1.1 Theoretical framework – Construction grammar

Young children learn the set of linguistic conventions used around them through interactions with members of their community. These conventions consist of hundreds of thousands of words and constructions, including the abstract constructional patterns that speakers of each language have grammaticalized.

Scientists have differed in their accounts of this acquisition process (see introduction to this Special Issue). In more recent years, a number of studies in developmental psychology, linguistics and cognitive science have suggested that children have at their disposal very powerful learning mechanisms integrated with social-cognitive skills that enable them to learn from the input. Theories representing this viewpoint come mainly from cognitive or functional linguistics and take a usage-based approach to language development. Such approaches take language structure to emerge from language use (Langacker 1987; Croft 1991; Goldberg 1995; Givon 1995; Bybee Reference Bybee1985; Tomasello Reference Tomasello2003). Adult linguistic competence is thus viewed as based on concrete pieces of language that undergo generalizations. Children are seen as using meaning and function to acquire linguistic constructions in context. Crucially, linguistic competence does not correspond to the mastery of a formal grammar based on semantically empty rules, but rather to the mastery of an inventory of meaningful linguistic constructions that each pair meaning with form. For Tomasello (Reference Tomasello2003), toward the beginning of the acquisition process these constructions are dominated by isolated, lexically specific, ‘item-based’ constructions. Acquiring grammar is viewed as the gradual generalization of such item-based constructions into more abstract patterns based on communicative function. More concretely, children first extract tangible elements (words, morphemes, unanalyzed strings, constructions) by identifying their communicative function; then, based on recurring items with similar functions across various situations in the input, they create more or less abstract categories and constructions.

In this study of the acquisition of constructions marking possession, we will illustrate how at the beginning of acquisition, some of the children's productions do not resemble adult language. Rather, the actual structures and representations involved in children's productions seem to change across different stages of the acquisition process. The main factors affecting language development in our view are 1) communicative intention, 2) frequency and saliency in the input, but also 3) children's very own affective, social and practical concerns. Those three main factors come into play to various degrees according to the specific linguistic item that is being acquired.

The development of the marking of possession serves as an especially interesting locus for a study of the way these factors intermingle. As shown by Marcel Cohen (Reference Cohen1924), children seem to construct their own transitory systems in the course of language acquisition. Since the beginning of the 20^th century, a number of scientists have tried to explain the mismatches between the input and some of the children's peculiar productions. Karmiloff-Smith (Reference Karmiloff-Smith1986) makes a theoretical proposal in terms of continual revision of form-function mappings illustrated by examples of how the child's system does not directly reflect the characteristics of the input. Children produce a number of nonstandard forms at the beginning of the learning process (Rondal et al. Reference Rondal, Esperet, Gombert, Thibaut, Comblain, Rondal and Seron2000); these are typical of what Clark (Reference Clark, Bowerman and Levinson2001) has named ‘emergent categories’. Some of those forms illustrate how children are able to move beyond frozen expressions thanks to productive analyses of the input, and might create non-adult constructions in the process of learning form-function pairings.

In the course of this study of a French-speaking child's forms used to mark possession, we will analyze her ‘errors’ or nonstandard forms, which have been referred to as ‘barbarisms’ by Egger (Reference Egger1879) or ‘incorrect forms’ by Buhler (Reference Bühler1926). We view these forms as revealing signs of her cognitive processing of the grammatical categories and the semantic features at play in the marking of possession.

1.2 Expressing possession in French

Even though possession is quite a stable notion cross-culturally, and all languages seem to have means to encode that notion linguistically, there is no universal linguistic structure common to all possessive constructions. These constructions often express other concepts, and various forms used to express other concepts might be borrowed to express possession (Heine, Reference Heine1995). It is therefore quite difficult to analyze the use of possessive constructions in a large corpus (especially spontaneous adult-child conversations), since automated searches are not easy when the forms are diverse, unless the notion of possession is specifically tagged.

Possession in French may be marked in many ways, either locally in nominal expressions (attributive possession) or more globally in the utterance using a predicate, and particularly the verb avoir / to have (predicative possession). In this paper, we will focus on attributive possession. Expressions involving mere juxtaposition are much less frequent in French than in English (as in New York buildings). Just as in English, however, French has not grammaticalized the distinction between inalienable (inherent) possession (son nez – her nose) and alienable (non-inherent) possession (sa voiture – her car).

For third-person possessors, there are two competing structures in French: [Noun1 + preposition de + Noun 2] (le sac de maman – mummy's bag) and [Possessive Determiner + Noun] (son sac – her bag). The selection of the construction depends on the discourse status of the possessor. If the possessor has been mentioned (given), the speaker will use [Possessive Determiner + N] (sa voiture – her car). If the information is not known (new), the speaker will use [N1 de N2] (la voiture de Martial – Martial's car). A third grammatical marking of possession is possible, with the use of an anaphoric possessive pronoun (la sienne - hers), when not only the possessor, but also the possessed object, are given information in the discourse context, though this pattern is less frequent in adult data.

For first and second-person possessors, the construction used is [Possessive Determiner + Noun] as in ma maison (my house), ton sac (your bag), since the reference to the possessor is constructed deictically. An emphasis can be expressed with the addition of [à + tonic pronoun] to the [Possessive Determiner + noun] structure, as in ton papa à toi (your own dad). This phrase can also stand by itself: the answer to questions such as c'est à qui? (whose is it?) can be à moi (it's mine).

In French, as opposed to English, agreement of possessive determiners marks both the possessed object and the possessor. These determiners must agree in number and gender with the noun they determine (referring to the possessed object/s) and in person with the possessor(s). According to Rondal et al. (Reference Rondal, Esperet, Gombert, Thibaut, Comblain, Rondal and Seron2000), their use is guided by an economy principle that reduces expressions such as ‘X belongs to Y’ or ‘X is part of Y’ to [Poss Det + X]. The choice of a possessive determiner is the result of four combined operations: 1) the selection of personal reference according to the interpersonal context of the verbal exchange (first person when the speaker is the possessor, second person when the interlocutor is the possessor, third person when the possessor is neither the speaker nor the interlocutor); 2) the selection of the agreement in gender and number according to the gender and number of the possessed object; 3) the selection of the agreement in number according to the number of the possessor (singular or plural); 4) the initial phoneme of the possessed element (in son amie, the possessive determiner is masculine because the noun begins with a vowel).

Possessive determiners can therefore be analyzed as an amalgam of several grammatical and phonological features. They seem to be simple markers on the surface but are quite complex at different levels.

1.3 How children acquire the marking of possession

Children start talking about possession from an early age (Brown, Reference Brown1973), for it plays a central role in their daily lives (Eisenbeiss et al., Reference Eisenbeiss, Matsuo, Sonnenstuhl and Mc Gregor2009). Slobin (Reference Slobin and Slobin1985) suggests that they start out with a higher-level category of location that includes possession. But Bowerman (Reference Bowerman and Slobin1985) argues that nothing proves which of location or possession is more basic. Children might focus on either of these semantic relations at different times.

Since French subdivides the semantic notion of possession into several constructions according to the various factors identified above (discourse status, person, gender, number), children learning French often produce ‘errors’ while in the process of sorting out these factors (Clark, Reference Clark, Bowerman and Levinson2001). More precisely, children have a preference for seeking consistent ways to mark certain distinctions as they create their own form-function pairings (Slobin, Reference Slobin and Slobin1985; Budwig, Reference Budwig1995). They may thus pick a single construction to express all instances of a category they mean to target. This may lead them to overgeneralization errors before they acquire all the conventional forms for that category and learn how to express finer distinctions. Besides, constructions might be acquired in stages with omission of certain grammatical markers. This leads to paratactic uses such as chemise papa (shirt daddy) instead of la chemise de papa (daddy's shirt).

According to Rondal et al. (Reference Rondal, Esperet, Gombert, Thibaut, Comblain, Rondal and Seron2000), French-speaking children don't use possessive determiners at first; they express possession in three different ways:

• – (preposition à) + possessor: à moi (mine)

• – (article) + possessed object + (preposition à/de) + possessor: la balle de Dédé (Dédé's ball)

• – possessor + possessed object: papa bic (daddy pen)

The construction [N + à moi] is quite frequent in some registers of French used in families from the working class (Grégoire, Reference Grégoire1937), but the child in the data we have collected is not surrounded by that kind of input. Those uses are not standard in her environment and could be considered to express possession as an emergent category (a single construction generalized to nonstandard uses). However, since children are so sensitive to the input, these non-standard [N + à moi] constructions are transitory and are soon replaced by the adult forms [N1 de N2] and [Poss Det + N].

In this paper, we examine the developmental pattern in our longitudinal data and compare it to the relatively rare mentions of the acquisition of possessive markers in the literature, as well as to the marking of possession in the child's maternal input. We mostly focus on the nonstandard forms used by the childFootnote ¹ and on her acquisition of the conventional standard alternatives.

Table 1.

Example of Coding

2.1 Data and Method

This paper focuses on Anaé’s longitudinal data (see the description of the Paris corpus, this issue). We used the three types of data at our disposal to study Anaé's linguistic development: 1) transcriptions in CHAT format from 1;06 to 4;0 linked to video; 2) the mother's diary from birth to 4;00; 3) detailed coding of eleven one-hour sessions from 1;07 to 2;11.

We coded all utterances that contained a construction marking possession. In the child's language, we found the following standard constructions embedded in her utterances:

• – [Poss Det + N]: possessive determiner + Noun (sa maison – her house)

• – [N1 de N2]: Noun 1 + preposition de + Noun 2 (la maison de Dora – Dora's house)

We also found the following constructions in isolation:

• – C'est à + tonic pronoun (c'est à moi – it's mine)

• – A + tonic pronoun (à moi - mine).

Very few of these constructions are used in the second and third person. Nonstandard constructions were also coded:

• – Le N de moi / Le N à moi

This construction was only produced in the first person.

The coding was done using Excel, based on the following features:

1. 1) Dataset, 2) Speaker, 3) The whole utterance, 4) Presence of a possessive structure (Yes/No), 5) Noun determined, 6) Possessive structure, 7) Concrete possessive phrase, 8) Personal reference of the possessor.

Examples such as la poche de ma maman (my mummy's pocket) were coded twice, once for the [N1 de N2] structure, once for the [Poss Det + N structure] (ma maman – my mummy).

2.2 Distribution of the various structures in the data

2.2.1 Possessive structures in the input

Our coding enabled us to count the number of occurrences of each construction in the input, in order to compare mother and child's uses. Overall, we found 271 occurrences of possessive constructions in the child's verbal productions and 552 in the mother's input in the recordings between 1;07 and 2;11.

We can observe in Figure 1 that there are two types of constructions in the mother's input, where the structure [Poss Det + Noun] is predominant. The total rate of [Poss Det + N] in the mother's productions in the eleven one-hour sessions is 91.5%. The construction [N1 de N2] is only used to refer to third-person discourse objects that are introduced for the first time in the discourse (new information), are not the focus of attention, and are not particularly emphasized for contrastive reasons, as illustrated in the examples below.

1. (1) - Anaé 2;05

   MOT: C'est Martial et Magali qui te l'avaient offert, le papa et la maman de Sacha. [Martial and Magali gave it to you, Sacha's Mummy and Daddy.Footnote ²]

Figure 1.

Distribution of possessive constructions in the input

In (1), the mother introduces the family relationship between Sacha and the two discourse objects mentioned in the first part of the utterance. The relation constructed with [N1 de N2] enables the mother to clarify who the discourse-objects are, and to locate them according to a discourse-object whose identification seems to be easier for Anaé. Sacha is a friend she knows very well, but she probably does not remember the names of his parents.

1. (2) - Anaé 2;05

   MOT: le bébé kangourou, il est dans la poche de sa maman. [the baby kangaroo is in his mummy's pocket.]

In (2), there are two possessive constructions in the same utterance 1) la poche de sa maman (his mummy's Pocket) (N1 de N2) and 2) sa maman (his mummy) (Poss Det + N). In the first construction, the relationship between poche (pocket) and maman (mummy) is presented for the first time in discourse. It is new information and gives the location of ‘bébé kangourou’. The second construction is made possible because the discourse-object has been mentioned in the first part of the utterance and is pointed back to in the second part with the third-person possessive determiner sa (his), which also plays an anaphoric function.

In six occurrences of the mother's 367 uses of the construction with a possessive determiner (0.01%), she adds the structure [à + tonic pronoun], which expresses a contrastive value.

1. (3) - Anaé 2;03

   The mother is moving Anaé's toy horse. The other horses belong to her brothers.

   MOT: Voilà ton cheval à toi qui arrive. [look, your own horse is coming.]

The marking of possession in the mother's input is therefore quite straightforward and follows the rules of the French grammatical system, with some rare occurrences of a construction specific to spoken French to mark contrastive possession.

2.2.2 Possessive structures in the child's data

As shown in Figure 2, possession is marked quite early in Anaé’s data and with a variety of forms. There is only one case of a paratactic juxtaposition of two nouns, which are linked semantically by possession in the data. At 1;09, Anaé holds out a plate to her mother (who is setting the table) and says c'est l'assiette (pause) Ael (It's Ael's plate, Ael is one of her elder brothers). Juxtaposition therefore is not a privileged means to express possession, even though the child does juxtapose nouns for other types of relations (carte Dora meaning a card displaying a picture of the character Dora; sirop pêche instead of ‘sirop de pêche’, meaning peach-flavored syrup).

Figure 2.

Distribution of possessive constructions in the child's productions

The child therefore mainly expresses possession with overt forms and in the first person (there is only one instance of à toi – yours, and one instance of à elle – hers - throughout the corpus, with a possessive sense). She uses the construction à moi and [N + à moi] a lot until 2;03. This construction is probably derived from the massive use of à moi in the input (by her mother, but also by her two elder brothers) in turn-taking situations (where ‘à moi’ means ‘my turn’). We did not consider those uses to be marking possession, even though these recurrent situations are strongly related to the marking of identity (and otherness) expressed in possessive relations. Example (4) illustrates the nonstandard possessive use of [N + à moi].

1. (4) - Anaé 2;00: c'est la puzzle à moi [it's my puzzle].

The recording at 2;00 is particularly interesting to study the blossoming of several constructions through Anaé's production of various nonstandard forms. In (4), on top of the nonstandard use of [N + à moi], the gender error (la puzzle instead of ‘le puzzle’) may be a sign that Anaé is constructing her utterances without always rigidly replicating the input and makes the wrong choice in a paradigm of determiners. Anaé proves to be particularly prone to gender errors (Leroy-Collombel, Reference Leroy, Neveu, Muni Toke, Durand, Klingler, Mondada and Prévost2010) such as un fleur (a flower), le poupée (the doll), le poule (the chicken), la ballon (the ball), une oiseau (a bird) but with non-systematic gender assignment.

Once possessive determiners are acquired (at 2;02, when the child's MLU is 3.0), they become dominant, as is the case in the input. Anaé's uses of possessive constructions are therefore quite similar to Brown's description of the use of possessive constructions in English. Genitive constructions and possessive determiners are stabilized in the course of what Brown refers to as ‘stage 3’, (between MLU 2.5 and 3) (Brown, Reference Brown1973). However a few occurrences of nonstandard forms such as [N + de moi] are also produced such as example (5) and (6).

1. (5) - Anaé 2;08: le caddie de moi il était bougé aussi. [my shopping cart was also moved.]

2. (6) - Anaé 2;08: ça c'est mon truc de moi [that's my own thing.]

In example (6), Anaé superimposes the [Poss Det + N] structure (mon truc – my thing) and the nonstandard [N + de moi] structure used to emphasize possession. This might also demonstrate that she has not entirely analyzed the role of mon (my) as being both a masculine determiner and a marker of first-person possession, so that she resorts to a second grammatical marker to express her intended meaning. There are two occurrences of such a phenomenon: one at 2;08 and one at 2;11.

The [N1 de N2] construction emerges at 2;02 and is used for all first mentions with a third-person possessor. The possessor can eventually (but rarely) be the child herself, in constructions such as la tête de Anaé (Anaé's head). As shown by Caët (this issue), the child sometimes uses her own name to designate herself in specific pragmatic conditions, which creates a certain distance between herself as a speaker and herself as a discourse-object. This particular case is interesting, because when she produces this utterance, she is playing a game where she is alternatively hiding and showing her head. This use of her own name in a possessive construction enables her to take on the other's perspective and might also, as Deutsch and Budwig (Reference Deutsch and Budwig1983) note for English, mark less ‘pragmatic control’ as opposed to the use of my.

At 2;02, [N1 de N2] alternates with [Poss Det + N] with a very clear advantage for the latter - 92.3% in the whole longitudinal data. There are 5.8% [N1 de N2] constructions and 1.9% à moi and c'est à moi productions in isolation. The nonstandard productions are quite frequent up to 2;00 and then become sporadic in the rest of the data.

It is interesting to note that when Anaé starts making a productive use of possessive determiners at 2;02, out of the 17 different nouns she produces with those determiners, 10 are also used in other contexts either bare (with no determiner), or with filler syllables, or with other determiners. The possessive determiner may therefore very well be chosen in a paradigm of alternative determiners, and not learned with the noun as a frozen expression or a ‘noun island’.

1. (7) - 2;02

   mon bébé / le bébé / ce bébé / [e] Footnote ³bébé (my baby / the baby / this baby / [e] baby)

   mon chien / le chien / ce chien (my dog / the dog / this dog)

In the same recording, the mother uses five different determiners (ton, le, un, les and ces) with bébé and seven different determiners (le, mon, ce, son, un, des and les) with chien. It is therefore interesting to note that at 2;02, similarly to her mother, the child already has a paradigm of determiners at her disposal for a number of nouns. She has moved beyond frozen expressions (a noun always used with the same determiner) and shows productivity and a sense of grammatical categories in her use of determiners in nominal expressions.

2.3 Possessive determiners among nominal determination markers

In the mother's data (Figure 3), definite articles are used predominantly (with an average of 53%), then come indefinite articles (29%), possessive determiners (14.6%) and very few demonstrative determiners (3%). Although the distribution and number of determiners depend on context (for example, there are more indefinite articles when looking at picture books and labeling objects), the number of possessive determiners (solid grey line in Figure 4) remains relatively constant throughout the data.

Figure 3.

Percentage of occurrences and types of determiners in the mother's utterances

Figure 4.

Percentage of occurrences and types of determiners in the child's utterances

Even at the end of the data, as shown in Figure 4, the child uses fewer tokens than her mother (73 tokens at 1;10 when the mother uses 482, and, at 3;01, 146 tokens for the child and 229 for the mother). The child uses filler syllables up to the age of 2;03. As soon as they emerge at the age of 2;00, she predominantly and constantly uses definite articles, just like her mother (53.4% of all her determiners throughout the data are definite articles). The number of indefinite articles increases at 2;05, a session in which they are more numerous than definite articles (possibly due to the prevalence of a labeling activity). Possessive determiners start being used a little bit later and stabilize at 2;02 with an average of 12%, which is quite similar to adult use. Demonstrative determiners are very infrequent and only start being used as of 2;05 (1% of the overall use of determiners). The child's profile is therefore very similar to her mother's. However, she has a tendency to use infrequent forms even less often than her mother, which is quite typical of that age range (Morgenstern et. al. Reference Morgenstern, Parisse and Sekali2009).

2.4 Personal reference of possessors in the data

In the mother's data, as shown in Figure 5, there are very few mentions of first-person possessors (7.2%). Second-person references are used most (54.4% overall). She therefore speaks mostly of her child as possessor. Many of the third-person references (which total 38.4%) are made during games with various dolls and figurines, or during book reading, but family members and friends are also mentioned. Very few references in first, second and third person are plural possessive determiners (16 occurrences out of 680 nouns, which represent 0.02%).

Figure 5.

Personal reference of possessors in the input

The distribution is very different in the child's data (Figure 6). There is a majority of reference to self as possessor (45.3%), but mostly in the first half of the data. Reference to the addressee as possessor is quite low (5.8%) and a parallel could be made with reference to the addressee as grammatical subject (see Caët, this issue). The rate of reference to third-person possessors is quite similar to the rate found in the input: both mother and daughter refer a lot to toys and characters in books, friends and family. There were no uses of plural possessive determiners before three.Footnote ⁴

Figure 6.

Personal reference of possessors in the child's productions

As far as personal reference of possessives is concerned, the child thus does not directly replicate the frequency of the actual forms observed in the input. However, it should be clear that both the mother and the child refer predominantly to the child as possessor. A pragmatic feature – focus on the child in adult-child interaction – therefore influences the person of expressed possessors. Personal reference can thus be considered ‘usage-based’ in the sense that it reflects the typical pragmatic experience in which possession is expressed in interaction.

4.1 Determiners

As noted earlier, possessive determiners in adult French mark possession, gender, number and person in one grammatical form. This amalgam of several features might make it difficult for children to handle those unstressed monosyllabic grammatical words. Anaé's analysis of gender marking is quite visible when she makes repairs of her own productions, as in le son son sa queue (the, his, her tail) or il met son son sa robe (he's putting on his his, her dress) at 2;02 and Timothé i veut pas me prêter mon v/, ton v/ le vélo de Timothé (Timothé doesn't want to lend me my, your, Timothé's bike) at 2;10.

Possessive determiners are first used at 1;08 in frozen expressions (tatête – yourhead – to designate the toy's head is not yet analyzed as two separate elements and the wrong person, second instead of third, is used). At 1;10, there is a very interesting use of donne ma main (give my hand) when she means ‘give me your hand’ or ‘hold my hand’ as if ma was used to amalgamate the role of patient (instead of using moi) and the possessive relation. During the same session, she also makes the opposite mismatch, saying toi chaussure (you shoe) to her father as she tries to take his shoes off, instead of ‘ta chaussure’ (your shoe). The handling of those grammatical markers thus seems to be quite confusing, since they amalgamate several functions in one term. At 2;00, there are two ‘correct’ occurrences of first-person possessive determiners (mon jeu – my game, and mon papa – my dad). At 2;01 Anaé starts using a third-person possessive determiner (sa maman – his mummy) but it is only at 2;02 that possessive determiners are used very productively, and ma, mon, mes take on the function that was marked by her nonstandard use of [Noun + à moi], which totally disappears from the data. At 2;02, there are 27 occurrences of possessive determiners with 17 different nouns. Some nouns appear with third- and first-person possessive determiners, both singular and plural (mes doudous, son doudou). Anaé makes sporadic uses of [Noun + de moi] up to 2;11. At 2;08, an interesting occurrence of mon truc de moi (my thing of me) shows that she might not consider possessive determiners to be semantically strong enough to mark nominal determination, reference to a first-person possessor, masculine gender and possession all at once, and therefore needs to resort to the nonstandard de moi to emphasize the identity of the possessor. This type of production, which is not found in adult language (though it may be derived from the [N1 de N2] construction), demonstrates how the little girl ascribes functions to forms that she has picked up in her analysis of the input. These types of unusual uses can be seen as ‘emergent categories’ (Clark Reference Clark, Bowerman and Levinson2001) and are transitional forms children use while in the process of acquiring the target linguistic forms. In acquiring their language, children must attend to all grammatical distinctions relevant in their own language, and might not employ them multifunctionally as shown for instance in Karmiloff-Smith (Reference Karmiloff-Smith1979)'s work on the acquisition of determiners. When several grammatical features are marked by the same term, this could lead to some confusion about which features are marked by what grammatical element, as shown by Anaé when she says mon truc de moi (my thing of me).

As illustrated in Figures 5 and 6, the distribution of first-, second- and third-person reference differs dramatically between Anaé and her mother, with very high percentages of first-person possessive determiners in Anaé's data.

It is also interesting to note that Anaé starts using possessive pronouns (le mien, le tien, le sien- mine, yours, his) later on, after the end of the data coded for this paper, in the mother's diary. A pronoun such as le mien (mine), for example when referring to one's dog, amalgamates reference to the noun (with its gender and number) and to the possessive determiner (with its person). We found examples in the diary such as j'ai trouvé ton mien (I found your mine) at 3;00 as she speaks about her brother's toy or votre mien (your mine) referring to her parents’ bed. In these examples, she seems to use the first-person pronoun with only one function (possession), instead of amalgamating person and possession. She therefore adds another grammatical marker to indicate person.

4.2 The [N1 de N2] construction

The ‘correct’ use of the [N1 de N2] construction appears only at 2;02, at the same time as the productive use of possessive determiners. At 2;01, there is one nonstandard production in the data with the preposition ‘à’ instead of the preposition ‘de’ (les épaules à papa – the shoulders of daddy). At 2;02, the session in which the possessive determiner becomes predominant, there are two occurrences (la tête de monsieur – the head of mister and la tête de Anaé – the head of Anaé). The child designates herself with her name instead of the deictic determiner ma, and treats the reference to herself in the same way as the reference to ‘monsieur’.

Her uses are more numerous and diversified as of 2;05. Her utterances are like building blocks made of several different pieces (at 2;05 dans le lit de moi – in the bed of me; at 2;06 le bébé de ours brun – the baby of brown bear; les poils de le bébé – the hair of the baby; at 2;11 des petites poils de Omer – the little hairs of Omer). The construction [N1 de N2] is therefore an excellent locus to observe the various nonstandard forms a child can create in the process of acquiring language. Since it is made of several blocks (nouns, determiners, a preposition), there is ample opportunity for creativity on the part of the child.