In order to quantify this notion, we have used the mean size of paradigm MSP measure proposed by Xanthos and Gillis in prep. In this section, we will first recall the general definition of MSP. Then we will show how it was implemented in the developmental framework of this research; in particular, we will introduce the distinction between total and cumulative MSP and will explain how successive evaluations of MSP were used to calculate the speed of development of morphological richness.
Then, we will consider the question of the dependence of MSP on sample size and describe the resampling procedure that was applied for size normalization. Finally, we will discuss the method that was used to set the starting and end point for each child corpus that was investigated. Mean size of paradigm MSP Despite frequent references to the intuitive notion of paradigmatic morphological richness, there is no widely accepted way to measure it. Arguably, the first proposal suitable for cross-linguistic acquisition studies was the inflectional diversity ID measure developed by Malvern et al.
An alternative approach was proposed by Xanthos and Gillis in prep. By construction, MSP ranges between 1 and F. Since the number F of different word-forms in a sample cannot exceed the size in tokens of that sample, it follows that the maximum value of MSP is dependent on sample size. Xanthos and Gillis in prep. It is claimed that ID inherits from D the important property not to be a function of sample size Malvern et al. Moreover, it can be assumed that it shares with it the property of representing the entirety of the relationship between morphological respectively lexical richness and sample size.
For this research, however, we chose to use MSP for two main reasons. First, we believe that the characterization of morphological richness in terms of an average number of inflected forms is a particularly intuitive one; this will be an important asset when we will add an extra layer of complexity by considering the speed of development of morphological richness. Second, in contrast to ID, MSP does not rely on a specific assumption as to the nature of the relationship between morphological richness and sample size; to that extent, it may be rightly considered a more agnostic perspective.
Developmental perspective Total vs. In order to account for this additional structure, we introduce a distinction between the total MSP, which is computed over a whole corpus, and the cumulative MSP at month m , which is computed over the whole corpus up to month m. The latter is denoted by a subscript: MSPm. This example shows that the cumulative MSP is not bound to increase over time. In particular, it increases between months m and m' if and only if the number of new word-forms is greater than the number of new lemmas times MSPm.
The use of a cumulative as opposed to monthly definition of MSP is an important feature of our methodology. It conveys our intent to capture paradigmatic oppositions between word- forms whose occurrences may span different months. Consequently, morphological richness is in principle evaluated over an ever growing amount of data. Given the dependence between raw MSP and sample size see previous section , this makes a strong case for carefully designing a normalization procedure.
For this research, we generally considered the case of successive months i. These are in turn averaged to yield the monthly speed in the corpus as a whole. This average value is the reported score of each corpus on the dependent variable - with the additional specification that it is computed on the basis of normalized cumulative MSP, as indicated in the next section. The average monthly speed is equal to 2 - 0.
As mentioned in section 2. Indeed, the unit of measurement of speed is an average number of word-forms per month. Thus an average monthly speed of 2 simply means that, on average, 2 new inflected forms of each lemma occur during a month. This observation leads these authors to advocate the systematic application of some normalization procedure. This is an even more stringent requirement in view of our cumulative definition of MSP see section 2. Specifically, Xanthos and Gillis propose the use of the following resampling method, inspired by the mean segmental type-token ratio Johnson, Let N denote the original sample size.
Thus, each token is expected 3 The reason why we calculated the monthly values before averaging them, rather than directly computing the overall monthly speed using formula 2 , is that it enabled us to estimate the variance of speed for a given corpus, and perform an analysis of it see Xanthos, this volume. Xanthos and Gillis report empirical evidence suggesting that, while the MSP S of a sample clearly depends on S, it is actually independent of the original sample size.
We adopted their procedure, with a few modifications in order to fit the cumulative definition of MSP and the calculation of speed. By construction, cumulative MSP is evaluated on the basis of an ever increasing amount of data; our concern was to ensure that this increase remained comparable between corpora. Since our goal was to compute the monthly speed of development of MSP, we decided to set the parameter S to a possibly different value for each corpus, in such a way that the expected number of tokens per month would be the same for all of them.
Consider a corpus of N tokens ranging over L months. Let Z denote the expected number of tokens per month. The number B of subsamples and the probability P of selecting a token when constructing a subsample are then calculated as in the original procedure. An important feature of the modified resampling procedure is that it preserves the sequential structure of the original corpus: when constructing a given subsample, each selected token is assigned to the month where it actually occurred.
Thus, each of the B subsamples is itself a sequence of L monthly sub- samples. We may then calculate the cumulative MSP for each month of each subsample this corresponds to B series of L values , and the average cumulative MSP per month. The reported values of cumulative MSP were systematically normalized that way, and served as a basis for calculating the monthly values of speed of development as indicated in the previous section.
These monthly values were eventually averaged over all months composing a given corpus, and the result was reported as the value of the dependent variable for that corpus. Starting and end points As mentioned in section 2. Age-related differences in the original data provided for this research e. We considered different possible candidates for a common base-line. Age was not qualified as a useful candidate, in view of the well-known inter-individual variation in onset time of language development Bates et al. Mean length of utterance MLU was rejected as well, on the grounds that it is not a 4 As noted by Xanthos and Gillis in prep.
As suggested by Devescovi et al. We therefore settled for the proportion of utterances with verbs PUV. The largest common interval of PUV between all corpora was found to be excessively small, so that we could not use this straightforward criterion for selecting months to be suppressed. For a given sample, the CDF of a variable X is defined as the proportion of individuals having a score lower than or equal to X.
This proportion was then evenly split between the two ends of the distribution of PUV. Notice that this formulation does not lead to the suppression of monthly samples exceeding these bounds when they occur between other samples that do not exceed them: it merely amounts to cropping the beginning and end of corpora when needed.
On average, the starting points are set one month later in the aligned corpora 1;8 instead of 1;7 , and the end points one month earlier 2;8 instead of 2;9. As an estimate of the resulting reduction in variation, the standard deviation of PUV is reduced from 0. Independent variables The main independent variable involved in this research is the degree of morphological richness of the input both from a paradigmatic and a syntagmatic perspective. In addition, several other independent variables were constructed which are susceptible to show an impact on speed of development of morphological richness in child speech, notably the transparency, uniformity and salience of word forms.
Input analyses were conducted for each child corpus separately; within each corpus, the input data are treated as a single sample i. It has to be noted that the amount of input data being analyzed differs between the three language groups. For the weakly inflecting languages see Laaha et al. In what follows, we will list the seven different independent variables constructed for this research see also Dressler, this volume and explain how they were calculated. However, in contrast to the dependent variable, it measures the total as opposed to cumulative MSP in the input, and it is considered from a static perspective i.
With regard to the issue of sample size, we applied the original normalization procedure described by Xanthos and Gillis in prep. In particular, we computed the normalized MSP over tokens recall that in this version of the normalization procedure, the total size of the sample is normalized, as opposed to the average number of tokens per month.
For samples where less than tokens were available i. It was computed by dividing the total number of inflectional suffix tokens by the total number of word tokens. Nouns and verbs were treated separately. It was calculated in the following way: we used the CLAN program FREQ to establish a total word type list of the input data and classified these words or rather: their inflectional paradigm as either transparent or non-transparent i. This means that any word belonging to a paradigm in which there is at least one 5 This decision was based on the empirical observation by Xanthos and Gillis in prep.
Clearly, this is at best a good approximation, which the availability of more data would enable us to discard. Then, we divided the number of transparent word types or lemmas by the total number of word types. We used FREQ to establish a total word form list of the input data and classified these forms as either transparent or non-transparent i. Then, we divided the number of transparent word forms by the total number of word forms tokens. We first classified inflectional noun and verb categories of the language being investigated as either uniform or non-uniform i.
Then, we used FREQ to calculate the token frequency of uniform categories in the input data and divided it by the total number of inflectional categories tokens. If the language being investigated was of type c , then we calculated the proportion of inflectional suffixes with full vowels out of all inflectional suffixes tokens. If the language being investigated was of type c , then we calculated the proportion of stressed inflectional suffixes out of all inflectional suffixes tokens.
Summary In this chapter, we presented the method used to investigate the hypotheses of this research. The main hypothesis is that there exists a proportionality relationship between the degree of morphological richness of the input and the speed of development of morphological richness in child speech. In addition, factors such as transparency, uniformity and salience should show an impact on speed of development see Dressler, this volume. We also considered the question of the dependence of MSP on sample size and discussed the method used to set a common base-line for this analysis section 2.
Then, we explained the calculation of the independent variables of this research: the main independent variables investigated concern the degree of morphological richness of the input, both from a paradigmatic and a syntagmatic perspective Variables A1, A2 ; the other independent variables are word and form transparency B1, B2 , uniformity C , phonological-segmental and prosodic salience D1, D2 section 2. Tables 2a and 2b below summarize the values of each variable for each of the 13 child corpora investigated. Empty cells indicate that the value could not be calculated for the given child corpus.
More details about the analysis of each language are given in the three language group chapters see Laaha et al. Table 2a. Values of the independent and dependent variables, by child corpus: nouns Language Child Indep. V Dep. Values of the independent and dependent variables, by child corpus: verbs Language Child Indep. Dressler 3. Typological characteristics French, Dutch and German are weakly inflecting languages. The two Germanic languages Dutch and German are very closely related genetically and typologically.
The Romance language French is even less strongly inflecting than Dutch and German due to a number of characteristics of the isolating language type Geckeler, In German, nouns can receive number and case marking. There are four cases: nominative, accusative, dative and genitive. German verbs encode the grammatical categories of person, number, tense, mood and voice. There is no grammatical category of aspect.
Person 1st, 2nd, 3rd and number Sg. Within the category tense, colloquial Austrian German distinguishes between present, future, perfect, and pluperfect. Perfect, pluperfect and future tense are expressed by periphrastic constructions, i. In Dutch, nouns can receive number marking. This is accomplished by suffixation: the suffixes -en and -s are productive, while the suffix -eren is not productive occurs with a limited number of words.
Case marking is non existent, except for some fossilized phrases. As in German, Dutch verbs encode the grammatical categories person, number, tense, mood and voice. The categories person 1st, 2nd, 3rd and number Sg. These are the only synthetic forms. The perfectum, plusquamperfectum, futurum, and futurum exactum are periphrastic. For instance, the perfectum and plusquamperfectum are formed by a combination of the present, resp. The imperative is formally almost indistinguishable from the present, and the conjunctive has disappeared from colloquial speech.
There are no cases and number is expressed by determiners Sg. In contrast, the verb system is typologically more mixed. Grammatical categories of the French verb are person 1st, 2nd, 3rd , number Sg. Aspect is not encoded separately from tense. Non-finite categories are infinitive, present participle occurring only as gerundive in the spoken language and past participle PP.
The infinitive is used in periphrastic constructions such as compound future and modal constructions. PP is part of the compound forms. Within the category tense, spoken French has four compound forms compound past, compound future, pluperfect, and past future , and two synthetic forms : imperfect parl-ais imperfective aspect and simple future parl e -ra. The simple past parl-a perfective aspect is used only in fairy tales. Within the category mood, subjunctive and conditional have a synthetic form and a compound form past subjunctive, past conditional.
They can be divided into two macroclasses like in the weakly inflecting-fusional languages German and Dutch. But the number of the microclasses is higher in French than in the two other languages. To that extent the French verb system can be considered as more inflecting-fusional than the verb system of German and Dutch Dressler et al. Thus, young Dutch and German children are confronted both with the separated and non-separated variant of these verbs.
For this reason, particle verbs with the same base were considered as one single verb type or lemma for the analyses presented in this chapter. For one analysis total input MSP , we calculated two different versions: a version in which particle verbs with the same base were merged, and a version in which they were not merged see section 3. A1 below. The data The data analyzed in this chapter stem from two Swiss children acquiring French, one Flemish child acquiring Dutch and two Austrian children acquiring German as their first languages; all children are monolingual, with no developmental or linguistic problems.
The French-speaking children Emma and Sophie are born in Lausanne Switzerland in upper class families. The parents recorded the children at home, in situations of free play or while looking at picture books. Both her parents hold a university degree. The recordings were made by a research assistant, who also 1 On parle instead of nous parlons. The data of Emma are more limited than the data of Sophie. Emma was recorded generally only twice a month and some of the recordings are very short e.
This diversity in the data of Emma is probably responsible for the greater heterogeneity of some of the findings on her language development. Unfortunately parent language addressed to the two triplets is not available. The German-speaking children Jan and Katharina come from high-to-middle class families in Vienna, Austria.
- Do We Have a Choice?.
- 803 German Verbs!
- On The Principal Articles Of The Christian Religion: Commenced By The Author Chiefly For The Purpose Of Forming A System Of Divinity (With Active Table of Contents).
- Does the simplicity of Esperanto verb tenses make its verbs too imprecise?.
- Cabo de Gata (German Edition).
- Circus Techniques?
For each of the three languages, the input analysis is based on the entire maternal input sample, i. Independent variables of the input In what follows, we will present the input analyses for French, Dutch and German. They constitute the independent variables of our research. Morphological richness The first input variable investigated is morphological richness. It was examined both from a paradigmatic and a syntagmatic perspective.
Paradigmatic morphological richness In order to give a first overview of paradigmatic morphological richness in the input, we made a list of all non-homophonous form-type categories in nouns and verbs occurring in the input of the three languages Table 3. As can be seen, in nouns, the number of categories is very low in all three languages, with the lowest number in French only 2 categories. In verbs, the number of categories is higher than in nouns in all three languages; French has more verb categories than the two Germanic languages 13 categories, as compared to Dutch 5 and German 9 categories.
Table 3. The values of total input MSP for each of the three languages are given in Table 4. In verbs, total input MSP is higher than in nouns in all three languages. For Dutch and German verbs, we calculated two different versions of total input MSP: a version in which particle verbs with the same base were merged, and a version in which they were not merged. In the case of merging, total input MSP in verbs is clearly higher in the two Germanic languages than in French; in the case of non-merging, only the Dutch input shows higher values than French. Further investigations are necessary to explain this difference.
A graphical summary of total input MSP in the three languages is presented in the two figures below Figures 1a, 1b. Syntagmatic morphological richness We now turn to syntagmatic morphological richness in the input. Figure 2a shows the distribution of inflectional suffixes in nouns, for the three languages. Distribution of suffixes in nouns for French, Dutch and German Examples 1 - 3 below illustrate the different form types for each of the three languages the number of suffixes is given in parentheses.
This difference between French and the two Germanic languages is mainly due to the fact that in French, all persons in the singular are expressed by the base form, e. Distribution of suffixes in verbs for French, Dutch and German Examples 4 - 6 below illustrate the different form types for each of the three languages the number of suffixes is given in parentheses. The average number of suffixes for each of the three languages is given in Table 5. As can be seen, both in nouns and in verbs, the average number of suffixes is lower in French than in the two Germanic languages.
Table 5. Transparency The next input variable is transparency. We investigated two different types of input transparency: word transparency Variable B1 and form transparency Variable B2. Examples 7 - 9 below illustrate transparent and opaque noun and verb forms for each of the three languages. This result suggests that nouns tend to show up in the German input in a transparent form, although some of the nouns belong to opaque plural classes. Form Trans. Word and form transparency in nouns for French, Dutch and German Figure 3b shows word and form transparency in verbs, for the three languages.
For French and German, we conducted two different analyses of form transparency: form transparency for all verbs V and form transparency for lexical verbs only VLEX. V Form Trans. Macroclass I e. Both in French and in German, form transparency is higher if one considers lexical verbs only; this difference is due to the high token frequency of opaque auxiliaries and modal verbs in the two languages. The verb categories qualified as uniform in French are Sg. German has a smaller number of uniform verb categories 1Sg. This difference is also visible in the input data.
As shown in Table 6, the proportion of uniform verb categories is higher in the French input than in the German one. Table 6. Salience The last input variable investigated is salience. We examined two different types of input salience: phonological-segmental salience Variable D1 and prosodic salience Variable D2. As shown in Table 7, French and the two Germanic languages differ with respect to phonological salience. Both nouns and verbs are phonologically salient in French, because French has no inflectional suffixes with a final reduced vowel.
By contrast, they are non- salient in Dutch and German: both languages have no inflectional suffixes with a final full vowel. Table 7. Both nouns and verbs are prosodically salient in French, because word-final inflectional suffixes are always stressed in French. By contrast, they are non-salient in Dutch and German where word-final inflectional suffixes are always unstressed. Table 8.
Common English Irregular Verb List
Figure 4a shows the development of MSP in nouns for the three languages. As can be seen, in the French-speaking children, MSP in nouns shows no increase at all; in the Dutch- speaking child, it shows a very slight increase; and in the German-speaking children, an increase, similar to that in the Dutch-speaking child. Development of mean size of paradigm in verbs for French, Dutch and German Figure 4b shows the development of MSP in verbs for the three languages. As can be seen, MSP shows a steeper rise i. In the French-speaking children, MSP in verbs shows a certain increase with age; in the Dutch-speaking child, it shows a slightly higher increase; and in the German-speaking children an increase similar to that in the Dutch-speaking child.
A comparison of the development of MSP in the seven Dutch children suggests that individual differences of speed of development within one and the same language are quite restricted see Figures 5a and 5b. Development of mean size of paradigm in verbs for the seven Dutch corpora To summarize, our results suggest that the French-, Dutch- and German-speaking children investigated show a faster development of MSP in verbs than in nouns. Furthermore, the children acquiring the typologically closely related languages Dutch and German show a faster development than the children acquiring French.
These differences in speed of development seem to fit to differences in morphological richness in the input, as presented in section 3. A above. Typological characteristics The languages to be dealt with in this section are three strongly inflecting Indo-European languages, namely West-Slavic Russian, South-Slavic Croatian, and Greek.
In the three languages, verb morphology is richer than noun morphology, although the difference is much bigger in Greek than in Russian and Croatian. While the case-number system of the two Slavic languages is very rich, that of Greek is relatively poor. Besides, Greek and Croatian, but not Russian, distinguish a vocative form of certain nouns. As far as language acquisition is concerned, the degree of syncretism of case forms is important. With the exception of the class of masculine nouns ending in -os, Greek nouns distinguish only two case forms in the singular as well as the plural for classification of the Greek noun see also Christofidou, There is much less syncretism in the two Slavic languages as compared to Greek.
While, in Russian, productive nouns distinguish five case forms to express six case categories in the singular as well as the plural, unproductive ones may be limited to three different case forms in each number. Animate masculine nouns, however, do not distinguish between the accusative and the genitive singular. In the plural, animate nouns show syncretism of the genitive and the accusative, while inanimate ones do this for the nominative and the accusative. In contrast to nouns, verb forms have no syncretism at all in Russian and very little in Greek and Croatian.
Whereas the inflectional system of the noun is very similar in the two Slavic languages, there is an important difference in the inflection of the verb. Croatian comprises both synthetic and periphrastic past tense forms, while Russian is limited to synthetic past forms. Besides, both Slavic languages possess imperfective present and perfective as well as imperfective synthetic past forms.
The distinction between the perfective and the imperfective aspect is expressed by the choice of a perfective vs. Greek differs from the Slavic languages by its strong grammaticization of verbal aspect. Nearly all Greek verbs oppose a perfective to an imperfective grammatical stem form. While the imperfective present and the perfective and imperfective past is expressed by synthetic verb forms, the future tense and the subjunctive mood are marked by future vs. Thus, while Greek combines a strongly grammaticized aspectual category with a rich tense system, Croatian possesses a rich tense system but less strongly grammaticized aspectual distinctions.
Russian resembles Croatian as far as the grammatical status of aspect is concerned, but has a poorer temporal system. In order to oppose the perfective and the imperfective aspect in the past, Greek uses two grammatical forms of the same lexical item Example 1a whereas, in Russian, this distinction is expressed by the choice of a perfective vs.
In Croatian, it is rare that an imperfective verb may also form a perfective past aorist Example 1c. Normally, perfective or imperfective verbs are used for these two forms, respectively. The data Two children have been investigated for Russian and one child each for Greek and Croatian see Table 1. As with the other children of the Project, tape recordings started with the onset of verbal production and lasted until about the second half of their third year. In everyday speech they are mostly replaced by the periphrastic perfect.
Three hundred utterances each were selected from the first, middle and last month of the observational period. Since the data of the two Russian children and the Croatian girl have been entirely coded, it was possible to include all of it into the computation of mean size of paradigm see section 4. Computation of the other variables is also based on a sample of utterances. Table 2. Liza and Filipp use rather different acquisition strategies.
The Croatian girl, an only child at the time of recording, is growing up in an upper middle class family in Zagreb speaking the Zagreb Stokavian dialect, one of the two major dialects of the Croatian capital. Antonia was recorded in her home during spontaneous interactions with her parents and grandparents. There were about three recording sessions per month lasting about 45 min. The Greek boy Christos was the only child of a Greek upper middle class family at the time of observation growing up monolingually in Athens.
He has been observed from the age of 1;7. The data analyzed for the present study consists of approximately 50 such recordings from 1;7 to 2;6. Independent variables of the input A. Morphological richness Morphological richness is studied both on the paradigmatic and the syntagmatic axis. Paradigmatic richness is determined by the number of form-type categories of nouns and verbs in each of the three languages as well as by mean size of a paradigm calculated 5 Direct repetitions, citations, frozen forms, utterances consisting of yes or no were excluded from the analysis.
Syntagmatic richness is demonstrated by the average number of suffixes of nouns and verbs. Paradigmatic morphological richness The paradigmatic richness of nouns in both the standard language and child-directed speech is much higher in the two Slavic languages than in Greek. The number of grammatical categories occurring in child-directed speech tends to be lower than that of the standard languages. Thus, standard Russian and Croatian possess 12 and 14 case-number forms of nouns, respectively, while only 11 of these are found in the input Table 3. Certain Greek nouns distinguish up to 7 case-number forms, but only 4 of these occur in the input.
The reason is that, in the three languages, the case distinctions of the plural are not fully represented in the input. Besides, in Greek, the inflectionally richest nouns have a minor role to play in the input. Thus, as far as the noun is concerned, the difference between Greek and the two Slavic languages is even greater in the input than in the standard languages. This is especially true of Greek. As with nouns, the number of verbal form-type categories occurring in the input is much lower than that of the three standard languages. The high number of Croatian verb forms as compared to both Russian and Greek is due to the numerous participles that have gender distinctions in both the singular and plural while Russian distinguishes gender only in the singular.
It shows that the caregivers are far from exhausting the inflectional potential of nouns and verbs as evidenced by the input for all lexical items Table 4. Many nouns occur in a single form only. Thus, in Russian, names of food or drinks either occur in the accusative or the genitive partitive and in Greek, many nouns are only used in a kind of all-purpose unmarked nominative-accusative singular or, 6 The table is limited to synthetic verb forms. Thus, Russian imperfective future and Croatian perfect have been omitted in spite of their important role in child-directed speech see fn.
Best English language images in | English classroom, English language, English
Even for verbs, the average number of inflectional forms per lemma is slightly below two for the Slavic languages and only between 2 and 3 for Greek. Table 4. In both Slavic languages, the MSP value is even lower for nouns and is at the same level as in the Germanic languages see Laaha et al. This means that in both language types, most nouns occur in only one or two different forms in spite of the fact that many more case-number forms are distinguished in the Russian and Croatian input 11 case-number forms each than in the Germanic languages.
Syncretism of nominal inflection typical of both Slavic languages cannot explain this state of affairs since, otherwise, the MSP values of verbs without such syncretism should be significantly higher than that of nouns. In Figures 1a and 1b the values indicated in Table 4 are presented in the form of charts. Total input MSP mean size of paradigm in nouns for Russian, Croatian and Greek Figure 1a demonstrates that, in spite of a big difference between the number of noun form-type categories in the two Slavic languages 11 types on the one hand and Greek 4 types on the other, the difference in MSP is very small.
An explanation coming to mind is the high degree of syncretism of case-number forms in the three languages which has been mentioned above. Total input MSP mean size of paradigm in verbs for Russian, Croatian and Greek In verbs, differences in the MSP do not directly reflect numbers of form-type categories found in the input.
Although the number of verbal form-type categories of the two Slavic languages is equal to or higher than that of Greek, MSP is much lower in Russian and Croatian than in Greek Figure 1b. Hybrid forms such as past participles functioning as adjectives, which do not entirely belong to the verbal paradigm, do influence the total number of form-type categories occurring in the input but have a minor role to play in mean size of paradigm. Syntagmatic morphological richness Turning from the paradigmatic to the syntagmatic axis of morphological richness Variable A2 , the verb again rates higher in all three languages than the noun.
While in the two Slavic languages nearly every noun carries a suffix, in Greek two thirds of noun tokens either end in a thematic vowel, not having been counted as an inflectional marker in the present study, or are uninflected foreign names Table 5. The corresponding value for verb forms is nearly the same in the three languages, with the exception of the Russian girl Liza. While inflectionally unmarked noun forms by far prevail in Greek partly for methodological reasons , most noun tokens have one suffix in both Russian and Croatian Figure 2a.
Distribution of suffixes in nouns for Russian, Croatian and Greek Although verb forms may carry two suffixes in all three languages, and in Russian even up to three although these very rarely occur in the input , verb forms with one suffix by far prevail Figure 2b. The higher number of zero endings in Croatian as compared to Russian is explained by the fact that, in Croatian, the third person singular consists of a bare stem.
Distribution of suffixes in verbs for Russian, Croatian and Greek B. Transparency Transparency has been calculated on the basis of a reduced sample of utterances of the input. Form transparency Variable B2 is even higher than word transparency in the three languages, which shows that caretakers seem to prefer transparent forms even when the lemma as a whole is non- transparent. As with nouns, form transparency of verbs is higher than word transparency in Russian and Greek, but not in Croatian, where non-transparent verb forms prevail.
If only lexical verbs VLEX are taken into consideration, with the highly irregular auxiliaries being excluded, form transparency increases slightly in the three languages. Uniformity Uniformly inflected forms Variable C have one and the same marker in all inflectional classes. This characteristic is untypical of inflecting languages. Thus, in both Slavic languages, the percentage of uniformly marked grammatical categories in nouns is zero or close to zero Table 6. The reason is that the only uniform categories e. Russian Dat:Pl and Inst:Pl very rarely occur in the input. As far as the two Slavic languages are concerned, a reverse tendency to use uniform or transparent forms can be observed in the input: While the latter are preferred, the former occur more rarely than in the standard language.
In Greek, nouns show a big percentage of uniformity because most inflected forms are suffixless in the strict sense of the term, i. For determining uniformity entire suffixes or suffix sequences representing bundles of grammatical categories were taken into consideration e. In Greek, entire verb forms comprising an aspect marker and an inflectional ending expressing tense, mood, person and number were taken into consideration. Since aspect markers vary to a much higher degree than inflectional endings, there were no uniform category bundles in the verbs of this language.
Matters would change if the inflectional endings were looked at separately e. Salience To determine phonological-segmental and prosodic salience Variables D1 and D2 only forms with inflectional endings containing a vowel have been taken into consideration. Such forms constitute very different subtotals of nouns and verbs in Greek as compared to Russian. While, in Greek, salience concerns only a small class of nouns7 but a big class of verbs, the reverse is true for Russian.
These vowels have been considered to be non-salient irrespective of the inflectional endings being stressed or unstressed. In Croatian and Greek, all vowels occurring in inflectional suffixes are unreduced. This leads to the values of phonological-segmental salience of inflectional suffixes represented in Table 7. Thus, in Croatian, all inflectional endings are unstressed but salient. Unlike the other independent variables, there is a remarkable difference in prosodic salience of inflectional endings of verbs in the input of the two Russian children.
However, more endings are stressed in verbs than in nouns in Russian. It must be kept in mind that the seemingly opposite situation in Greek is based on very different subtotals of nouns and verbs. However, the functional distinction between syncretistic case forms resulting from vowel reduction do not play an important role in some of their uses at least. In other cases, the functions of syncretistic forms may be disambiguated by prepositions e.
Thus, a low percentage of salient case forms may be quite sufficient from a functional point of view. Since consonants are more important than vowels for distinguishing Russian verb forms we may predict that prosodic salience will be of minor relevance for the development of verb paradigms. Development of mean size of paradigm in nouns for Russian, Croatian and Greek The development of MSP for nouns ranges from 1 to little more than 1.
After that point, development becomes more gradual, except for the Croatian child. The largest difference exists between the Greek boy and the Slavic children, since he stays at an MSP of less than 1. The Russian boy Filipp crosses the 1. It is important to note that the Russian boy and the Croatian girl reach the MSP of the input shortly before 2;9, while the Greek boy attains the input level already before the end of his second year.
This is not surprising, since so little inflectional diversity of nouns exists in the Greek input, namely 4 form-type categories as opposed to 11 in the two Slavic languages. The endpoints of the developmental curves of the three children who were investigated through the second half of their third year roughly coincide with the different MSPs of the input.
While the difference between the development of noun inflection of the Greek child in comparison to the Slavic children may be attributed to differences in the MSP of the input GRK 1. Also, word transparency is much higher in Croatian and there is no vowel reduction. The general conclusion to be drawn from a comparison of the three strongly inflecting languages as far as noun inflection is concerned seems to be the following: whether there is less or more to acquire, children all get off the ground of one form per lemma equally quickly, although they start at different points in time.
It is only after the 1. Development of mean size of paradigm in verbs for Russian, Croatian and Greek The developmental curves of MSP for verbs Figure 4b show both parallels and differences in comparison to those for nouns. All children reach higher MSP values earlier for verbs than for nouns.
Again, the Greek boy differs from the group of children acquiring Slavic languages. But this time, it is his MSP which develops most and also more quickly than that of the Slavic children. As with nouns, the endpoints of the developmental curves more or less coincide with the MSP of the input, except for the Russian girl Liza, who was not observed beyond 2;5. None of the independent variables except MSP can furnish an explanation. MSP is much higher in the Greek input 2. Thus, in the domain of the noun, Croatian ranks before Russian, and Greek is last, both as far as the number of form-type categories and MSP are concerned.
Typological characteristics Turkish, Finnish and Yucatec Maya are agglutinating languages. Turkish has the purest agglutinating morphology whereas Finnish and Yucatec Maya have some inflecting-fusional properties; Yucatec Maya is also polysynthetic to some extent. Turkish and Finnish are nominative-accusative languages whereas Yucatec displays split ergativity. Turkish and Finnish allow for subject and object ellipsis, in Yucatec the corresponding pronouns usually are not omitted.
Turkish and Finnish are characterized by vowel harmony, in Yucatec, however, vowel harmony is limited to certain noun and verb categories. In Turkish, nominals can receive case, number and possessive marking. In predicate position they also get tense-aspect-mood and subject-verb agreement markers. There is no grammatical gender. Case marking is synthetic and stem based. Except for the nominative which has no phonological realization, cases receive distinct morphological marking on nouns, question words, pronouns, and nominalized forms of the verb or the adjective.
The case system consists of a single paradigm that is fully productive and regular. Phonetically homophonous forms in the inflectional paradigm are rare, but exist. The accusative and the 3rd person singular possessive have the same phonological shape -i on nouns ending in consonants, for example.
Verb PNG Images
The inflectional verbal affixes mark negation, tense-aspect-mood, number and person. Turkish derivational morphology is also very rich; voice particles such as the causative or the passive , when present, are interposed between the verb and the tense-aspect-mood morphemes. Each morpheme is syllabic, the typical stress pattern is word-final though there are exceptions to this pattern in the environment of certain suffixes Lees, ; Sezer, among others. Finnish can be typologically characterized as an agglutinating language.
Finnish is also synthetic: derivational and inflectional suffixes are attached to the word stem, e. However, Finnish also displays some inflecting-fusional properties: there are a few portmanteau-morphemes e. The derivational morphology of Finnish is also rich, e. Yucatec has agglutinating, mildly polysynthetic, but also inflectional-fusional properties. Yucatec is a head-marking language: the verbal complex can function on its own as a complete sentential proposition Lucy, Verbs are distinguished between transitive and intransitive verbs according to their argument marking properties.
In the verb complex, person, mood, and aspect are represented by sets of inflectional affixes; non- emphatic personal pronouns do not occur in surface structure, and the lexical arguments can be dropped in the adult language. Both occur within nominal and verbal constructions. In Yucatec, nouns can receive number and possessive marking. Plural marking is optional; the singular is marked by numerals with classifiers. The noun phrase can be modified by demonstratives or by enumeration with numeral classifiers.
In predicate position nouns get tense-aspect-modality and optionally subject-verb agreement markers. The data The data analyzed are taken from one child acquiring Turkish, two children acquiring Finnish and one child acquiring Yucatec Maya as their first language. The Turkish data come from a monolingual Turkish child of middle class background, Deniz, recorded longitudinally between the ages 1;3. Recordings were done approximately twice a month and each of them was about 20 minutes long.
During the sessions, the child was engaged in various natural everyday activities. The Finnish data come from two monolingual Finnish children of middle class background, one boy Tuomas and one girl Tuulikki. Recordings have been made once or twice a month and each of them was about 30 minutes. During the sessions, the child was engaged in everyday activities. The child was recorded at home, twice a week, in interaction with his mother, grandmother, aunt, the observer and other children.
For the input, three sessions corresponding to three developmental points were sampled. Table 1 presents the age ranges of the children studied, as well as the number of noun and verb tokens in the output. Furthermore, since younger children are often taken care of by their older siblings, the Yucatec input data consist of few adult expressions as well as input of siblings older than 12 years. Paradigmatic morphological richness In terms of the paradigmatic morphological richness of the input data, the three languages differ with respect to the number of non-homophonous form-type categories.
In nouns, Yucatec has the least number of categories, only 5. Turkish and Finnish are about equally rich with 15 and 14 noun categories, respectively. In verbs, there are more inflectional categories than in nouns in each of the three languages. Turkish has the highest number of categories, Finnish has the least, 19, and Yucatec is in between with 30 categories. The categories delineated in each language are presented in Table 3 below. It is larger for verbs than for nouns in all the three languages. These findings are presented graphically in the following figures Figures 1a, 1b.
Syntagmatic morphological richness We now turn to syntagmatic morphological richness of the input data and first present the distribution of suffixes in nouns and verbs. In Finnish, suffixless forms are singular nominatives. It is observed that Turkish presents the richest input both for nouns and verbs, Finnish and Yucatec are comparable in terms of the number of suffixes on verbs whereas Turkish and Finnish are more similar in case of nouns.
Transparency The three languages differ in terms of the morphophonological processes that affect word and form transparency Variables B1 and B2. But these stem alternations do not occur very frequently in the input, and form transparency is again higher than word transparency, yet not as high as in nouns. Uniformity Turkish has no non-uniform noun categories.
The non-uniform categories of Yucatec are the possessive and inalienable. In verbs, Turkish displays some level of non-uniformity, particularly due to the behaviour of the aorist inflection. Non-uniformity is more pervasive both in Finnish and Yucatec. The verbal categories that show non-uniformity in Finnish are 3SG, passive, past participles, past, 1st and 2nd infinitives. In sum, in Yucatec there is a high proportion of non-uniform categories for verbs, in Finnish the verbal form categories are predominantly non-uniform.
The percentage of uniform categories Variable C in the three languages is given in Table 6. Salience As far as phonological-segmental salience Variable D1 is concerned, there is no vowel reduction in suffixes in Turkish, but in Finnish and in Yucatec some vowel reduction occurs both in nouns and in verbs see Table 7. In Turkish stress is mostly word final; deviations are mostly due to the clitic nature of some of the suffixes which cause the primary stress in the word to shift to the syllable preceding it.
So the suffixes in both languages are prosodically salient. However in Finnish stress is word-initial and the suffixes are prosodically non-salient see Table 8. Development of mean size of paradigm in child speech The characteristics that have been summarized above for the three languages constitute our independent variables. The dependent variable is the speed of development. Below we present the curves representing the development of MSP for nouns Figure 4a , and for verbs Figure 4b.
Development of mean size of paradigm in nouns for Turkish, Finnish and Yucatec It is observed that the curves for the development of MSP in nouns for the three languages have almost the same shape. Each curve rises steeply at first indicating a fast development at the beginning, until each child reaches MSP 1. After this point, development is slower for Finnish and particularly for Yucatec as compared to Turkish.
Development of mean size of paradigm in verbs for Turkish, Finnish and Yucatec For verbs, all independent variables related to morphological richness make Yucatec and Finnish more similar to one another as compared to Turkish. A similar grouping of the languages is observed with respect to the dependent variable: the developmental curve for mean size of paradigm for verbs is far steeper for Turkish than for Finnish and Yucatec, showing faster development.
To conclude: the development of nouns is more uniform across the three languages whereas the child acquiring Turkish goes ahead for verbs. The noun paradigms have more limited size than the verb paradigms, and this may result in a more similar pattern of acquisition in nouns than in verbs. General results Aris Xanthos 6. Introduction In this chapter, we will discuss the results of this research at a more general level; in particular, we will examine our main hypothesis, i.
We applied two standard statistical tools to the data, namely analysis of variance ANOVA and correlation analysis. The former was used to ensure that observed differences in speed of development between languages and language groups were significant; the latter specifically addresses the hypothesized correlation. Each method and its results will be discussed in a separate section. Also, a more thorough discussion of the presented results will be deferred to the concluding chapter Dressler et al. Analysis of variance 6. Although this does not directly document our main hypothesis, it sheds some light on the cross-linguistic and typological relevance of the dependent variable, speed of development.
The individuals entered into this analysis were monthly samples, characterized by their score for speed of development of MSP nouns and verbs. Figures 2a and 2b show the corresponding values obtained by clustering languages into weakly inflecting, strongly inflecting, and agglutinating language groups. We examined the dependence between speed of development and each factor language and language group using Welch's variant of ANOVA Welch, , which does not rely on the assumption of the homogeneity of variances.
In particular, we performed 4 separate one-way ANOVAs: language 9 and language group 3 factors were examined separately, and this was done for nouns and verbs separately. Mean speed of development of MSP for each language: nouns 0. Mean speed of development of MSP for each language group: nouns 0.
- Area guide book Taipei Zhongshan (Japanese Edition).
- Classic Parts.
- German Present Tense - Verb wheel template.
As can be seen, the pattern of results differs between nouns and verbs. P 1 2 Language Nouns The degree of this dependence varies in a way that corresponds to the intuitive examination of Figures 1a to 2b. Nouns display a more contrasted profile than verbs, and a better correspondence with typological expectations in relation to our main hypothesis. At this point, it is difficult to say whether this diverging behavior of nouns and verbs should be explained by other factors than the particular sample of languages and children that was used.
Nevertheless, more differences between the two subsystems will be reported in the next section. Correlation analysis 6. Method The second analysis was aimed at determining how well each of the seven independent variables A1: paradigmatic morphological richness, A2: syntagmatic morphological richness, B1: word transparency, B2: form transparency, C: uniformity, D1: phonological-segmental salience, D2: prosodic salience accounted for the observed differences in speed of development of MSP.
This time, the individuals were whole corpora as opposed to monthly samples. Pearson correlation coefficient r is by far the most widely used measure of correlation. However, it is highly sensitive to the presence of outliers individuals located far from the rest of the data , to the point that the addition of a single outlier to an otherwise random sample can yield a very high correlation coefficient. Instead of that, we used Spearman correlation coefficient rS , which is defined as the Pearson correlation coefficient between the ranks corresponding to the original scores.
Results The correlation coefficients between each independent variable and speed of development are plotted on Figures 3a nouns and 3b verbs below, where hatched bars represent values that are not significant at the 0. Spearman correlation between each independent variable and speed of development of MSP: nouns 1 Spearman correlation coefficient 0.
Correlation analysis results Nouns Verbs Independent variable Corr. This is a strong confirmation of the hypothesized relationship between speed of development and the degree of morphological richness in the input. In the case of verbs, no other independent variable is significantly correlated with speed. Therefore, it is possible that the correlation of one or the other with speed is a by-product of their relationship with A1.
Bavaud, We first considered the case of variable A2. We conducted the same verification with variable B2. The results are very similar, i. Thus, it is likely that, here as well, A1 is the underlying factor. As was noted in the previous section, it is not clear whether the differences found between the nominal and verbal subsystems would be confirmed using another or a larger sample of children and languages.
From a purely statistical point of view, given the less contrasted values measured for speed of development for verbs, it is not a surprise that the corresponding correlations with independent variables are generally weaker. Nonetheless, with regard to the data at hand, we may state that only variable A1 paradigmatic morphological richness is consistently correlated with speed of development. Summary In this chapter, we have surveyed the results of the general analyses that were applied to our data.
Using Welch ANOVA, we found that the speed of development of MSP varied significantly as a function of the language and language group factors, though this effect is stronger for nouns than for verbs, and for the first factor than for the second one. A second analysis focused on the Spearman correlation between speed of development and each independent variable at the level of corpora. It revealed that speed was highly correlated with variable A1 paradigmatic morphological richness , both for nouns and for verbs, as predicted by our main hypothesis.
For nouns, other significant correlations were found for variables A2 syntagmatic morphological richness and B2 form transparency ; however, the examination of partial correlations suggests that these may be spurious effects of the relationship of A2 and B2 with A1. This hypothesis has been embedded in a typological framework which considers noun and verb inflection as entities of a two-fold comparison: 1 a comparison of nine languages according to three different language types, 2 a comparison of the morphological subsystems of noun inflection and verb inflection.
In the preceding chapter Xanthos, this volume classificatory typology has been used to relate the three language types of weakly inflecting, strongly inflecting and agglutinating languages to speed of development, i. The expected significant differences between language groups have been found with respect to speed of development, i. It must be kept in mind, however, that these results only represent a first approximation, due to the oversimplified concept of typological classes of languages see Dressler, this volume , which informs us only about cross-class differences, but not about cross-language differences.
In order to compare all nine languages of our sample individually, they were ordered following the model of ordering typology, namely, according to the degree to which they display characteristics of the ideal agglutinating, inflecting and isolating types. To make sure your classic car will always be in good shape we remanufacture highly demanded parts.
That makes us able to offer thousands of new parts in perfect original quality. Wear part in particular like tooth- or fan belts and components regarding the brake are constantly completed in our assortment by well proven partners who are or were dealers of initial-equipment. Since these spare parts are strictly checked they are delivered by us using the unopened original manufacturers packaging.
Even if original or initial-equipment spare parts are not available anymore we make sure to help you with the part you need. We have the best quality in offer taken out of other outlets. Here again we send these parts in the original providers packaging to show our transparency. It helps us to understand what kind of content you might like and to show advertising messages that could be interesting for you. Here you can always withdraw your agreement. Volkswagen Classic Parts Search: Search.
My Wishlist. Cart 0. Recently added item s You have no items in your shopping cart. Business Customer Private Customer. One step back. Product details. VAT excl.