Tinkertoy Panlingua representation of "Galileo miffed the cardinals."

Three layers can be seen in this photograph.  The top layer consists of
the written "words" on cardboard tags, the second layer consists of the
words themselves, and the  third layer consists of the meanings of the
words written on cardboard tags at the bottom.  These layers can be
thought of as the phonological, the syntactic, and the semantic planes
respectively.  The nodes of the phonological layer are not actually words,
but SYMBOLS.  These symbols are the external representations of words that
are seen and heard in the real world, and not the actual words themselves.
 For example, in some sentence or other, the word for "water" may appear.
 In English, the symbol for this word is either the spoken or the written
form of "water."  But in Malay the symbol for the same word is "ayer,"
while in Greek it may be "hydor."  These external symbols are NOT part of
the Panlingua representation, but I have left them attached because these
symbols are what trigger the formation of actual words within our minds.
 The vertical rods supporting these symbols are NOT linguistic links, but
only a means of attaching each symbol to its underlying word.

All the rest of this Tinkertoy representation models a pattern of
linguistic links and the nodes at which they connect to one another.
 These links are patterns which can be represented in sketches on paper,
in Tinkertoy structures, or in the neurons of the human brain, but whereas
the symbols used to represent words in language are audible or visible,
linguistic links are nonmaterial PATTERNS, which are completely MASSLESS
and INVISIBLE in and of themselves.

All of the internal data structures of language can bemodeled using
nothing but linguistic links, which are connected to one another
end-to-end at connecting points called NODES.  Each linguistic link,
hereafter just called "link," consists of three elements: a SOURCE, a
TYPE, and a DESTINATION.  They can thus be thought of as tiny vectors or
arrows of various types.  Panlingua theory recognizes four generic link
types: syntactic, semantic, radical, and lexical, as follows:

1. Syntactic links (synlinks for short) link words (dependents) to their
regents (other words within the same phrase).

2. Semantic links (semlinks for short), which link words to their
meanings.

3. Radical links (radlinks for short), which link meanings to each other.

4. Lexical links (lexlinks for short), which link meanings to words in the
lexicon.

The nodes, or connecting points, of the syntactic layer are words.  The
nodes of the semantic plane are meanings, also called semantic nodes, or
semnods.  The nodes of thephonological layer (the lexicon) are called
lexical nodes, or lexnods.  Lexnods have a one-to-one relationship with
the spoken or written symbols usually called "words," but it should be
remembered at all times that these "words," as they are popularly known,
are not really words but only symbols, real words being the nodes of the
syntactic plain.  This terminological confusion might have been avoided by
calling words "syntactic nodes," but to do so might cause confusion by
calling real words something other than just "word."

So the second layer of this Tinkertoy model consists of representations of
the links and nodes of the syntactic plain.  A node is actually an
invisible connecting point, but in this Tinkertoy model we have used
Tinkertoy wheels to model them in the syntactic plane.  Thus each
Tinkertoy wheel represents an actual word.  And each word is nothing more
than a single node from which emanate (come out) two links.  The first is
a syntactic link, usually to another word, and the second is a semantic
link to the meaning of the word.  The syntactic links exist only within
the syntactic plane, whereas the semantic links bridge the gap between the
syntactic and semantic planes.  Unfortunately for our model, tinkertoy
rods have no DIRECTION, and do not look much like arrows, so we will have
to give them direction in our minds.  The semlinks from words to their
meanings are ALWAYS directed downwards from the syntactic plane to the
semantic plane.  The synlinks always go from each word to its REGENT,
except for that emanating from the top word of the sentence, which has a
link source and a link type but no link destination.  In other words, the
synlink of the top word goes NOWHERE.

So now let us examine our Tinkertoy model more closely.

From "Galileo" emanates a synlink to "miffed," which is its regent, and a
semlink to "a man," which is its meaning.  The type of the synlink is
"subject," and the type of the semlink is "proper noun."

From "miffed emanates a synlink that goes nowhere because "miffed" is the
top word of the sentence and has no regent, and a semlink going down to
"to anger."  The type of the synlink is "declarative," and the type of the
semlink is "past-tense verb."  Note that the synlink type of the top word
of a sentence is the type of the whole sentence.

From "the" emanates a synlink to "cardinals," and a semlink to "known
item."  The type of the synlink is "determiner," and the type of the
semlink is "article."

From "cardinals" emanates a synlink to "miffed" and a semlink to "Catholic
church official."  The type of the synlink is "object," and the type of
the semlink is "plural noun."