One common difference between descriptive grammars written by linguists and pedagogical grammars is how they use technical terminology. For instance, in French textbooks, the term adjectifs possessifs ‘possessive adjectives’ is commonly used. This might suggest to students that these words are actually adjectives. In reality, they are determiners. Hence, we will be calling them possessive determiners to be consistent with our final conclusion. We can analyze how adjectives, determiners, and possessive determiners are used in sentences to demonstrate this.

Before showing why possessive determiners are not adjectives, it might be useful to explain why someone might think that they are adjectives in the first place. First, adjectives in French have to agree in number (singular vs. plural in French) with the noun that they modify. When they fail to agree, the result is an ungrammatical phrase. As shown in (1), since the noun personnage ‘character’ is singular, adjective principal ‘main’ in its singular form can modify it. However, if we try using the plural form, principaux ‘main’, the result is an ungrammatical phrase.

The same is true of possessive determiners. They must also agree in number with the noun that they modify. In (2), if we use singular possessive determiner mon ‘my’ with the singular noun stylo ‘pen’, the result is grammatical. Again, however, if we try using plural possessive determiner mes ‘my’ with the singular noun, we get an ungrammatical phrase.

Additionally, both adjectives and possessive determiners need to agree in noun class (masculine vs. feminine in French) as well. As shown in (3)^[1], the feminine noun catin ‘doll’ must go with feminine possessive determiner ma ‘my’ and feminine adjective blonde ‘blonde’. If either is replaced with its masculine variant, mon ‘my’ and blond ‘blond’, then the phrase is ungrammatical.

In terms of agreement, then, adjectives and possessive determiners are quite similar and so it is not unreasonable for someone to group the two together in some way. However, other determiners, such as articles, must also agree in number and noun class with their corresponding noun.

For number agreement, in (4), singular noun cheval ‘horse’ must go with singular determiner le ‘the’ while plural noun chevaux ‘horses’ must go with plural determiner les ‘the’.

For noun class agreement, in (5), masculine noun livre ‘book’ can co-occur with masculine determiner le ‘the’ but not feminine determiner la ‘the’.

At this stage, adjectives, determiners, and possessive determiners are all behaving the same way. Intuitively, however, speakers of French know that determiners and adjectives do not belong to the same category. As such, possessive determiners can only logically be adjectives, determiners, or some third category.

One way to test the category of possessive determiners is to look at ways in which adjectives and determiners behave differently and then see if possessive determiners act more like adjectives, more like determiners, or in a way different from each of the two.

The first test concerns the relative word order of adjectives, determiners, and possessive determiners against cardinal numbers. As a reminder, cardinal numbers are used for counting things (one, two, three, etc.). This can be contrasted with ordinal numbers for example, which order things (first, second, third, etc.). In French, determiners such as les and possessive determiners such as mes go before cardinal numbers, as shown in (6) and (7) respectively.

Conversely, adjectives come after cardinal numbers. As shown in (8), the adjectives jaunes ‘yellow’ and jeunes ‘young’ come after the cardinal number trois ‘three’. This is true whether the adjective comes before or after the noun that it modifies.

The second test that can be used is the deletion test. In a sentence, removing an attributive adjective such as petit ‘little’ does not make a sentence ungrammatical, as shown in (9).

However, sentences do become ungrammatical if we remove a determiner (10) or a possessive determiner (11).

Finally, we can use the substitution test. Typically, you can replace words of one syntactic category with a word of the same category but not of another category. In French, if we try to substitute an adjective like bon ‘good’ for another adjective like grand ‘big’, the result is grammatical (12a-b). Conversely, if we try to replace an adjective with a determiner (12c) or a possessive determiner (12b), the result is ungrammatical.

Additionally, if we have a sentence with a determiner in it, it can be freely substituted by another determiner (13a-b) or a possessive determiner (13c) but not an adjective (13d).

Note that the examples directly above also show that possessive determiners can be freely substituted by determiners but not adjectives.

Based on the three tests above—the relative word order compared to cardinal numbers, the deletion test, and the substitution test—it can be seen that determiners and possessive determiners behave in one way whereas adjectives behave in a different way. This suggests that possessive determiners are in fact determiners and not adjectives. This section has shown only three tests that provide such evidence, but there are in fact more.

Hopefully this section has shown you that pedagogical grammars often have analyses that do not fit well into modern morphosyntactic analysis and that even provide more difficult explanations in some cases. In terms of possessive determiners, if they were adjectives, there is no natural explanation for why they always go before nouns, come after cardinal numbers, cannot co-occur with other determiners, can be substituted by determiners and not adjectives, and cannot be deleted when “modifying” a noun. Under the analysis that possessive determiners are determiners, learners know that they can simply put the possessive determiners in place of where they would put a determiner. The only reason why these problems seem to seldom come up among anglophone learners is that these constructions are quite similar to what we have in English.

Check yourself!

Consult the following data in answering the check yourself questions: