Welcome to Adobe GoLive 6

I view language as a means of communicating information to one or more speakers, and a means of interpreting sentences that have been uttered. To this extent linking theory is a form of information theory.

I am postulating that language has three major components and at least two submajor components that are required to link the three major components. A lexicon including grammatical information is also required. The three major components are lexical or conceptual semantics, form, and sign. Form includes syntax, morphology, and phonology. Sign is either phonetics or signing by the deaf. I will not offer any insights in the sign component.

This framework is adopting certain theories of transformational grammar, tree adjoining grammar, set theory, and some properties of other grammars. Linking is not new—I have borrowed this notion from tranasformatinal grammar, but I am proposing that links are the medium through which information is transmitted.

A link connects two nodes of any type. Suppose there are two nodes: A and B. It is possible to create a link between A and B, such that some kind of information can be transmitted from one node to the other. Indeed, I am proposing that information can go either way depending on the situation. The link between A and B will take the form of a double headed arrow:

If we assume that the left side of the arrow points toward lexical semantics and the right side towards phonetic sign, then A transmits information to B, and in the reverse B seeks interpretive information from A.

Suppose we add more nodes. Let us say 3 more for a total of 5, for example. If each node were connected, we would have a highly undesirable situation. Information could b e transferred all over the pace and information inerpreted. But the result would be chaotic. We need to a system of linking constraints. To do so, we will add two lesser components: argument structure which links lexical semantics to syntax and the morphophonemic component which links form (basically morphology but some syntax) to the phonemic component.

3. Linking the morphophonemic component and the phonologica component.

Here, I will start with linking from the morphophonemic component to the phonological component. I will use Russian as an example. Let me start with the simple word kniga /knig+a/ book. The stem is /knig/ and the suffix /a/ is an inflectional ending marking the nominative case singular for A-stem nouns. The stem is listed in the lexicon as 'knig', marked for the A-stem declension, gender feminine. The stem is linked to the 12 inflected forms marking case and number. These forms originate in the grammaticon, which is either a subcomponent of the lexicon, or it is an indendent component. We prefer the former. This form is regular; the stem is stressed throughout. The default links are established between the underspecified entry and the spcified entry. We will provide a link initially to the nominative singular and the nominative plural, knigi /knig+i/. Each phoneme is linked to the phonemic set established in the grammar:

I propose the following phonemes without argument here. These phonemes form a set enclosed in curly brackets:

Each lexical item is linked to the appropriate phoneme. I consider linking to be of very small cost the grammar. The information of each phoneme is transferred to the appropriate place in the lexical entry, which is then passed down to the appropriate fully specified word form.

The phonemic stem for kniga is {knig}. In a technical sense, there are four positions, each one linked to the phonemes /k/, /n/, /i/, and /g/, respectively:

In the first position in {knig} there is place with a connection to /k/ in the phoneme list. The information is stored in phonemic entry, and it can be activated when a connection is made. Four phonemic entries become activated in the production phase.

The inflectional ending /´/ is determined in the syntactic componet and in the morphological component. Let me suppose that we somehow know that the form is the nominative singular. The default is {á, ´}. Which ending is chosen depends on the stress class of the stem. If a given syllable is stressed, then the stressed /á/ cannot be selected since Russian permits only one stressed syllable in a word. /´/ must be selected. The stress falls on the root 'KNIG' throughout the declension. Another rule of Russian stress is that every word must have one stressed syllable. Thus, /´/ is selected:

The phonemic component is linked to the phonetic component, which produces the requisite sounds.

I have now added a morphophonemic form: {á, ´}. Rather than derive it from some morphophonemic, i.e. /á/ --> /´/. I believe that the two forms form a set, a morphophonemic set. When the link feeding this morphophonemic set is activated, a choice has to be made where to activate /á/ or activate /´/. This decision is based on stress. There are two rules or filters that rule out the incorrect phoneme.

The first bars a word from having more on than one main stress; the other bars unstressed words:

Let it be understood that "'..V" represents the onset of a stressed syllable plus the onset of the syllable plus the peak vowel of that syllable. Three dots "..." represents an indefinite string with the proviso that that if boundary markers are included as in (1b), they cannot occur in the indefinite string.aq And let it be understood that if if there is no stress mark in a word, the syllable is stressed. This rule must include clitics adjoined to verbs. Monosyllablic clitics cannot be stressed on their own, but the stress may shift onto a clitic leaving the wordform without a stressed syllable. Clitics are marked by a single '#':

(2a) passes since there is one stressed syllable, even though it is on the clitics. (2b) and (2c) pass since each string has one stressed syllable. (2d) fails because the string contains no stressed syllables. And (2e) fails since more than one syllable is stressed.

I now return to the selection of the nominative case-ending for kníga. If /á/ is selected, then Filter (1a) is violated. Stress is inherently assigned to the stem, and stress is selected for the ending. Selecting /´/ does not violate either filter. Whenever there is a morphophonemic alternation within a given morpheme, that alternation must be posited as a morphophonemic set. Such sets belong to the lesser morphophonemic component.

In the following figure, I trace the connections activated between the morphophonemic component and the filter:

Once the link to the 2nd Declension in the grammaticon is activated and the link to {á, e} is activated, a choice has to be made. Either /á/ or /´/ must be selected. For the moment, suppose /á/ is selected, marked in red. This information /á/ is sent to the Nominative Singular Suffix. The noun stem and the noun suffix are both members of the set noun and are linked. Now let me adopt tiers. Stress has its own tier, and I will identify this feature specifically, since stress will determine the choice of the morphophoneme. All words must pass through the filter. nIn this case each syllable is linked to any filter/rule that contains a syllable. If a syllable is selected, then everything that it contaions (dominatres) is affected. If /á/ is selected, then both the stem syllable and the desinence syllable are linked to the filter. The first syllable in the word, /kníg/ is connected to the ffirst syllable in the stress filter given above. The next syllable is then connected to the next position in the filter. The filter states that if two syllables within a word (between ##...##), the word fails, which will render the sentence containing the word ungrammatical.

The linking shown above does not really show how this works. At some point it may become too detailed to show everything, but I will attempt to show this here.