Replication types

[selpa'i]

There seem to be two types of default place filling, though neither the
RefGram nor the dictionary are at all explicit about them. This is
chiefly my personal assumption, which I shall present in short here:

Hardcoded replication is what we find in infinitives and in restrictive
phrases. It causes the automatically filled place to "disappear" from
the place structure and can only be overriden by use of an explicit case
link "so":

(1a) \ji /daw \cyr \dowu
"I want to go home."

Which is the same as:

(1b) \ji /daw \vo cyr \dowu

Normally, "\dowu" would have to fill the *first* case of "cyr", but it
doesn't, because the "daw" causes its first case to be hardfilled into
the first case of the infinitive in its second case. It disappears from
"cyr"'s place structure. This way we don'T have to repeat arguments in
infinitives, and we don't have to skip cases with case links.

Hardfilling also occurs in restrictive phrases:

(2a) \ji /fan \kqua |zgy
"I like water that is old."
"I like cold water."

This is the same as:

(2b) \ji /fan \kqua |vu zgy

The restricted phrase (kqua) gets replicated into the first case of the
restricting phrase (zgy). This is hardfilling because again, the only
way to explicitly fill the first case is by using the case link "so".

(3) \ji /fan \xa pso |[vu]fan \ji
"I like every person that likes me."

In the subordinate phrase, "ji" fills the second case, not the first,
because "xa pso" got hardfilled in the first case.

This should have been straighforward so far, but there *seems* to exist
a sort of "softfilling" as well. Here are some examples:

(4) ^:i fel
"Hooray." (from Refgram "Organization - Subordinate Clauses")

In (4), the first case is filled by "ji" due to the following rule (from
the same section):

"'ji' is provided by default in the first case (before conversion) of
any subordinate clause or top-level sentence whose first case ends up
vacant."

What I take this to mean is that this mechanism gets applied only after
the sentence is complete, unlike the hardfilling mechanism which occurs
immediately upon use of "infinitive-vo" or a subordinator. This in turn
means that in (4), adding an argument to the sentence would not put it
into the second case, but into the first.

(5) ^:i fel \ju
"You are happy" (not "I am happy about being you.")

This is what I mean by softfilling.

Another case of softfilling can be found in a few predicate words, for
which the dictionary indicates a default value for some of their cases.
"pli", for instance, is such a word:

(ji) X1 feels pleased that X2: does (vo) X3:+2

Here, "ji" softfills the first case of "pli".

Looking through the dictionary a bit, I noticed that there is a quasi
pattern to the words that do employ softfilling and the ones that don't.
A good rule of thumb seems to be that predicate words that lend
themselves well to vi-subordinates or va-subordinates are very likely to
have softfilling. Here are some examples: jai, qnu, faw, kam, glu...

Was this presentation convincing? What do think about these replication
types (if you happen to accept my analysis)?

I hope this is useful to some, as it might remove some of the confusion
that can arise when reading through the examples in the Refgram and
looking through the dictionary.

|jai \qo pranay