Graph b
1.2 Fundamental processes of sentence generation
Sentence generation is performed by enumeration of the morphemes
according to the directed edges. The processing can be executed not only
in order of the directed edges, but also by the morpheme alternation
under a certain rule in order to proceed the enumeration with morphemes
that have no edges directed from the last morpheme (q.v. 1.3).
The sentence generation requires an ability to select signifiés which
construct the meaning of the sentence. This ability and the process,
which we refer to as the signifiés selection, deeply concerns with one’s
cognition. Videlicet, the selection of the signifiés that describe a
phenomenon depends on the speaker. Differences in the use of languages
are created in the phase the speaker selects the signifiés, before the
selection of the morphemes. Diction is considered to derive from the
differences in cognition. A photograph of a room can be living
room to someone while the others describe it as home .
In the following phase, the morphemes which construct the sentence are
selected from the morphemes which correspond to the signifiés. This is
the morpheme selection. Unnatural sentence structures, such as
unsuitable conjugation, are obviated as the directed edges for those
structures are not formed.
A morpheme which follows another morpheme is hereinafter referred to as
a subsequent morpheme. The morpheme in the beginning of a sentence is
equally considered to be a subsequent morpheme of another morpheme; the
morpheme is referred to as the α-morpheme (alpha-morpheme). The
α-morpheme has no influences in the meaning of the sentence and has no
signifiants. It has a function to lead the morpheme in the beginning of
the sentence as its subsequent morpheme. Some morphemes in a sentence
behave as subsequent morphemes of a morpheme before those (not
the previous morpheme) e.g. the concord in Early Middle Japanese, present
perfect in German. Those subsequent morphemes are referred to as the
quasi-subsequent morphemes. The sentence generation stops when there are
no morphemes needed.
The following are examples of the processes of sentence generation of a
sentence: ʻI saw a yellow house.ʼ in English, Japanese and ʻI've seen the yellow house.ʼ in German. In
each graph, the morphemes that are selected through the signifiés
selection and the morpheme selection to be required to generate the
sentences are highlighted in yellow.
Graph c describes the example sentence generation in English. First, the
suitable signifiés are selected. Subsequently, the morphemes that
correspond to the signifiés are selected. The morpheme ʻhouseʼ is a
subsequent morpheme of the adjective ʻyellowʼ, and the quasi-subsequent
morpheme of the article ʻa.ʼ In the enumeration phase, all the morphemes
are ordered starting with ʻIʼ led by the α-morpheme in order of the
directed edges of the graph.
Graph d describes the sentence generation in Japanese:
(私は) 黄色い 家を 見た.
I.NOM yellow house.ACC saw.
ʻI saw a yellow house.ʼ
Processes of the sentence generation are identical in any languages; however different morphemes and signifiés are used in different
languages. For instance, a subject of a verb in a sentence can be
determined by the context in Japanese. There are other morphemes whose
corresponding signifiés are determined by the context such as
demonstrative adjectives and demonstrative pronouns. Therefore, the
signifié graph is considered to be consisted of the long-term signifié
graph as the knowledge of a language and the short-term signifié graph
which is similar to the dynamic memory of computers. In this example,
the subject ʻIʼ is included in the verb with reference to the short-term
signifié graph.
Graph e shows the processes in the sentence generation in German. The
conjugated verb ʻgesehenʼ is a quasi-subsequent morpheme of ʻhabe.ʼ