Category Archives: Subjectless Sentences

The Mystery Of The Missing IS: Or, Had John Duns Scotus Been An Ordinary-Language Philosopher Working In Tagalog

Below, I have tried to start incubating the suspicion that there is something fishy about treating ‘is’ as a predicate with two parameters accepting one argument each, i.e., a two-place relation.

Tagalog doesn’t have a verb ‘is’, no verb ‘to be’.  Given that more literal translations of Tagalog sentences often display the phrase ang noun phrase structure as:


phrase [is] ang noun phrase

For example:

Titser ang babae.

Maganda ang lalaki.

Umalis ang babae.

gets rendered as:

Teacher [is] the woman.

Beautiful [is] the man.

Having left [is] the woman.

or as I prefer (see my attempt below at eliciting the ‘aha erlebniss’):

Some teacher one  [is] the woman.

Some beautiful one [is] the man.

Some having left one [is] the woman.

…given that, one might think that, always, the suspect verb aka predicate aka relation is implicitly in effect in sentences with that structure.  The lack of a verb ‘to be’, of an ‘is’ in Tagalog that so perplexed the first Spanish grammarians of the language (so that, in their total confusion and lack of understanding, they tried to interpret the Tagalog inversion marker ‘ay‘ as the verb ‘is’, a confusion and misinterpretation that has had hilarious consequences lasting to this day), is always there, just unpronounced (or unwritten).  The space between ‘maganda‘ and ‘ang lalaki‘ in the written sentence, or the lack of interruption in the string of sounds (if that is how maganda ang lalaki gets pronounced — I am not strong enough presently in Tagalog to know) or the glottal interruption (if one exists between the ‘maganda‘ and ‘ang lalaki‘)  … the space, or lack of interruption in the continuous stream of sound, or the glottal, these are, as the case may be, an implicit sign of the two-place relation ‘is’.

Following Naylor, Schachter, and my own intuition, I have been treating the space, the lack of interruption in the continuous stream of sound, the glottal as an implicit equals.  For example, I prefer to translate the above three Tagalog sentences as:

Some teacher one  = the woman.

Some beautiful one = the man.

Some having left one = the woman.

Unlike ‘is’, however, which is (if there is such a critter) a two-place relation, ‘equals’ (alternatively, ‘=’ ) is, as I am about to show, a one place relation.  It is not just that the sign corresponding to ‘is’ is lacking in Tagalog:  the (real or putative) semantics of ‘is’ is lacking in Tagalog as well.  Tagalog is working with something completely different.

Clearly the ‘equals’ that is in play here is not given by the ‘equals’ in the following two-place relation:



Morning Star Evening Star
3 3
Rose With Barcode 3185321 Rose With Barcode 3185321
Clifford Wirt Clifford Wirt
The murderer of Jones The butler

…because in sentences such as Maganda si Taylor Lautner, the word ‘Maganda’  does not, at the moment of its utterance, specify, identify, locate, expose, or pick out any one particular thing.   ‘Maganda’ is equivalent to ‘Some beautiful one’, or the part of the formal sentence below that occurs before the ‘=’:

∃x ∈ MAGANDA: x = si Taylor Lautner.

The x that belongs to the set MAGANDA is left unspecified, unidentified, unlocated, unexposed, un-picked-out at the start:  Maganda … though it does get specified at the end:  …si Taylor Lautner.  But a two-place relation requires two identified, specified arguments for its two attributes.

Let me try to capture in D the sentence ‘∃x ∈ MAGANDA: x = si Taylor Lautner’.  Let me posit the following 1-place relation:

Taylor Lautner
Sunset at time t and place p
Rose With Barcode 3185321
Wine Red
The Taj Mahal
Haendel’s Umbra Mai Fu

Taking this relation as my springboard, I capture ∃x ∈ MAGANDA as MAGANDA{} (which gives us TABLE_DEE, or TRUE, or YES), then do a CARTESIAN PRODUCT of that with a restriction of MAGANDA:

MAGANDA{} as t_sub_0,
MAGANDA{MAGANDANG_BAGAY} where   MAGANDANG_BAGAY= ‘Taylor    Lautner’ as t_sub_1:
t_sub_0 X t_sub_1

CARTESIAN PRODUCT is a special case of JOIN.  TABLE_DEE JOIN r, where r is any relation, yields r.  So the D statement above yields:

Taylor Lautner

which expresses the semantics of the sentence ‘Maganda si Taylor Lautner’.  In this way, we get rid of the doubtful (I think) verb aka two-place relation ‘is’.

To sum up, a bit impishly:  the semantics of ‘is’ is different in Tagalog than in English because Tagalog really doesn’t have an ‘is’.  Later, I will try to develop this into part of an argument that Tagalog lacks a subject.  Tagalog’s lacking a verb ‘to be’ is related to its lacking a subject.

To stray back for a moment to philosophy:  were Duns Scotus an ordinary-language philosopher working in Tagalog, it may never have occurred to him to try to find a single relation (e.g. ‘contracts’ ) between the entity Beauty, as the argument on one side of the predicate ‘is’, and Taylor Lautner as the argument on the other side of the predicate, and so on for every other proposition formed by supplying arguments to the parameters x and y in the predicate x is y.

11/10/2012:  Updated to make a point a bit more clearly.

11/10/2012:  Updated to parenthetically add some snark about the first Spanish grammarians of the Tagalog language in the 1600’s.


Update:  11/25/2012:  Post grayed-out because I am dissatisfied with it.


Naylor’s Argument That Tagalog Lacks A Subject

The most persuasive and clearest argument I’ve found for the claim that Tagalog lacks a subject is Paz Buenaventura Naylor’s in her contribution to SUBJECT, VOICE AND ERGATIVITY (ed. Bennett, Bynon, Hewitt).  The majority so far of the posts in this blog have been attempts to wrap my head around this argument and to state the argument in my own words; I am placing this set of arguments in the category ‘Wrapping My Mind Around The Argument That Tagalog Lacks A Subject.’ In trying to gain a maximal grasp on this argument, I’ve learned a tiny bit of linguistics; but any illusion I may have produced (not likely anyway) of having any authoritative voice at all on the subject is just that, an illusion.  I am writing to learn, not to force-feed the world from my <this is meant ironically>vast store of knowledge</this is meant ironically>.

What follows is Naylor’s argument in a nutshell.  Further posts will be going into the details and articulating some disagreements.  I reserve the right to go back to her article and find out I have horribly misrepresented her position.

1)  A verb is a relation whose relata are (in the case of 2-place relations) subject and direct object and (in the case of 3-place relations) subject, direct object, and indirect object.  Conversely, a subject and object (direct and indirect) are always relata in a verb.

2)  Tagalog doesn’t have verbs.  What look like verbs are really something else:  They are names of actions rather than syntactical verbs.  We know that Tagalog doesn’t have verbs because ‘ng‘ is always a genitive.

3)  Since Tagalog doesn’t have verbs, and since subjects and objects are relata of those relations that are verbs, Tagalog does not have subjects.  (Neither does it have objects, direct or indirect.)

4)  This should be enough to show that Tagalog does not have a subject, but one can’t resist pounding in an additional nail in the coffin by pointing out the claims about syntax made by linguists such as Paul Kroeger are just wrong.

Many of my further posts in the blog will be elaborations on and criticisms of the above outline of an argument.


Maganda Without The Ang


1) Birfers!

with any of the following:

2) Si Robert Pattinson ang maganda.  (The beautiful one is Robert Pattinson.)

3) Ang maganda si Taylor Lautner.  (Taylor Lautner is the beautiful one.)

The exclamation “Birfers!” is a speech act naming a set, namely, the set of people who think that Obama is not a citizen of the United States.  (This is a substantial subset of ‘people who are seriously detached from reality’.)  Following Max Black, I submit that a set is just ‘things named all at once.’  ‘Birfers!’ thus establishes a set by naming all at once all people who are Birfers at the time of the utterance.

Robert Pattinson ang maganda” and “Ang maganda si Taylor Lautner” show a peculiarity about Tagalog:  one can turn an adjective or verb into a noun denoting a particular, discrete entity by pre-posing ‘ang‘ to it.  The sentences do not translate as “The beautiful is Robert Pattinson;’ and ‘Taylor Lautner is beautiful’;  they translate as “The beautiful one is Robert Pattinson” and “Taylor Lautner is the beautiful one”.  The ‘ang‘ in ‘ang maganda‘ signals that some entity has already been identified or will readily be identified (when it is pre-posed to to the topic) or that this definite entity and not some other is identical with the topic (when it is pre-posed to the predicate).

One sees, for example, a person in a group who stands out because of his beauty and you are confident he will stand out this way for your audience.  You know that Robert Pattinson is identical with this person, but you aren’t sure that the people you are talking with know this, so you say “The beautiful one is Robert Pattinson.”  “Si Robert Pattinson ang maganda.” You are assuming an identity.

You see someone in a group whom you know to be Taylor Lautner; you are overwhelmed by his beauty, but you are not totally sure that everyone else is (Dr. Forsberg for example is constantly casting aspersions on your taste in guys); there are other arguably beautiful men there…Brad Pitt, for example, or Matt Damon or Jude Law or Ashton Kutchner… so you say “Taylor Lautner is the beautiful one”.  “Ang maganda si Taylor Lautner”.  Not Brad Pitt, not Matt Damon, not Jude Law, not Ashton Kutchner, but Taylor Lautner — just to set the record straight.  So you are asserting, stating, not just assuming an identity.  This is the predicate, after all.

That ang either assumes or asserts an identity suggests the possibility of interpreting ‘maganda‘ as ‘some (currently unidentified) beautiful one.’  Taking out the ‘ang‘ takes out the identity, the ‘this specific one’, the ‘this one and not that other’.  What one is left with then is some (currently) not yet specified beautiful one; there exists some member of the set of beautiful entities, but we do not yet know which one, nor do we have enough of a handle on the entity to say it’s this one and not that other.

Of course, without the ‘ang‘, ‘maganda‘ cannot be the topic of the sentence. (The ‘*’ indicates a sentence that would strike competent speakers as a word salad.)

* Si Robert Pattinson maganda.

So in the canonical Tagalog sentence, the ang-less maganda could only be at the left of the sentence — on the left side of the scales, so to speak:

Maganda si Robert Pattinson.

This would then translate (on an extremely literal level) as:

Some beautiful one = Robert Pattinson.

Or again:

Some member of the set of beautiful entities = Robert Pattinson.

And this would be the easiest, cleanest way to cash out the intuition that the canonical Tagalog sentence has a PREDICATE = TOPIC structure.  Coming up with the cleanest way of doing this is the motivation for suggesting that we look at ‘maganda‘ as ‘ang maganda‘ stripped of the assumed or asserted specific identity signaled by the ‘ang‘.  (And understanding the arguments for the claim that Tagalog does not have a subject is the motivation for trying to cash out the PREDICATE = TOPIC intuition.)

If this way of cashing out the intuition is correct, then, ‘maganda‘ does not name a set in 2), as I suggested in a previous post.  Instead, it names some member of the set of beautiful entities that is unidentified at the moment of the utterance of the word and will remain unidentified until we get to ‘si Robert Pattinson.’  In this way, it differs from the ejaculation ‘Birfers!’ because, unlike ‘Birfers!’, it names not a set but some (unidentified) member of a set.

Above, I say ‘suggests the possibility’ rather than ‘shows’ because of course this interpretation of ‘maganda‘ has not been demonstrated.  To show that ‘maganda‘ in 2) names at the time of utterance some (currently unidentified) beautiful entity, I would have to show that stripping away the ‘ang‘ in ‘ang maganda‘ does not radically alter the function of the word ‘maganda.’  This of course I have not shown.  The constituents of ‘ang maganda‘ might not be that atomistic, partes extra partes.

So far all I have is ‘I have the intuition that the canonical Tagalog sentence is an equality (and Naylor and Schachter also have this intuition, so  nyah nyah nyah), and interpreting ‘maganda‘ in the way I have just suggested would be the simplest way to describe this equality should the intuition turn out to be correct.



The English Laughs Names A Relation; The Tagalog Tumatawa Names A Set

Saying  ‘Kha laughs’ I do not pretend to have full control over how to refer to an utterance. I have this feeling that a whole can of spaghetti awaits me here, and disentangling each strand of spaghetti will take some doing is akin to typing the letters “K” “h” “a” onto the screen of a semantic machine and into an underlined area of the screen that precedes the string “laughs”, The hell with it — I am going to use the British orthographic rules for combining quotation marks with other punctuation; the American orthography for this is horribly misleading   then hitting ENTER.  “Laughs” names a predicate that possesses an agent parameter.  Entering a string argument for that parameter creates a proposition — something that can be true or false.  Hitting ENTER completes the utterance:  the function delineated in the previous post gets called; the entity bearing the name “Kha” gets selected; the truth value, represented by the string “T”, associated with that entity gets returned.  Kha is indeed laughing now; the proposition is true.

The semantic machine works a little bit differently in the case of the corresponding Tagalog sentence ‘Tumatawa si Kha’, at least if the machine has been constructed guided by the notion that the canonical Tagalog sentence has a predicate = topic structure in the way outlined in the first post below.  The Tagalog semantic machine has on the screen “∃x ∈ tumatawa: x = si _____”.    But should the user type in “K” “h” “a”, the machine will work the exact same way as in the case of _____ laughs and return the result.  The function { Kan → F, Chris → T, Kha → T, Guile → T, Juan → T, Corliss → T, Ralph → T, John → T, Robert → F, Ken → F } gets applied to Kha.  The truth value T, represented by the string “T”, gets returned.

Notice now the difference between laughs in _____ laughs and tumatawa in ∃x ∈ tumatawa: x = si _____.  [L]aughs is a relation because it is a function.  (All functions are relations; not all relations are functions.)  It is a 1-place function, taking as its sole parameter an agent.  (A more intuitive example of a relation would be _____ throws _____,  a 2-place parameter which takes an agent argument in the first parameter and a direct-object parameter in the second parameter.)  [T]umatawa, on the other hand, is a set, not a relation.  A verb is a relation.  So, naming a set, the string “tumatawa” is really a noun, not a verb.  That the Tagalog “verb” is really a noun is a position Naylor argues for; I hope to get into that argument later.

Trying To Understand The Subjectless Tagalog Sentence Umuulan (Rains)

Looking at the semantics for a 1-place predicate such as the English laughs (an “intransitive” verb, i.e., a verb with a transitivity of 1, having only an agent parameter into which arguments can be plugged naming the agent) suggests a way to understand at least one subjectless, 0-place predicate Tagalog verb:  umuulan.

The semantics for laughs can be described as a function from an agent to a truth value.  For the sake of simplicity, I will restrict the members of the set of entities who can laugh to just 10 — this is our pretend universe of people who are able to laugh.

[[ laughs ]] = { Kan → F, Chris → T, Kha → T, Guile → T, Juan → T, Corliss → T, Ralph → T, John → T, Robert → F, Ken → F }

A truth value T or F is assigned to each member of the set according to whether the person laughs now or not.

Imagine now a machine on whose screen one sees “_ laughs.”  The machine takes as input an argument for the agent parameter.  The argument is a string representing a particular person who can laugh.  Suppose we enter the string “Chris,” and hit ENTER.   The machine produces two outputs.  One output is the string “Chris laughs,” which is the syntactic effect of the operation.

The other output is the truth value T, which is its semantic effect.  Calling the function with the argument string “Chris” selects the entity Chris denoted by the argument and returns the associated truth value T.   (See  This truth value might be represented, say, by the string “T” appearing on the screen, or, less prosaically, by a shaft of light coming down from the heavens accompanied by the sublime choir of Caravaggio-esque angels.

How would one construct a similar machine for the Tagalog predicate umuulan, which has no agent parameter accepting an argument….in fact, has no parameters at all?    Unlike the English speaker, who has to plug in a phony agent to produce the sentence “It is raining” (there has to be SOMETHING that is raining, a close friend of mine said, apparently basing his intuitions on his first language Vietnamese and his second language English), the Tagalog speaker simply says umuulan — rains.

(Let me write umuulan when I am talking about the complex of the string “umuulan” and the semantic machine, and “umuulan,” as I have just done twice, when I am talking about just the string, stripped from its context in the semantic machine.)

Let’s construct the function this way:

[[umuulan]] = { Locality around Jamby at time t0 → T, Locality around Juan at time t1 → F, Locality around Jamby at time t1 → T … }

The assignment of the truth values T or F to each locality is determined by whether drops of water can be felt dropping in that locality.  How ‘locality’ is to be defined — how many miles or fraction of a mile around Jamby etc. is enough to count as a locality, is a sorites issue which I will ignore for the moment (though of course it cannot be ignored forever).

Jamby logs into the semantic machine using the string “Jamby.”  The machine captures the time of his logon with a timestamp, which happens to be t0.  The semantic machine calls the function using the string “Jamby” and the timestamp,  selects the locality determined by those two values, and returns, on the syntactic level, the string “umuulan,” and on the semantic level, the truth value T (represented however which way).

Construed this way, umuulan would function as a kind of indexical.  The indexical character of the predicate enables one to show how the semantic machine can select a  particular member of the set of localities and return its associated truth value without the aid of an argument plugged into an agent parameter.  Without an agent parameter — without any parameter — the predicate lacks a subject. It is a subjectless sentence.  It is a sentence nonetheless — actually, a proposition — because even in its naked agent-parameter-less subjectless state it is either true or false.

If anything could be a well-formed sentence without a subject, umuulan would be it.  Rain has a pervasive quality — in all the grayness, mist, and clouds, it is difficult to point to any specific thing that is the agent of the the raining, that is doing the raining.  The case is different in the unusual and maybe somewhat comical situation in which a single dark cloud above one’s head is raining on one, because that cloud is identifiable.  At least in the less rational parts of one’s mind, the cloud is an agent doing something, initiating an action, even as one’s rational mind discounts this.

(In any attempt to deal with the vagueness of the term ‘locality,’ I would want to work with the notion ‘point at which nothing gets identified as the agent of the raining.’ )

So this is my account of one Tagalog subjectless sentence, umuulan.  No subject is required for the hypothetical semantic machine to do its work in the case of umuulan.  But what about the other subjectless Tagalog sentences, a category which according to Paz Buenaventura Naylor, Paul Schachter, and others, includes all Tagalog sentences?  What sort of account can be given for all of these sentences?  Can a way be found to construe the workings of these sentences as not needing a subject, in which case the notion of a subject is not a linguistic universal after all, or will we be forced to re-introduce the notion of a subject in order to understand them?  Can a way be found to make the subjectless character of these sentences (if in fact they are subjectless) intuitive, as (maybe — my friend may dispute this) was done with umuulan?

The attempt to answer these questions is what is motivating these posts on Tagalog.

Predicate = Topic Structure In Tagalog Sentences

Titser ang….

Imagine a Tagalog speaker in the act of uttering a sentence, beginning with those words.  Not having yet uttered the noun phrase that is about to come after the ang, the speaker hasn’t yet come to the topic — what the utterance is about.  So far we, the speaker’s audience, have no handle on any definite person or object being talked about — this will come a bit later, after the ang noun phrase.  What we can reasonably guess, though, is that the person being talked about is identical with some member(s) of the set titser, i.e., the set of teachers.  (We can guess that, that is, if we are familiar with the notion of sets.)  To use the database theorist C.J. Date’s terminology,  we already know that the person, whoever he or she is, putatively belongs to the type titser.

Titser ang babae.

If the speaker’s utterance has not misfired — if his confidence is not misplaced that we will know whom he is talking about — then at the utterance’s conclusion we, his audience, will have identified from the context which person is being talked about.  The context can be either perceptual (say, the person has just walked into the room) or spoken (the person, say, was previously mentioned in conversation).  The identification is something happening at the pragmatic level.  (Warning:  I have recently picked up just enough linguistics to be a danger both to myself and to society at large.  I am now able to persuade myself that I can distinguish between the pragmatic, semantic, and syntactic levels of an utterance.)  Assuming the statement is true, we now know which member of the set of teachers is being talked about.  So the utterance has the effect of:  [Do I need to cash out this ‘has the effect of’?]

I embraced the summer dawn

One member of {A, B, C…} = this particular woman. (Where A, B, C … are members of the set of teachers.)

I embraced the summer dawn

I advance this analysis of Titser ang babae in an attempt to cash out the linguist Paz Buenaventura Naylor’s claim that the canonical Tagalog sentence (comprising predicate on the left + topic or ang phrase on the right) is an equality.  Consider the following English sentences, identical with or almost the same as the examples Naylor uses:


Teacher = the woman.

Beautiful = the man.

Left = the woman.


These express in English the force of:   [Can I get away with the metaphor ‘force’?]


Titser ang babae.

Maganda ang lalaki.

Umalis ang babae.


At first sight, Naylor’s claim is, I think, more intuitive in the case of Titser ang babae than it is in the other two cases. How is beautiful equal to this particular man?  What could that possibly mean?  Likewise, how is left (as in ‘left the room’ ) equal to this particular woman?  But if we cash out thse equations as:


One member of {A, B, C…} = this particular woman.  (Where A, B, C, etc. are particular teachers.)

One member of {A, B, C…} = this particular man.  (Where A, B, C… are particular beautiful objects or people.

One member of {A, B, C…} = this particular woman. (Where A, B, C…are particular people ((or animals, or anythings else with agency)) who have left some place, e.g., a  particular room, a city, a country.) [Can I get away with restricting ‘left’ in the third example to leaving a particular place, as opposed to a job, a wife, a party, and so on?]


… then the equality becomes much more intuitive in the case of the second and third examples.

If, like me, you like to think in terms of wildly undisciplined, not completely respectable pictures and metaphors, picture at the start of the utterance — Titser ang…. — a crowd comprising all the teachers in the world.  [How strict do I need to be in specifying this set?]  Our view of each teacher is fuzzed out or grayed out so that no teacher can be distinguished from another.  The utterance completes:  Titser ang babae.  The moment the utterance is understood, our view of one teacher in the crowd resolves itself.  We now clearly see one particular woman who is a teacher.  Likewise, gather in one’s imagination all the beautiful objects or people…and all the entities with agency who have just left a place…..  The same picture we used in the case of  Titser ang babae applies, mutatis mutandis, to the other two utterances.  When the utterance completes, our fuzzy picture gets resolved, and a particular beautiful man (Brad Pitt in A River Runs Through It or even in The Tree Of Life) appears, or a particular woman (Marlene Dietrich, say) who has just left this particular room.  Brad Pitt pops into view.  Marlene Dietrich pops into view.

The equation makes sense now in these two latter cases because one particular gets identified with another particular when we utter the sentences.   ‘This particular thing (veiled at the start of the utterance) is identical with that particular thing (known to speaker and audience through the context, i.e., pragmatically). ‘


∃x ∈ titser: x = ang babae.

∃x ∈ maganda: x =  si Brad Pitt.

∃x ∈ umalis: x = si Marlene Dietrich.


We make sense of the claim that the canonical sentence in Tagalog has a predicate = topic structure when we regard not just titser, but also maganda and umalis as names of sets.  One member of each set equals an entity named in the topic that has been identified pragmatically.  This will also start to make sense of Naylor’s intitially counter-intuitive claim that in Tagalog what look like verbs are actually nouns, i.e., names.

To use an even less respectable conceit (‘conceit’ used here as in the sense it is applied to the technique used by English Metaphysical Poets), it is as if we identified an unknown star in the morning with an already identified Evening Star.  The ang phrase codes old information:  we already know the Evening Star is the planet Venus.  The predicate codes new information:  we now know, by the end of the sentence, which bright object showing up in the morning is in fact identical with the Evening Star is in fact identical with the planet Venus.  It is as if something like this were happening with every Predicate = Topic Tagalog sentence….

All right.  Enough of strained metaphors … although this one at least lets me picture the function of the Tagalog predicate as coding new information and the Tagalog ang phrase as coding old information.  And lets me picture ‘start of utterance’ (morning) with ‘completion of utterance’ (evening).


When I first started suffering under the delusion that I had some grasp of how Tagalog works vs. English, I pictured the canonical Tagalog sentence as a weighing scale:  one puts the predicate on the left side of the scale, then places the topic, the ang phrase, at the right side of the scale so that now the two sides are completely balanced, are completely level.  I contrasted this with the standard English sentence, say “The man threw the ball” which is “transitive” — i.e., energy flows into the ball from the man, energy gets transmitted from the man to the ball.

This first picture contrasting with the second was my first Aha-Erlebnis regarding Tagalog.  Balance and equality vs. transmission.  This Aha-Erlebnis gained strength when I encountered Naylor’s claim that the canonical Tagalog sentence has a Predicate = Topic structure, though I did not completely understand what that equation meant in the cases of (ostensible) adjectives and verbs.  (I say ‘ostensible’ because Naylor persuasively argues that the ‘verbs’ at least are really nouns — on the syntactic level — in Tagalog.)  I submit that we can understand this this equality by thinking in terms of sets, i.e., of types.  The canonical Tagalog sentence works, as it moves from start to completion, first, by restricting the range of things possibly being talked about to members of a set, then by changing the status of one object of that set from ‘currently unidentified’ to ‘identified’  by equating it with an object known from the context by the time the utterance completes to be the object being talked about.  Typically, that object, coded by the topic, constitutes old information of some sort:  everyone has seen Brad Pitt enter the room, for example, or he has been talked about previously.  And typically, the predicate codes new information, or information that hits one with a renewed force that calls for a likewise renewed predication: Brad Pitt is beautiful.  ∃x ∈ maganda: x =  si Brad Pitt.  Maganda si Brad Pitt.

UPDATE (12/10/2011):  Beiged out a metaphor that is, while still useful to me as unfinished lumber, is likely to be confusing to anyone else.  Tried to clarify the concluding sentence in red.

UPDATE (12/14/2011)  Added the ∃x ∈ <<name of set>> statements.