Nouns such as belief, dogma, fact, order, philosophy, sign, etc. governing finite complement clauses are discursive devices for the speaker to characterize a proposition, an event or a state of affairs (cf. Biber et al. 1999 or Schmid 2000). Such nouns are part of a closed semantic class that undergoes a strong selectional constraint. As illustrated in example (1), the challenge for linguists remains to explain why philosophy, doctrine or dogma can govern a that complement clause whereas religion or linguistics cannot. (1)This rejects the philosophy [doctrine/dogma/*linguistics/*religion] that the market can deliver health or community care efficiently, effectively or fairly. (BYU-BNC)In the literature, linguists have often related head noun selectional constraints to abstractness (Quirk et al. 1985, Francis et al. 1998), derivation (Nomura 1993, Huddleston and Pullum 2002) or modality (Perkins 1983, Ballier 2007, and anonymous reference). Taking into account these studies, this paper aims at arguing that head noun selectional constraints cannot be explained through a single particular feature. For instance, one can notice that neither all the head nouns are abstract (message, sign, picture), nor all abstract nouns are head nouns (inflation, literature, peace): the result of the vote is at least a hopeful sign [*peace] that sexist dinosaurs are on their last legs in the 20th century, (BYU-BNC). Similarly, two abstract nouns can have a co-hyponymic relation but differ in that-clause governing: though philosophy and linguistics can be co-hyponyms of FIELD OF STUDY/DISCIPLINE (example 2), they do not have such a relation in that-clause governing as in example (1). (2)My Lords, Mr. Page was appointed as a lecturer in philosophy [linguistics] in the University of Hull with effect from 1 October 1966. (BYU-BNC) As for derivation, it can neither be a sufficient property to explain head noun selection in all cases (to write that P vs. *the/a writer/writing that P, to learn that P vs. the/a learner that P). As a result, instead of describing head noun selectional constraints on the basis of isolated features, we propose to deal with the phenomenon as depending on a set of common semantic features that enable nouns to take that complement clauses. In this vein, we work out five properties that nouns must have in order to govern that-clauses. These criteria consist of three lexico-semantic features: abstractness [+ABSTRACT], inanimate [−ANIMATE], nonhuman [−HUMAN] and two semantic-pragmatic features: modality [+MODALITY] and reference to discourse [+REF. TO DISCOURSE].Such an approach pinpoints why nouns like believer, messenger or philosopher – contrary to belief, message and philosophy – cannot govern that-clauses. In comparing examples (3 and 4), one can see that the former are not head nouns since they do not meet any of the five features listed above. (3)We had a very strong belief [message/philosophy] that independence could be achieved through interdependence and interaction. (BYU-COCA)(4)*We had a very strong believer/messenger/philosopher that independence could be achieved through interdependence and interaction.It also enables us to explain why attitudinal/modal nouns like coward, liar, scum, swine, etc. do not govern that-clauses. At least two of the criteria above (abstractness and reference to discourse) are not satisfied: these nouns refer to specific extra-linguistic entities (“first order entities” in Lyons’ 1977 terms). Furthermore, the paper aims at showing that among the five criteria, [+MODALITY] and [+REF. TO DISCOURSE] seem to be the most predominant features, since, in addition to being part of the lexico-semantic defining set, they also determine the discursive function of the whole that-clause construction. Finally, the analysis reveals that modality and reference to discourse are two interdependent features. Intrinsically, any lexical or grammatical modal marker expresses the speaker’s attitudes or opinions in relation to a discursive or propositional content, which is to say that modalization and referring to discourse are interrelated semantic operations. ReferencesBallier, Nicolas 2007. “La complétive du nom dans le discours des linguistes”. In D. Banks (éd.), La coordination et la subordination dans le texte de spécialité. Paris : l’Harmattan, 55-76.Biber, Douglas et al. 1999. Longman Grammar of Spoken and Written English. Essex: Longman.Francis, Gill et al. 1998. Collins COBUILD grammar patterns 2: Nouns and adjectives. London: Harper Collins.Huddleston, Rodney and Pullum, Geoffrey 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press.Lyons, John 1977. Semantics. Cambridge/New-York: Cambridge University Press.Nomura, M. 1993. “The semantics of the content clause construction in English”. English Linguistics, 10, 184-210.Nomura, Masuhiro 1993. “The semantics of the content clause construction in English”. English Linguistics, 10, 184-210.Perkins, Michael 1983. Modal Expressions in English. London: Frances Pinter. Quirk, Randolph et al. 1985. A Comprehensive Grammar of the English Language. London/New York: Longman.Schmid, Hans-Jörg 2000. English Abstract Nouns as Conceptual Shells: From Corpus to Cognition. Berlin: Mouton de Gruyter.Salkie, Raphael 2008. “Modals and typology: English and German in contrast”. In M. L. A. Gómez-González, J. L. Mackenzie and E. M. González Álvarez (eds.), Current Trends in Contrastive Linguistics: Functional and Cognitive Perspectives. Amsterdam/Philadelphia: John Benjamins, 77-98.[Anonymous reference], 2010. “Mood and modality in finite noun complement clauses: A French-English contrastive study.” In M. Stefania; K. Heylen and G. De Sutter (eds.), Corpus Studies in Contrastive Linguistics. IJCL 15: 2. Amsterdam: John Benjamins, 267-290.CorporaBYU-BNC: Brigham Young University – British National Corpus.BYU-COCA: Brigham Young University – Contemporary American English Corpus.