a  low ratio of errors), whose mastering of Swedish morphology clearly resulted in their approaching an attractor state as early as the second semester (see Figure 6.25 – right diagram). Their intra-individual variability was considerably lower than it was for the other group of students. Variability in the develop-ment of the morphological systems is also clearly connected with the stability of the system: the process of internalisation is reflected in such variability.

But even this variability should be relativized. The highest error ratio in the group was 0.46, which means that the learner made a morpholog-ical error in every second T-unit. Such high values did not occur in other students and experiments, where the frequency of the inaccurate use of morphological features was no greater than one error in every third T-unit. Thus we can conclude without any doubt that morphological ac-curacy develops early.

Figure 6.24. Learners achieving a considerable decrease in morphological errors during the second semester