How a Lexical Unit Profiles a Schema

From the frame's grounding to the word's own profile#

A frame's grounding gives every one of its lexical units access to the same schema, but two lexical units of one frame rarely mean exactly the same thing — they differ in which of the frame's own frame elements each one actually brings forward. That selection is the lexical unit's own profile, and because the frame element itself is already bound to a schema element, the profile reaches all the way down: a lexical unit doesn't just foreground a frame element, it foregrounds a specific piece of conceptual structure.

The profile can be ordered. The highest-ranked frame element in a lexical unit's profile is its trajector — the figure the rest of the profile is read against — and a genuinely symmetric relation is simply left unranked, with no trajector at all. Rank is what lets two lexical units of one frame that both bind the same frame elements still differ in meaning, by disagreeing about which one is figure and which is ground.

Dimension or value: which one a word actually names#

The clearest, most frequent split a conceptual profile records is whether a word names a dimension itself or a position on it. FrameNet routinely places both in one frame, correctly — they really are one structure — and the profile is what tells them apart:

For a value produced by a change, the pattern extends to a third member — the verb that produces it. Cansaço ("tiredness") is the dimension; cansado ("tired") is a value on it; cansar ("to tire") is the change that reaches that value. It's worth being precise about what kind of word cansaço actually is here: it names the dimension itself, not a nominalized event. Two independent facts confirm this. O cansaço durou dois dias ("the tiredness lasted two days") describes a state persisting, where o cansaço aconteceu ("the tiredness happened") is simply ill-formed — an event-denoting noun tolerates that kind of sentence readily, a dimension-denoting one doesn't. And the morphology agrees: -mento productively nominalizes an event (casar → casamento, "to marry" → "marriage," which does take aconteceu), while -aço in cansaço names a condition instead. The real evidence for the underlying change sits with the verb, cansar — a nominalization may or may not exist alongside it, and its absence or presence says nothing about the dimension/value split itself.

Some dimensions lean toward one of their own poles rather than sitting neutrally between them — cansaço, beleza, inteligência ("tiredness," "beauty," "intelligence") all presuppose a preferred direction, where altura and temperatura ("height," "temperature") don't presuppose tall or short, hot or cold. That's a fact about the dimension itself being valenced, not a different kind of profiling — the lexical unit still simply profiles the dimension, exactly as it would for a neutral one.

One relation, three characters#

A value's own character depends entirely on what dimension it sits on, not on any special device of its own: a value on a class is categorial (casado, "married"); a value on an unmeasured, ordered dimension is gradable (alto, "tall"); a value on a measured one is measurable (3kg). All three are the identical structure — a bearer holding a value on a dimension — differing only in what the dimension itself turns out to be.

Whether a value is read as inherent or resultant is a similar, single-relation distinction: an inherent value is simply held (alto), where a resultant one sits on a trajectory some change produced (cansado, from cansar). Because a language can always construe a resultant value as if it were simply inherent, this isn't two different structures needing two different devices — it's the same one, with that connecting relation present or left out of view.

One construal, or two words#

A frame can carry more than one core frame element in view of a single lexical unit at once — several frame elements profiled together as one configuration, not competing readings of the same word. Where two distinct senses of one form each foreground a genuinely different frame element depending on how the word is used, that's ordinary polysemy instead: two lexical units, each with its own narrower profile, sharing one frame and one schema between them.

What this adds beyond the ontological type#

Every lexical unit also carries an ontological type — a coarse classification of what the concept itself denotes, described in the ontological dimension's own LU. It's worth being precise about how that type relates to the profile described here, because the two look similar and record genuinely different facts. The type records what a word denotes; the profile records what it foregrounds — and for a verb, those two come apart completely. Comprar ("to buy") denotes a buying event and is typed accordingly, while its own profile foregrounds two of that event's participants, a buyer and goods. Reading the type off the profile's binding target would say comprar names a buyer, which is simply wrong. Only where a lexical unit profiles an element that belongs to another frame entirely — the way comprador ("buyer") profiles the buyer role of the very frame comprar evokes — does the profile actually determine the type, because only then does the lexical unit denote the thing it's pointing at rather than the scene it's part of. This is one instance of a systematic pattern — an entity LU in a classificatory frame profiling an event frame's FE elsewhere in the network — worked through in full in Profiling.