This is the second chapter which deals with complementizers (like English that). In this feature, we investigate a different verb class from Chapter 95, namely verbs of knowing, such as 'know', 'forget', or 'learn' in sentences like We knew that they had come. Verbs of knowing belong to the group of factive verbs where the content of the knowledge – expressed in the complement clause – is entailed. This fact puts factive verbs apart from non-factive verbs like ‘think’, ‘believe’, or ‘trust’, where the complement clause can be questioned or denied. (For ‘think’ complements, see Chapter 98.)
Complementizers are defined here as elements that link the embedded clause to the verb of knowing, not belonging either to the verb of knowing or to the embedded clause. Examples are fa in Fa d'Ambô (see ex. 1), ité in Mixed Ma’a/Mbugu (see ex. 2), and kay in Zamboanga Chabacano (see ex. 3):
Indirect questions like 'I don't know where...' or 'I don't know whether...' are disregarded in this feature.
As in Chapter 95, quite a few languages use a form of the verb 'say' as a complementizer with verbs of knowing. As these serial verb constructions have been intensely discussed in the creole literature, we are interested in their distribution as well.
In this feature we distinguish four values:
|Complementizer identical to bare ‘say’||7||18||25|
|Complementizer consists of ‘say’ plus some other marker||0||3||3|
|Complementizer not synchronically related to ‘say’||11||31||42|
In 25 APiCS languages, we find a complementizer which is identical to a ‘say’ verb (value 1). Seven of these languages have this type as their only option (out of which San Andres Creole English is one):
In the Surinamese creoles, the complementizer is taki/táa ‘say’ (< English talk):
One could classify the predicate marker a- in Sango as illustrating value 2 (‘say’ plus some additional marker, see next value). But as the predicate marker expresses person information, we interpret it as being part of the verb complex and therefore put it under value 1.
The next two values are by far the most frequent values in this feature. 42 languages show a complementizer which is not synchronically related to ‘say’.
Complementizers can be complex, as for instance in Vincentian Creole da hou (where both parts, da and hou, can be used on their own as complementizers).
In Korlai, one finds a circumpositional complementizer composed of the general complementizer ki and the element puris:
Fifty-four APiCS languages allow for no complementizer after verbs of knowing (value 4), 16 of which have this as their only option:
Interestingly, all pidgins in APiCS show exclusively this value.
An interesting question is whether languages which have the bare 'say' construction in Chapter 95 (with verbs of speaking) also have a bare 'say' construction in this feature. And indeed, it is striking that the values for the APiCS languages in Chapters 95 and 96 are nearly identically distributed, not only regarding the bare ‘say’ constructions. But there are some languages which show differences between complements of saying and knowing, e.g. Angolar, Ambon Malay, and Vincentian Creole—where the bare 'say' construction is not possible with verbs of knowing whereas it is possible with verbs of saying. In Papiá Kristang and in Hawai‘i Creole, verbs of saying show both strategies (with the complementizer ki/ dæt and no complementizer) whereas verbs of knowing only show no complementizers. In Diu Indo-Portuguese and Batavia Creole the option of a zero complementizer is only found with verbs of saying, and not with verbs of knowing.
APiCS languages show far fewer possible construction types with verbs of knowing than with verbs of saying (1.63 vs. 1.82 average value choices per language).
For a substratal explanation of the bare ‘say’ constructions, see § 3 in Chapter 95.