2.1. Meaning Contribution

We stated above that ons ‘our’ signals family membership. This is a slight oversimplification. There are indeed dialects that use ons ‘our’ only for family memberships, including for spouses and in-laws. This is, for example, the case for the Brabantic dialects spoken in the area west of Brussels. Other dialects might be more or less restrictive in this respect. The Northern Brabantic dialect spoken in the Dutch town of Vught is an example of a less restrictive dialect. Kroon (Reference Kroon2015) shows that the younger speakers of this dialect do not only use ons ‘our’ with proper names of family members, but also with those of close friends.

The Brabantic varieties spoken in Brussels and the neighboring areas to the North and East are examples of more restrictive dialects. In these dialects, ons ‘our’ can only introduce blood relatives of the speaker. Relatives that are part of the family of the speaker through marriage (for example, spouses and in-laws) cannot be introduced by ons ‘our’. In order to appreciate this, consider the family tree in 5.

1. (5)

   

As illustrated in 6, Speaker Katrin can, for instance, use ons ‘our’ in combination with the names of her daughters, Emma and Anna, and of her father, all of whom are blood relatives. The examples in 6, however, also show that Katrin cannot use ons ‘our’ with all the members of her family. For instance, she cannot use ons ‘our’ in combination with the names of her husband, Guido, or her sister-in-law, Sammy. This shows that Katrin cannot use ons ‘our’ to refer to those family members who are related to her by marriage.

1. (6)

   

Unlike her mother, speaker Emma is a blood relative of everyone who is represented in the family tree in 5. She can use ons ‘our’ in combination with the name of anyone in 5, as is illustrated by the examples in 7.

1. (7)

   

Based on these data, we conclude that ons ‘our’ in these dialects can function as a marker that signals relations between blood relatives. We therefore refer to ons ‘our’ as a marker of consanguineous possession. In the remainder of this paper, we focus on dialects with consanguineous ons ‘our’. However, the account that we develop is flexible enough to also be applied to the less restrictive dialects, as we show in section 2.3 below.

2.2. Other Properties

In addition to the semantics outlined above, consanguineous ons ‘our’ has a number of other unusual properties. First of all, consanguineous ons cannot receive contrastive stress, as shown in 8.Footnote ⁵

1. (8)

   

This incompatibility between consanguineous ons ‘our’ and contrastive stress is not the result of a more general incompatibility between 1st person possessive pronouns and contrastive stress. In the relevant dialects, ons ‘our’ is not only used as a marker of consanguinity or family membership, but it can also be used as a regular 1st person possessive marker. On this use, it can carry contrastive stress, as in 9.

1. (9)

   

Proper names are not only compatible with the consanguineous use of ons ‘our’, but also with its normal possessive use. It is possible to combine a proper name with ons ‘our’ carrying contrastive stress. In that case, ons ‘our’ can only be interpreted as the regular possessive, not as the consanguinity marker. We illustrate this with the example in 10. In this example, Emma is not necessarily a consanguineous kin member, but instead she can be a member of any group that can be marked as ours. So Emma in 10 could belong to our class, our softball team, our friends, our colleagues, etc.Footnote ⁶

1. (10)

   

In the introduction, we already discussed another property of the family/consanguinity marker ons ‘our’: its incompatibility with productive adjectival modification, as shown in 11 (see also 2).

1. (11)

   

The unacceptability of the adjective in 11 has its source in the presence of consanguineous ons ‘our’. It is not due to an incompatibility between normal productive adjectival modification and proper names. This is shown by the observation that the same adjectives can modify the same names in 12, without giving rise to unacceptability.

1. (12)

   

We propose an analysis of consanguineous ons that straightforwardly accounts for these properties.

2.3. Analysis: Consanguineous Ons as a Constructional Idiom

As pointed out above, ons ‘our’ also functions as the regular 1st person plural possessive pronoun. In case the regular 1st person plural possessive ons ‘our’ combines syntactically with a common noun such as huis ‘house’, the resulting phrase ons huis ‘our house’ means something like ‘the house that is in a possessive relation with a group that includes the speaker’. The possessive pronoun thus introduces a possessive relation. The semantic nature of this possessive relation is rather flexible. Ons huis ‘our house’ might be the house that we designed, built, or painted, the house that we own, live in, etc. Depending on the context, different interpretations of the possessive relation are possible. In order to account for this flexibility, Barker (Reference Barker1995) proposes that the possessive relation denotes an underspecified relation R whose exact nature is filled in by the context of the utterance it appears in. Using this relation R, the interpretation of the regular possessive phrase ons huis ‘our house’ formally corresponds to 13, in which the iota operator encodes definiteness (that is, an exhaustively identified specific individual).

1. (13)

   

By contrast, when ons ‘our’ is used as a consanguineous marker with a proper name, it is not the case that the individual denoted by the proper name is in some possessive relation that needs to be filled in by the context. Rather, this relation strictly represents the consanguineous relation between the individual denoted by the proper name and the plural entity denoted by ons ‘our’. So, ons Emma in 1a above, repeated here in 14a, refers to the unique individual called Emma, who is in a consanguinity relation with a plural entity that includes the speaker, as shown in 14b.

1. (14)

   

This means that the difference between the regular possessive ons ‘our’ and consanguineous ons ‘our’ is that the possessive relation is underspecified in the former, but specified as consanguinity in the latter. This additional meaning of consanguinity is noncompositional. Ons ‘our’ does not by itself entail consanguinity, as shown by its regular possessive use. The same holds, of course, for proper names: Uttering the proper name Emma by itself does not entail the existence of a consanguineous relation. Put differently, the consanguinity of the phrase in 14 is a semantic property of the whole construction that is not traceable to any of its parts.

Semantic noncompositionality is the hallmark of idiomatic expressions. In this case, however, one is not dealing with what is traditionally called an idiom, that is, expressions such as English kick the bucket, of which all parts are fixed. In the case of consanguineous ons ‘our’, the proper name that occurs after ons ‘our’ is not fixed. Instead, it can be any first name. Booij (Reference Booij2002:302) refers to such a syntactic expression with noncompositional meaning of which only a part is fixed as a constructional idiom, following work by Langacker (Reference Langacker1987), Jackendoff (Reference Jackendoff, Everaert, van der Linden, Schenk and Schreuder1995, Reference Jackendoff1997, Reference Jackendoff, Nooteboom, Weerman and Wijnen2001, Reference Jackendoff2002) and work done in construction grammar (Fillmore et al. Reference Fillmore, Kay and Catherine O’Connor1988, Goldberg Reference Goldberg1995, among others).

We propose that the combination of consanguineous ons ‘our’ and a proper name is a constructional idiom in this sense. Following Booij Reference Booij2002 and work cited therein, we consider constructional idioms to be stored in the lexicon as a fixed expression, that is, a lexical template that has a variable part and that is associated with its own (often noncompositional) meaning. The variable part of the expression in the relevant case of consanguineous ons ‘our’ is the first name following ons. Consanguinity is part of the meaning associated with the constructional idiom as a whole. Since consanguinity is only represented at the level of the entire constructional idiom, it need not be traceable to any of the idiom's parts. The syntactic and the semantic properties of the constructional idiom that is at the basis of instances of consanguineous ons ‘our’ are schematized in 15.

1. (15)

   

‘the unique individual who bears the proper name Y and who is in a consanguinity relation with a plural entity that includes the speaker’

The representation of the constructional idiom in 15 accounts for those dialects in which ons ‘our’ in the context of a proper name signals a consanguinity relation. In section 2.1 above, we noted that there is dialectal variation with respect to what this use of ons ‘our’ marks. There are dialects, like the varieties spoken West of Brussels, in which it signals all family relations, including nonconsanguineous ones. There are also dialects in which it marks membership of a peer group, as in Vught Dutch. These dialects have constructional idioms that are minimally different from the one in 15. The sole aspect in which they are different is that the consanguineous kin relation that is part of the meaning of the constructional idiom is replaced either by a general family relation or a peer group membership relation. Since the constructional idioms in these dialects are identical with respect to all other properties, below we continue to exclusively focus on the consanguineous ons dialects.

2.4. Accounting for the Properties of Consanguineous Ons

Above we discussed two peculiar properties of consanguineous ons ‘our’ constructions. We now show how our analysis accounts for them. One of these properties is that names introduced by consanguineous ons ‘our’ resist productive adjectival modification, as shown in 16, repeated from 11a.

1. (16)

   

The unacceptability of the utterance in 16 is straightforwardly accounted for by our proposal. As explained by Booij (Reference Booij2002), parts of constructional idioms resist modification because they do not syntactically project in the usual manner. The reason for this is that the parts of a constructional idiom do not combine in the syntactic component. Instead, the constructional idiom corresponds to a lexical template that is stored in the lexicon. The variable part of that template is specified by the lexical component of the grammar prior to insertion of the constructional idiom in syntax. The constructional idiom is thus formed in the lexical component and will therefore behave as a single atomic syntactic unit. Modification of a noun by an adjective is a syntactic process. Given that the entire constructional idiom is an atomic syntactic unit, modification of its parts is impossible. So in order to derive 14a—that is, 16 without the adjective—the first name Emma replaces, in the lexical component of the grammar, the variable Y shown in the lexical template in 15. This results in the syntactic unit [_DP ons Emma]; although it consists of two words, it is syntactically atomic in that it does not have an internal syntactic structure. Given the absence of this structure, it is impossible to perform syntactic modification of its composing parts. This explains the unacceptability of adjective slimme ‘smart’ in 16.Footnote ⁷

This renders combinations of consanguineous ons ‘our’ and a proper name akin to other complex elements that are stored in the lexical component instead of being derived in the syntactic component. Noun-noun compounds are a case at hand. Crucially, such other complex lexical items also do not allow their nominal subparts to be productively modified by an adjective.Footnote ⁸ For instance, the adjectives in 17 can only modify the entire compound, but not the first nominal part of the compound.

1. (17)

   

This parallel with adjectival modification between compounds and consanguineous ons ‘our’ confirms our lexical treatment of the latter.

The other peculiar property of consanguineous ons ‘our’ we introduced above is its incompatibility with contrastive stress, even though ons ‘our’ is compatible with contrastive stress in other uses. This incompatibility has been shown in 8 and is repeated here in 18.

1. (18)

   

This property also follows straightforwardly from our proposal that consanguineous ons ‘our’ is part of a constructional idiom. In order for an element to receive contrastive stress it needs to have an independent meaning. Otherwise it is impossible to contrast its meaning with the meaning of something else. On our proposal, the consanguineous ons ‘our’ is part of the lexical template in 15 above. This template is stored in the lexicon with its associated meaning. Crucially, this meaning is a property of the whole template rather than of its parts. This means that the ons-part of the constructional idiom does not carry any meaning of its own at any level of representation.Footnote ⁹ As a result, it is impossible to put contrastive stress on ons ‘our’: It cannot be contrasted since it has no meaning of its own.

We have shown above that our constructional idiom approach accounts for three properties of the combination of consanguineous ons ‘our’ and proper names: its noncompositional meaning, its incompatibility with productive adjectival modification, and the impossibility to contrastively stress ons ‘our’ in its consanguineous use. We now turn to another construction in the same dialects that combines ons with a kinship term. We show that this construction needs to be analyzed along the same lines as combinations of consanguineous ons ‘our’ and proper names.

4.1. The Head Restriction and Its Exception

Standard Dutch has multiple strategies to express adnominal possession. One of these strategies is to use a prenominal possessor that is marked with the suffix -s, as in 25.

1. (25)

   

This strategy bears some resemblance to the English Saxon Genitive ‘s. In both cases, there is a prenominal possessor that is marked by a suffix -s. There is, however, also a crucial and well-known difference between the two. The possessor occurring in the Saxon Genitive can be phrasal, as shown in 26a. Most speakers of Dutch, however, do not allow phrasal possessors to be marked with -s, as shown in 26b.

1. (26)

   

Instead, the possessor taking the -s suffix must be a syntactic head (see Weerman & de Wit Reference Weerman and de Wit1999, Corver Reference Corver, Koster and van Riemsdijk2003, Kampen & Corver Reference Kampen, Corver and Vliegen2006, Broekhuis & den Dikken Reference Broekhuis and den Dikken2012, among many others).Footnote ¹² As a consequence, Dutch prenominal -s possessors are restricted to proper names, as in 25, kinship terms that can be used without an article, like vader ‘dad’ in 27a, and quantifiers such as iemand ‘somebody’ in 27b.

1. (27)

   

Prenominal -s possessors are not the only way to express adnominal possession in Dutch, but it is the only construction that displays a head restriction. Dutch has two other adnominal possessive constructions. The first of these also involves a prenominal possessor. The possessor is, however, not marked by the -s suffix. Instead, it is doubled by a possessive pronoun, as in 28a. This strategy is also known as possessor doubling (see Grohmann & Haegeman Reference Grohmann and Haegeman2003, Georgi & Salzmann Reference Georgi and Salzmann2011, Schoorlemmer Reference Schoorlemmer2012 for a discussion of similar cases in Norwegian and German). The second of these additional strategies is to express the possessor postnominally through means of a van- PP, as in 28b.

1. (28)

   

Unlike prenominal -s possessors, the possessor in possessor doubling constructions and van-PPs does not need to be a head, but can also be an XP, as shown in 29.

1. (29)

   

Broekhuis & den Dikken (Reference Broekhuis and den Dikken2012:837–838) note that there seems to be a curious exception to the head restriction on prenominal -s possessors. Kinship terms introduced by a 1st person singular possessive pronoun readily occur as -s possessors in Standard Dutch, as in 30.Footnote ¹³

1. (30)

   

Example 30 seems to involve a phrase as prenominal -s possessor. Although it was not discussed by Broekhuis & den Dikken (Reference Broekhuis and den Dikken2012), the obvious question that arises is whether exceptions to the head restriction are limited to kinship terms accompanied by a 1st person singular possessive pronoun or whether other combinations can also constitute a complex -s possessor.

The data in 31, 32 and below show that only the combination 1st person singular possessive pronoun + kinship term can be a complex -s possessor. It is not possible to combine a nonkinship term, such as kapper ‘hairdresser’ with a 1st person singular possessive pronoun, as shown in (31).

1. (31)

   

It is also impossible to combine kinship terms with possessive pronouns other than 1st person singular. This is shown in 32a. A 2nd person singular possessive pronoun is only marginally acceptable. A 3rd person singular possessive pronoun leads to an even worse degradation. Finally, plural possessive pronouns lead to sharp unacceptability. Examples 32b,c show that there are no restrictions of a similar sort on the possessor doubling and prepositional strategies to express adnominal possession.

1. (32)

   

The data in 31 and 32 confirm that only the combination 1st person singular possessive pronoun + kinship term violates the head restriction.

4.2. Analysis of the Apparent Exception to the Head Restriction

The question that now arises is why the combination mijn ‘my’ + kinship term is the only exception to the head restriction on prenominal -s possessors. Our answer to this question is that these are only apparent exceptions. In reality, they are syntactic atoms, that is, my and a kinship term together form a syntactic head. As a result, the head restriction is not violated. These combinations of mijn ‘my’ and a kinship term are syntactic atoms because they are instances of a constructional idiom, as we argued above.

Recall from our discussion that there are two observations showing that one is dealing with constructional idioms in these cases. First, it is impossible to add material such as adjectives to a complex -s possessor, as shown in 4 and 23 above and illustrated here with another example in 33. In addition, it is also impossible to put contrastive stress on the possessive pronoun of a complex possessor, as shown in 24 above and 33c below.

1. (33)

   

Since these two observations demonstrate that one is dealing with a constructional idiom, we propose that the combination mijn + kinship term is an instance of the constructional idiom, as shown in 34.

1. (34)

   

In the case of 33a, the variable part Y of the constructional idiom is filled in by the kinship term moeder ‘mother’ prior to insertion into the syntax. Consequently, the two-word combination mijn moeder will enter the syntax as an atomic unit, that is, a head. It can therefore freely combine with the possessive -s suffix without violating the head restriction. In this way, there are no real violations to the syntactic head restriction on prenominal possessors in Dutch. The only case that looks like one is a multiword expression that is formed in the lexicon, not in syntax. In this way, the apparent exception to the head restriction, and the bans on contrastive stress and adjectival modification receive a unified account.

Note that the meaning of the constructional idiom in 34 is fully compositional. It is the same as if mijn ‘my’ and the kinship term were combined in the syntax. It is therefore, just like the ons ‘our’ + kinship term combinations discussed above, another illustration that noncompositional meaning is not a necessary property of a constructional idiom.