The notion of structured unit itself consists of two subparts, viz. being structured and being a unit. The claim that the referent of the noun phrase must be “structured” can be illustrated by the contrast between the examples in (170) involving proper nouns. A proper noun like Europa can easily be preceded by the predeterminer heel, because the geographical entity “Europe” is normally understood as constituting a structured set of basically equivalent objects, i.e. member states. A proper noun like Jan, on the other hand, cannot be combined with heel because a person is not normally understood as a structured, homogeneous set of objects such as cells, organs, or limbs.

The same is normally true for a noun phrase headed by a [+animate] common noun like de manthe man in (171a); however, this example also shows that once a context is provided which allows the animate noun phrase de man to refer to the set of a person’s limbs, a slightly marked but acceptable result arises. Examples in (171b) show that a similar but somewhat stronger contrast can be found with [-animate] noun phrases like zijn armhis arm. Example (171c) shows that the result is perfectly acceptable without contextualization with the [-animate] noun phrase zijn lichaam, which can be seen as a structured unit consisting of a set of body parts.

The examples in (171) alternate with fully acceptable sentences with the adverb helemaalaltogether; cf. De man/Zijn arm/Zijn lichaam zat helemaal onder de schrammenThe man/his arm/his body was completely covered with scratches’. Example (172) shows that helemaal can also be used with proper nouns, which leads to a severely degraded result with heel; the syntax of helemaal will be discussed in more detail in Section 21.2.4.

The fact that some noun phrases readily allow an interpretation as a structured unit with heel, while other noun phrases require a special context for this interpretation to become available, suggests that it is the speaker’s conceptualization of the material world that is responsible for the difference: lichaambody is stored in the mental lexicon as a noun denoting structured units consisting of body parts, whereas manman is stored as a noun denoting atomic units (viz. individuals). The same distinction applies to proper nouns: Europa is stored in the mental lexicon as the name for a structured unit consisting of member states, while Jan is stored as the name of an individual.

So far we have focused on the requirement that the referent of the noun phrase must be structured, i.e. must be understood as consisting of several subparts. That the referent of the noun phrase must be a unit is emphasized by the interpretation of example (173a); the indirect object is conceived as a unit, all parts of which are equally and collectively affected by the event expressed by the verb phrase. The unit reading expressed by (173a) can be primed by the paraphrase in (173a'). In (173b), we find a similar example that involves a PP-complement; again it is possible to provide a paraphrase with in z’n geheel.

The primeless examples in (173) contrast with the examples in (174), despite the fact that the syntactic function of the heel phrases in these examples is the same, viz. indirect object and complement of a PP-complement of the verb.

The difference in acceptability between (173) and (174) thus seems to be of a semantico-pragmatic nature. Insofar as the examples in (174) are acceptable, the referent of de film receives an intrinsically distributive interpretation, with attention being equally distributed among its subparts (e.g. the various scenes of the movie). It is apparently difficult for heel de film to receive such a distributive interpretation, which seems to lend support to the importance of the unit part of the semantic characterization of predeterminer bare heel; whereas the house in (173) is conceived of as a unit whose parts are collectively affected by the event expressed by the verb phrase, the events in (174a&b) do not affect the subparts of the movie as a group or unit, but only in a distributive way.

The structured-unit requirement is also reflected in the fact, illustrated in (175a), that plural noun phrases cannot normally be combined with predeterminer bare heel; in the general case, plurals do not constitute a unit, but a set of units. Systematic exceptions to the ban on predeterminer heel construed with plural noun phrases are pluralia tantum, like de tropen in (175b), and formal plurals denoting a conventionally fixed unit, like de Antillen in (175c).

Although the plurals in (175b&c) behave like regular plurals in their external syntactic distribution in triggering plural agreement on the finite verb, their denotation is that of a unit. The fact that predeterminer bare heel can quantify pluralia tantum and plurals denoting a fixed unit confirms the structured-unit ingredient of its semantics.

The partitioning part of the semantics of heel can be illustrated with reference to example (176), which seems to imply that all of the rooms (and other relevant subparts) of the house have been cleaned, i.e. all of the constituent parts that together make up the house have been affected.

Since heel partitions the entity denoted by the noun phrase it quantifies into its constituent parts, we expect an anomalous output if these parts cannot be affected individually by the event denoted by the verb phrase. This seems to be confirmed by (177): (177a) is unacceptable because the motorboat can only rock as a unit, i.e. its parts cannot be affected individually; (177b) is awkward because mud typically covers the house as a unit, not all of its constituent parts (i.e. the individual rooms) separately.

Some caution is needed here, however, since not all speakers agree that (177b) is indeed anomalous. This may have to do with the fact that the event of a house being buried under the mud in (177b) evokes a partitioning of the exterior of the house in question, whereas an event of a house being cleaned in (176) usually evokes a partitioning of the interior of the house; we leave this issue for future discussion.

The third concept involved in the meaning of predeterminer bare heel is exhaustivity; cf. Postma (1996). This is exemplified in (178a), where heel prompts a reading in which all of the individual office spaces that make up the office block have been rented out. It seems that the addition of an “except”-clause, which overrides the interpretation “in all of its constituent parts” assigned by heel, leads to a somewhat awkward result (indicated by the dollar sign) The examples in (178b&c) illustrate the same point; the addition of an “except”-clause leads to a somewhat awkward result. Our acceptability judgments on the examples in (178) seem to be supported by a Google search (August 1, 2022); the strings [heel de * behalve] and [heel de * met uitzondering van] returned fewer than 40 hits in total, including several instantiations not relevant to the present discussion.

The exhaustivity part of the meaning of predeterminer bare heel is eminently present in the Dutch rendering of the introduction to Asterix the Gaul by Goscinny and Uderzo, given in (179). The italicized part is of interest for our present discussion: the assertion that the whole of Gallia (heel Gallië) is occupied is refuted by the reference to a small settlement that continues to offer resistance.

Consistent with the characterization of the semantics of predeterminer bare heel in terms of exhaustive partitioning of structured units, we find that bare heel does not normally alternate with halfhalf. This is especially the case when the noun phrase hosting heel/half is “totally affected” by the event denoted by the verb phrase, as in (180): bezaaid liggen met means “to be totally covered with” and leegroven means “to rob empty”, i.e. to rob in such a way that the whole object is emptied as a result. Such “totally affecting” verbal predicates only allow partitioning of their surface subject if the partitioning is exhaustive: substituting half for heel yields an anomalous result, since half differs from heel precisely in not being exhaustive.

The verb phrases in (181), on the other hand, are not “totally affecting” in the sense that they apply only to the inhabitants of the island/village, and now the modifier heel alternates with half. The reason for this is that these verbal predicates allow, but do not require, an exhaustive partitioning of the surface subject.

The percentage sign in (181) is used to indicate that some speakers find half in predeterminer position categorically impossible, but examples such as (181) are easily found on the internet; is actually an attested example (as an alternant of the more common form het halve dorp liep uit, to be discussed later). It should also be noted that we have also seen several cases with half on the internet that seem to be of a different nature than the examples in (181), so further investigation of the alternation between heel and half would certainly be welcome.