A recent observation from watching a toddler (20 months) learn to speak: she seems to treat words and names as equivalent, and watching her do so makes the distinction in my mind apparent. She is at a stage in her linguistic development where she’s learned to pay attention to speech, can distinguish individual words in a sentence, will interpret words as attaching to objects based on their use in the context of the sentence, will request the words for objects as a sort of game, and will parrot words as they’re given to her, or when prompted to name an object.
But she doesn’t yet have much in the way of grammar. She can form some two word phrases, in particular combining manual signs with spoken words to form compounds (e.g. sign for “more” with spoken word “cracker”). But these compounds are few in number and used frequently, making them seem more like two syllable words for a single concept (the act of giving her a cracker) rather than a combination of two concepts (act of giving + object to be given).
Related to that, she is also confused when items don’t have specific words that describe them. For example, she has a toy made up of plastic doughnuts that can be stacked. When she asks for the word for one such donut, I don’t have one to give her because I don’t think of these objects with a single word, but rather a collection of words that describe it: a hollow yellow plastic donut or ring. Since she doesn’t have a grasp of noun phrases, she can’t parse that, and just looks confused.
From these interactions, I wonder if what she is asking for are perhaps not words, but rather names, and I have been supplying names for objects. In my mind some objects don’t have names, but can nevertheless be described by words. In adult speech, many words can be applied to an object depending on their salience, so a toy may be a stuffed animal and an elephant. But her use of words seems limited to their naming use, rather than their more abstract use as words. She can recognize e.g. a dog as a dog, but probably can’t recognize that a dog is at once a dog and a pet and an animal. It isn’t clear that she will assign multiple words to the same object, which makes sense if they are acting as names rather than words (though this is hard to test: if she doesn’t understand a request, e.g. “bring me the doll”, she will guess, and her guesses draw on both the words used for the request and the gestures and attention of the requester).
Evidence against her treating words as names is her readiness to generalize, which she does both rapidly and accurately, e.g. identifying most dogs correctly after only seeing two or three (though not perfect; she confuses horses and cows despite several attempts to distinguish them). Names are specific to an object rather than used for any member of a class. However, it is consistent with a hypothesis that the distinction between words and names is not sharp for her, and thus learned rather than inborn. This also fits with known a lack of abstraction in early speech: names for members of the class ‘dog’ are understood, but attribute classes such as color, shape, texture, and number are more abstract, and must be built from concrete concepts.