Woordvormanalyse Door Kinderen: De Rol Van Klemtoon en Vormovereenkomst

Early word form representations are assumed to be unanalyzed 'routines'. Around age 2 1/2, when the first 50 to 100 words have been acquired, the organization of the mental lexicon starts to change. Word form representations are segmented into their constituent linguistic substructures: syllable and phonemes. Gradually the network-like structure which is thought to capture the mature mental lexicon emerges. Peters (1983, 1985) has proposed two heuristics that may be employed by children for segmenting words, both during this 'reorganization' and afterwards, when new words are acquired and inserted into the rapidly expanding lexicon. The first employs word stress, the second is based on matches between a new form and lexical items that have already been acquired.This study investigates in an experimental fashion whether children use stress and formal similarities in segmentation. Children (4, 5 and 6 yrs. old) were asked to alter syllable onsets in trisyllabic words with stress on either of the three syllables, and in trisyllabic complex words in which at least one morpheme could be expected to be known by the subjects.It appeared that word-internal onsets of stressed syllables were more often altered than onsets of unstressed syllables. These results are supportive of the 'stress' heuristic. Also, syllable onsets which coincided with word-internal lexical morpheme boundaries (i.e., boundaries between known and unknown parts which can also occur independently) were more often altered than non-boundary onsets. However, syllable onsets at root-suffix boundaries were not altered more frequently than non-boundary onsets. This difference between lexical morpheme boundaries and suffix boundaries is not predicted by a segmenting heuristic based on formal match. It is therefore suggested that in fact prosodic characteristics of the final morphemes in the complex words factors could be responsible for the difference: all first (or only) syllables of the right-hand lexical morphemes were stressable, whereas the suffixes were not.This conjecture was tested in a second experiment in which 7 and 8 year old children were asked to alter syllable onsets in derivations consisting of a bisyllabic root and a monosyllabic suffix. Some suffixes were stressed, others were unstressed and still others caused stress in the stem morphemes to be shifted to the second syllabe. It was found that stressed suffixes yielded more alterations than unstressed suffixes. Second, stressed syllables, in general, yielded more onset alterations than unstressed syllables, regardless of whether stress was assigned by the stem morpheme or by derivation with a stress-shifting suffix. Finally, syllables which had 'lost' stress as a result of dervational stress shift produced more alterations than unstressed syllables that were not stressed, in the stem morpheme either.These results support the conjecture that the apparent effects of formal match on segmentation behavior in the first experiment can be explained by reference to stress. Consequently, it is concluded that stress is the primary cue in word segmentation. This conclusion is in accordance with 'prosodic bootstrapping' theory, which claims that children are 'programmed' to use prosodic features as clues to linguistic structure.