In the equative constructions in (118), the neutral form langlong is compared to the non-neutral form smalnarrow; this yields a degraded result.

The examples in (119), in which two non-neutral forms of the measure adjectives are compared, are also unacceptable. This leads to the conclusion that the comparison of two adjectives is only possible with the neutral forms of the measure adjectives, i.e. the form of the measure adjectives that can be modified by a nominal measure phrase; cf. twee meter lang/*korttwo meters long/*short and één meter breed/*smalone meter wide/*narrow.

In comparative constructions we see essentially the same thing. First, (120a) shows that the neutral forms of measure adjectives can be compared without difficulty. Second, since comparatives of the non-neutral forms of measure adjectives can also be modified by a nominal measure phrase (cf. Section 25.1.4, sub II, example (256)), it is not surprising that example (120b) is also acceptable. Finally, the primed examples show that the result is unacceptable when the adjective in the dan-phrase is a non-neutral measure adjective. In short, it seems that being susceptible to degree modification is a necessary condition for being able to occur in the construction under discussion.

Example (121) provides the semantic representations of the acceptable primeless examples in (120); again, the predicates are in bold to indicate that we are dealing with the neutral meaning of the adjectives. Example (120a) does not imply that the table in question is actually long or wide; neither does (120b) imply that the referent of deze tafel “table1” is short nor that the referent of die tafel “table2” is wide.

The examples in (122) show that as soon as the construction includes an adjective other than a measure adjective (i.e. an adjective that cannot be modified by a nominal measure phrase), the construction yields an unacceptable result.

We conclude this subsection by noting that Kennedy (1997) gives English (123a) as acceptable, but examples such as (123b) as (semantically) anomalous. This contradicts our hypothesis that the two adjectives should both be modifiable by a nominal measure phrase: since this is the case for shorter (10 cm shorter) but not for low (*2 meters low), (123a) should be ungrammatical; since both shorter and high can be modified by a nominal measure phrase (2 meters high), (123b) should be grammatical. Our English informants do not (fully) share Kennedy’s judgments: some are simply confused by these examples, while others consider (123b) merely marked and sometimes even better than (123a), especially when shorter is modified by a measure phrase like two meters. Example (123c), which is not discussed in Kennedy (1997), is considered by all of our informants to be the best way to express the intended proposition. This would be consistent with our hypothesis, since both less tall and high can be modified by a nominal measure phrase.

Our judgments on the corresponding Dutch examples in (124) are similar to those of the English informants who prefer (123b&c) to (123a), and thus consistent with our hypothesis. Example (124a) is unintelligible to us and clearly worse than (124b), and the best way to express the intended proposition is to use the minorative form of the adjective langlong, as in (124c). Our German informants offer similar judgments about the German translations.