Sergio,, 

Thank you for a very thoughtful discussion of relative "that" and the cognitive and semiotic framework it functions within.  I admit to not having the background in semiotics that you obviously and manipulate with such insight and clarity.

I have argued previously on this list for the superiority of prototype analysis over discrete categories, and I have acknowledged the possibility that "that" is one of those words that doesn't fit any single category well, that it may have properties that characterize more than one category.  As I noted, this discussion has a long history on the list, and Craig has been very helpful in clarifying--or complicating--my own thinking on the matter.  This appears to be a case where strong syntactic arguments can be made for non-pronominal status but those arguments fly in the face of the perception of native speakers that relative "that" is pronominal.  I was trained back in the '60s when we took native speaker intuition very seriously, perhaps to excess, and my training makes me take seriously the opinion of thoughtful and well-trained native speakers like Craig.  This is also a case where my reliance on native speaker intuition and my use of syntactic analysis result in contradiction.
I'm no longer surprised at such contradiction.  Other linguists, including myself, have dealt with such contradictions in multiple languages as well as English.  Let me give a few examples from Yoruba, a Niger Congo language I've done a bit of work on.  

First there is the categorial question of verbs and prepositions in Yoruba.  Yoruba verbs, with certain systematic exceptions consist phonologically of an initial consonant, a vowel, and a tone.  Given that Standard Yoruba has 18 consonants and 10 vowels (7 oral, three nasal), and 3 tones, the set of verbs in the lexicon is limited to 18*10*3=540.  There are certain systematic gaps:  nasal consonants cannot  be followed by oral vowels.  There are also arbitrary gaps as there are in any language--English, e.g., has no word "blick"--so that the actual total number of verbs in the Yoruba is about 334.  So how does the language manage with such a paucity of verbs? English, in contrast, has upwards of 3500, not including all those words that can be functionally shifted to verb.   Syntactically Yoruba makes extensive use of serial verb constructions.  Grammatical relations beyond subject and object are marked by verbs.  "fun" (give) marks indirect objects; "ba" (help) marks  benefactives; "fi" (take) marks instrumentals; "lO" (go) and "wa" (come) mark locatives.  There are more of these for other grammatical relations and other distinctions, for example "take" (<pick up> sthg "go") can use "mu" (pick up light object), "gbe" (pick up heavy object), or "ko" (pick up multiple objects, gather).  (Pardon me for not marking tone or the distinction between tongue root retracted and tongue root advanced mid vowels, but I find HTML doesn't always handle IPA well.  I'm using /O/ for [-ATR] [O] and /o/ for [+ATR) [o], likewise /E/ and /e/.)  Here's a typical serial verb sentence:

Ade ba me gbe iSu lO si Oja fun baba mi.
Ade help me pick up yams go to market give father my.
Ade on my behalf took yams to market for my father.

Five verbs, not all of which correspond to verbs in English.

There are also two words, "ni" (be stationary) and "si" (motion toward), that have very general, semantically bleached, grammaticalized meanings.  They are frequently called prepositions, but they behave in all ways like verbs, morphologically, phonologically, and syntactically, and they are related diachronically to the verbs "ni" (have) and "si" (exist, exclusively in negative clauses).  Now the question of category vs. prototype is whether these have changed sufficiently to be considered prepositions, implying a new category, or whether they are verbs of a restricted sort.  This question continues to be debated by Yorubaists in a debate that really is a question of whether categories exist or prototypes.  Another question this whole issue raises is whether all monosyllabic lexical words in Yoruba are verbs and what to do with certain problem cases.  But enough of that here.

Second, Yoruba extends its class of verbs by means of many verb+nominal contractions.  Most nouns in Yoruba have a vowel prefix, although there is also a class of consonant initial nouns that are a special morphological type.  When a verb and a noun come together, two vowels are in contact and generally one of them must elide.  If the vowel of the verb elides, the resulting word is semantically analytic, transparently the sum of its parts.  If the prefix vowel of the noun elides, the result is a compound, with unpredictable consequences for meaning.  Thus "gbe EsE" "lift foot" can contract to "gbEsE" "lift foot" or to "gbesE" "kick."  Where things get interesting analytically is when such a word is transitive and has a direct object pronoun.  Yoruba has three sets of pronouns, subject, object, and possessive.  To say "pick it up" Yoruba says "gbe e" where the 3s object pronoun is a lengthening of the vowel of the verb with a tone polarization.  To say "pick up his/her foot" Yoruba says  "gbEsE rE" which means that some one raised his/her own foot.  However "gbesE rE," the compound, means "kick him/her/it."  The possessive pronoun maintains its morphological relationship to the noun EsE, but its meaning is "direct object of "gbesE."  Meaning and structure do not coincide.  This is a frequent structure in the language as there are many such verb+noun compounds.

Third, Yoruba has a morphological category of "ideophone."  These are words that involve onomatopoeia, reduplication, and phonological contrasts like vowel length that are not otherwise found in the language.  Ideophones can be any grammatical category, noun, verb, or adjective/adverb, a category distinction that the language distinguishes functionally but not formally.  It is impossible to tell from the form of an ideophone what grammatical category it functions as.

That's probably the longest excursus on a language other than English this list has seen in some time, but I wanted to demonstrate from a formal analytic perspective that discrete categories are problematical beyond English, if there was ever any doubt about that.  It's a case where Craig and I reach pretty much the same linguistic stance from very different directions..  I acknowledge this. of course, with great curmudgeonly reluctance.

Still, there are one or two points that I'm not clear on in your presentation.  Throughout your presentation, you never argue that relative "that" is  pronoun.  You assume it.  I don't find that assumption satisfying.  Second, your entire analysis deals with the written language, not with spoken English, and the subordinator "that" differs consistently from pronominal "that" in that it is almost never stressed, except perhaps when we want to contrast it with another subordinator, as in "I mentioned THAT I went, not WHY I went."  It's true that we can have unstressed pronouns when they are topical, as in "Have you seen John?  Yes, I saw'm at the grocery store."  Subject pronouns are typically unstressed since subjects tend to be topical.  WH-pronouns are generally at least weakly stressed, not unstressed, and relative "that" is odd in that it is regularly unstressed  whatever the grammatical function of the relativized NP.

On another matter, the deletion of complementizer "that" is a complex and subtle matter that I'll have to review again.  I fear I'm pretty rusty on the literature on that, although back in '75-'76 when I was working on my Language paper "Which that," I was pretty well up on it.  I'm looking forward to rereading Bolinger's "That's that," a typically thoughtful and subtle piece of analysis.

If you're not familiar with Johan van Auwera's "Relative that--a Centennial Dispute" Journal of Linguistics 21 (1985) pp. 149-179, I recommend it as a very thorough critical review of the non-pronoun analysis and defense of the pronoun analysis.  Unfortunately, the paper predates Huddleston and Pullum, and I don't know how van Auwera would handle their additional arguments.

Herbert F. W. Stahlke, Ph.D.
Emeritus Professor of English
Ball State University
Muncie, IN  47306
[log in to unmask]
________________________________________
From: Assembly for the Teaching of English Grammar [[log in to unmask]] on behalf of sergio [[log in to unmask]]
Sent: Wednesday, May 22, 2013 7:10 AM
To: [log in to unmask]
Subject: Re: relative "that" revisited

Herb,
Now, I see your point better. I apologize for the length of this
comment in advance.
And again, it is primarily to clarify the matter for myself, rather than else.
I have broken down my argument and the final goal is the statement in
the first section.

FROM ARISTOTELIAN CATEGORIES TO COGNITIVELY PLAUSIBLE ONES
As to the ongoing discussion about the status of "that" in relative
clauses, I think that most of the difficulties rest on the
Aristotelian categorizing system, which is often called the necessary
and sufficient conditions. Not only does relying on such a
categorization system engender many logical aporias, but it has also
been proven rather unrealistic since psychologist Eleanor Rosch's
studies in cognitive processes involved in category/concept building
in the 1970s.
After her studies a more cognitively plausible system of
categorization, named semantics of prototypes, has been recognized to
explain more efficiently many "oddities" in the way we put objects
under one category or the next.
Basically, all this is to state what Craig has posted on May 9, "I
think it's more a matter of where we draw the boundaries with our
definitions...I also grow impatient with approaches to grammar that
imply there are strict rules about how language can act."

THE COMPOSITE NATURE OF RELATIVE CLAUSE MARKERS
In a prototypical way, the issue of the status of "that" could be
restated as follows:
the objects that can be used to introduce a relative clause have a
more or less apparent pronominal function. Using “marker” as a neutral
term that encompasses “subordinator,” “complementizer,” or “pronoun,”
we can build a continuum as follows:
1) Zero marker = NO pronominal feature, Japanese does not have any
marker at all to introduce relative clauses.
2) Omissible marker = close to null pronominal feature, like "that" in English.
3) Marker sensitive to some semantic trait of the antecedent = light
pronominal feature, like the opposition “who-that” in English.
4) Marker sensitive to the syntactic function in the relative clause =
heavy pronominal feature, like the today almost superseded opposition
“who-whom,” the opposition “qui-que” in French ("qui" for subject,
"que" for non-subject function) or German's relative markers that are
case sensitive.
5) Marker sensitive to the presence of a preceding preposition =
Heavier pronominal feature, like in the opposition “that-which” in
English ("The knife with which she caught the cake" does not admit
"with that")
6) Marker sensitive to gender and number of antecedent = Full
pronominal feature, like il quale (masc.sing.), i quali (masc. pl.),
la quale (fem. sing.) le quali (fem. pl.) in Italian, my
mother-tongue, but also in French and German at least.

Within the semantics of prototypes, we would state that within the
same category some items are more or less central to the category, or
are good or bad representative of that category. In English, if we
only take the pronominal feature, we can say that “that” is not a
central representative of the relative marker, because of the presence
of markers of type 3, 4, and 5, which have a stronger pronoun-hood.

THE A-TAD-BIT-LESS COMPOSITE NATURE OF SUBORDINATORS
Along the same lines, though, if we tried to build a continuum for the
category “subordinator,” we would state that “that” as a relative
clause marker is definitely not prototypical of the category because
it has some pronoun-hood features though extremely light. But
interestingly enough, also the other “that,” the one traditionally
dubbed subordinator (as in “I think that you should go”) isn't a good
representative of the category either. The reason rests on the fact
that there is no other more “typical” subordinator that can be
dropped. Thus far I have tried to think of any example in which we can
drop subordinators such as “while,” “because,” “although,” “even
though,” “since” and didn't come up with anything. If you have any
example, it is more than welcome. Only in elliptical sentences,
subordinators can be dropped as in the following exchange:
A- Why didn't you go to the party?
B- [I didn't go to the party because] My car didn't start.

In this sense, one line of argument of Huddleston's seems to be rather
weak, because “that” as a relative clause marker becomes a
subordinator on the basis of a feature that is not so salient in
defining the category itself. It would be like saying that a rat is an
elephant because they are both gray, where “gray-ness” is not exactly
the most central and representative feature that characterizes either
an elephant or a rat. Nonetheless, in the semantics of prototypes we
could still maintain that there's some family resemblance that could
justify ascribing the relative “that” to the category subordinator. As
a matter of fact, also the relative “that” introduces a clause
subordinated to the main clause.

MY SUBSTITUTION TEST
My substitution test with “which” was not much aimed at pointing out
the similarity between the relative “that” and wh-markers. It was
rather aimed at pointing out the difference between relative “that”
and subordinator “that.”
The book that {which} is on the desk is very interesting.
They said that {*which} it is interesting.

When we have “that” as a relative marker the substitution is
structurally and semantically possible even though, according I do
agree that there are some differences. But the point is that when we
have “that” as a subordinator, the substitution is totally impossible.

AREN'T THERE TWO DIFFERENT SEMIOTIC REASONS FOR DROPPING THE TWO “THATs”?
The dropping itself of “that” as a relative marker (when in
restrictive clauses and not in subject position) and “that” as a
complementizer seems to be justified by two different semiotic
reasons.

- Dropping “that” as a relative marker
The possibility is ingrained in language because there are languages
that can do perfectly well without any relative clause marker (at
least Japanese). Moreover, in English this dropping is also possible
when the marker replaces the subject of the relative clause provided
that it is in the passive voice and the auxiliary “to be” is also
dropped (some linguists call them whiz-clauses):
The rooms [that were] reserved for the trip are very expensive.

- Dropping “that” as a complementizer
This omissible complementizer introduces the direct object of the main
clause and often the verb can be followed by direct speech:
They maintained
They said
They believed                     (that) it was the most useful action
They doubted
They thought


Neither direct objects nor direct speech are introduced by any marker.
So it is possible that, cognitively speaking, the absence of “that” as
a subordinator does not create any difficulty/ambiguity in our
perception of the sentence.
Notice that in the last sentence (“It is possible that the absence of
“that” as a subordinator does not create....”) the extraposition of
the subject creates still a different behavior of “that,” which, I
think but am not sure, cannot be omitted:

- That the absence of “that”as a subordinator does not create…. is possible
- It is possible that the absence of “that” as a subordinator does not
create …..
- *?It is possible the absence of “that” as a subordinator does not create

Ciao,
Sergio



2013/5/8 Stahlke, Herbert <[log in to unmask]>:
> Sergio,
>
> You're in the midst of a long, intermittent discussion of the status of "that" in relative clauses.  Grammarians from Otto Jespersen to Rodney Huddleston have argued that relative "that" is, in fact, a subordinator and not a pronoun.  The substitution you suggest is misleading, because that-relatives behave differently from wh-relatives, in ways I can't go into just now.
>
> Herb
>
> Herbert F. W. Stahlke, Ph.D.
> Emeritus Professor of English
> Ball State University
> Muncie, IN  47306
> [log in to unmask]
> ________________________________________
> From: Assembly for the Teaching of English Grammar [[log in to unmask]] on behalf of sergio [[log in to unmask]]
> Sent: Wednesday, May 08, 2013 1:42 AM
> To: [log in to unmask]
> Subject: Re: relative "that" revisited
>
> Dear Dr. Stankle,
>
> I might be missing the point and for the sake of my better
> understanding, I was wondering whether a simple substitution test is
> possible here.
>
> "He avoids whatever roads might cross this desolate valley and stays
> on the open land, so there's no risk of turning a bend and ramming
> head-on into innocent motorists, with all the physical and moral
> consequences that(replace it with "which") would ensue."
>
> Therefore in "...with all the physical and moral consequences WHICH
> would ensue", the original "that" is a relative pronoun referring to
> "all the physical and moral consequences" and subject of "[THEY=the
> consequences] would ensue". It is not a subordinating conjunction as
> in,
> "I think that they would ensue"
> because here "which" cannot substitute "that".
>
> Does this make any sense?
>
> Sergio Pizziconi
>
> 2013/5/8 Stahlke, Herbert <[log in to unmask]>:
>> I came upon an interesting "garden path" sentence today in Dean Koontz's One
>> Door away from Heaven (Bantam 2001), p. 287.
>>
>> "He avoids whatever roads might cross this desolate valley and stays on the
>> open land, so there's no risk of turning a bend and ramming head-on into
>> innocent motorists, with all the physical and moral consequences that would
>> ensue."
>>
>> When I got to the last three words, I anticipated that "that" would be a
>> pronoun referring to "turning a bend and ramming head-on into innocent
>> motorists," and I expected a verb like "entail."  However, the verb "ensue"
>> stopped me cold and forced me to reread and interpret "that" as a
>> subordinating conjunction.  We've discussed that status of "that" in
>> relative clauses at some length, and I've taken the position that it's not a
>> pronoun but rather a subordinating conjunction with no referential function.
>> In this case, one could write, "that that would entail," but Koontz is a
>> better writer than that.  The choice, however, is between a demonstrative
>> pronoun and a subordinator.  The fact that they can be used together
>> supports the claim that they are two different words with very different
>> functions.  Very likely the preference for only the demonstrative in this
>> case, rather than both, is an example of haplology.
>>
>> Herb
>>
>> Herbert F. W. Stahlke, Ph.D.
>> Emeritus Professor of English
>> Ball State University
>> Muncie, IN  47306
>> [log in to unmask]
>> To join or leave this LISTSERV list, please visit the list's web interface
>> at: http://listserv.muohio.edu/archives/ateg.html and select "Join or leave
>> the list"
>>
>> Visit ATEG's web site at http://ateg.org/
>
> To join or leave this LISTSERV list, please visit the list's web interface at:
>      http://listserv.muohio.edu/archives/ateg.html
> and select "Join or leave the list"
>
> Visit ATEG's web site at http://ateg.org/
>
>
> To join or leave this LISTSERV list, please visit the list's web interface at:
>      http://listserv.muohio.edu/archives/ateg.html
> and select "Join or leave the list"
>
> Visit ATEG's web site at http://ateg.org/

To join or leave this LISTSERV list, please visit the list's web interface at:
     http://listserv.muohio.edu/archives/ateg.html
and select "Join or leave the list"

Visit ATEG's web site at http://ateg.org/


To join or leave this LISTSERV list, please visit the list's web interface at:
     http://listserv.muohio.edu/archives/ateg.html
and select "Join or leave the list"

Visit ATEG's web site at http://ateg.org/