Millikan is one of the most important naturalists in the philosophy of mind. She is known as an outstanding defender of “teleological” theories of content, but she also has defended a naturalistic theory of concepts (²⁰⁰⁰) and a biological account of the nature and development of inner representations (²⁰⁰⁴). In Language: A Biological Model, Millikan continues with this general naturalistic-biological approach, now focusing on the language faculty and some other issues of interest to philosophers of language, such as the relation between language and thought, the utility of conceptual analysis, the nature of illocutionary acts, and the distinction between semantics and pragmatics.

The book is composed of ten different articles. Two of these are entirely new while two others are substantial revisions of previously published papers; the remaining six are more or less the same as in previously published versions. Although the biological perspective gives some unity to the collection, the topics addressed across the ten articles are various and divergent, and it would be impossible to discuss them all here. For this reason, I will focus on three of the main ideas proposed by Millikan across the book: the idea that natural language is conventional but doesn’t require regular conformity or prescriptive rules (discussed across most of the papers but mainly in “Language Conventions Made Simple”, “In Defense of Public Language”, “On Meaning, Meaning and Meaning” and “Proper Function and Convention in Speech Acts”); the idea that there is a useful notion of public language (mainly discussed in “In Defense of Public Language”); and the idea that understanding language doesn’t require thoughts of speaker intentions because it is a form of direct perception of the world, and therefore children don’t have to have a theory of mind to learn language (mainly discussed in “Semantics/Pragmatics: (Purposes and Cross-Purposes)”).

Millikan shares with some linguists and philosophers the idea that conventions are central to language, but she has her own view about how we should understand “convention”. She questions the idea that conventions involve prescriptive rules and/or regularities. Contrary to much philosophical opinion, Millikan claims that conventions need not be adhered to either by the majority of the members of a com- munity or by an average member of it. According to her, conventions in language are not a matter of what speakers do most of the time. Rather, conventions are patterns of action that fulfill a function and involve both hearers and speakers, and that have been reproduced partly due to weight of precedent (and not because they are the only or the best patterns capable of performing the function they are performing). A pattern has been reproduced in this sense (and not for example, reproduced due to genes) so long as “its form is derived from a previous item or items having, in certain respects, the same form, such that had the model(s) been different in relevant respects the copy would have differed accordingly” (p. 3).

Thus, all that is required for a linguistic pattern to be conventional is that it be repeated often enough to survive, and that such repetition be somewhat arbitrary (other patterns may have served as well for the function that the pattern is realizing). There is no requirement that a speaker be aware of rules or norms. For Millikan, the copying that occurs in reproducing conventional linguistic patterns is a purely causal matter; so it is not required that the speaker be conscious, or intend, to reproduce the pattern.

Take, for example, the use of the word “book”. The word “book” in English conventionally means book because often enough the members of a certain group of language users utter that sound and not some other when they want to talk about books. The reason why that particular group of sounds is used is simply that it has been used for the same purpose in the past, and not because it is intrinsically better than any other for that job or because there is a rule that tells one that it should be so used. At the same time, thinks Millikan, speakers will usually reproduce it without even giving it conscious attention; i.e., without explicitly intending to follow the convention according to which people call books “books”.

Now, this view of what conventionality involves is not without problems. Its main problem is that under it, practices that we wouldn’t consider to be conventions turn out to be conventions. Consider first a case of something a person does out of habit, such as using the same cup every morning for her first coffee. This is a behavior that has been reproduced due to “weight of precedent”, and it seems that some other pattern of behavior, i.e., using a different cup, could fulfill the purpose of the behavior, namely getting coffee. Are we going to call this a convention? It seems not. It seems that in order for a pattern of behavior to be considered a convention it should be present in more than one person. This is something Millikan seems to accept. The problem is that it seems that the fact that a pattern of behavior has been copied socially doesn’t seem to be enough for that pattern to be conventional. It seems that social norms have to be involved, that some notion of doing things right or wrong has to be involved.

Millikan doesn’t accept that conventions have to do with social norms, but it seems that something less than that won’t do. To see this, consider cases in which a pattern of behavior is copied from one person to the other, but there are no social norms involved, and the person reproducing the pattern has no intention to reproduce it. Suppose for example that you start scratching your nose very often when you are talking to strangers because you unconsciously copied this gesture from your partner. We can say that this gesture is reproduced due to weight of precedent, we could even claim that it has a function that other patterns of behavior could perform as well (for example, it helps you and your partner deal with tension in social situations). Would we claim that scratching your nose is a convention? Of course not; and it seems that the reason not to consider it a convention is precisely because you really don’t intend to copy this gesture, and because there is no sense in which we could say that by scratching your nose in the presence of strangers, you are doing something that is right or wrong within a social group. By contrast, consider the convention of distributing cigars among your friends if you become the father of a boy (this is one of Millikan’s examples of a convention). It seems that the reason that this is a convention is not only that you are copying this social behavior but that you are doing it because you want to do what you think is done within your social group, what you think is right.

Coming back to the case of the person that uses the same cup every morning for her first coffee, suppose that this person uses the same cup every morning because this is what she saw her parents do while she was growing up. Now it seems that we could consider this behavior as being conventional. Think why. If we consider this behavior conventional -within this family- it is not only because one member of the family has copied it from another member, but because, within that family, the right thing to do is to use the same cup, your cup, every morning, and the person uses the same cup every morning because she intends to do things right. It is this, and not only the fact that the person has copied the pattern of behavior from someone else, which makes the pattern conventional.

Thus, in conclusion, it seems that social norms, and an intention to follow them, are involved in conventional behavior. This is also the case for language conventions. They seem to involve intentions, since speakers have, at least, the intention to communicate, and norms seem to play a role, since there are right ways and wrong ways to say things.

Let’s turn now to Millikan’s discussion of the notion of “public language”. Against many language theorists, who think of a pub- lic language in terms of an ahistorical abstract system of symbol-types governed by discrete sets of rules, Millikan holds that accumulated facts about past linguistic performances are determiners of contemporary semantic and syntactic facts. She claims that she agrees with Chomsky in that there are no “fixed entities” which a child slowly approaches when learning a language, and in that public language shouldn’t be understood just as some property that two people can share. However, against Chomsky, she believes that there is a useful notion of public language, one according to which a public language is “a sort of stuff in the real world” (p. 38), consisting of all of the conventional patterns of interaction between speakers and hearers within a language community. The set of these structures, claims Millikan, is what constitutes a public language. She claims that “a language consists in a tangled jungle of overlapping, crisscrossing traditional patterns, reproducing themselves whole or in part for a variety of reasons... ” (p. 18). This is, she thinks, a legitimate way of looking at language as a public object, as something that can be learned and doesn’t consist in merely acquiring an internalized language in Chomsky’s sense. Learning a language is, under Millikan’s view, coming to know various public conventions.

Chomsky probably would agree with Millikan on this point, if he took conventions in the same sense as Millikan does. Giving this meaning to the notion of “public language” makes it wide and innocent enough. There are, however, some questions one may want to ask. Considering that there are different public languages, how are we going to distinguish among them? It is no use to claim that what distinguishes one public language from another is the linguistic com- munity to which it belongs, since there seems to be no non-circular way of determining what would count as a linguistic community. At the same time, it seems impossible to delimit a number of conventional patterns that give identity to a language, since lots of linguistic conventional patterns are shared among people that speak different languages and people that speak the same language don’t share all of their linguistic conventional patterns. These considerations may make one wonder what the use of a notion of “public language” in Millikanian terms can be, if public languages are not specifiable things.

Finally, let me comment on Millikan’s idea that language learning doesn’t require a theory of mind because language understanding doesn’t depend on accessing the intentions of speakers.

Millikan claims that during normal conversation, it is not language that is most directly perceived by the hearer but rather the world that is most directly perceived through language. This, she says, implies that children learn what patterns of language mean exactly as they learn what is “meant”, for example, by the patterns of ambient light in which their eyes are bathed.

Unlike what the Gricean pragmatic approach to language under- standing suggests, Millikan’s proposal is that, in interpreting what a speaker is saying, the hearer doesn’t have to reconstruct what the speaker has in mind. What the hearer does in interpreting the speaker is to recognize which reproductive family a particular linguistic to- ken comes from. For example, when listening to an utterance of the word “bank”, the hearer has to decide whether it is reproduced from previous utterances in which “bank” referred to the place where you keep your money or whether it is reproduced from previous utterances in which “bank” referred to the bank of the river. To do this, claims Millikan, the environmental conditions in which the communication takes place are usually enough. The idea is that the hearer can interpret what the speaker is saying on the basis of their shared context. The hearer does not need to know anything about the intentions of the speaker. The meaning of what the speaker is saying is not in the speaker’s head; interpreting utterances is just another way of perceiving the world. She says that “interpreting the meaning of what you hear through the medium of speech sounds that impinge on your ears is much like interpreting the meaning of what you see through the medium of light patterns that impinge on your eyes” (p. 205). Thus, when communication proceeds normally, the hearer directly perceives the world through the words, and not the speaker’s thoughts and intentions. Since recognizing intentions of thoughts of speakers is not necessary for understanding a language, claims Millikan, children don’t have to know anything about thoughts and intentions to learn a language, so they don’t need to have a theory of mind.

Although it is probably true that children don’t need to have a developed theory of mind to learn a language, as shown by the fact that children learn a language before they can distinguish their own beliefs from the beliefs of other people, it is not at all clear that they don’t need to have some idea of the intentions of others in order to develop the degree of proficiency at language that normal adults have. It is maybe true that they can learn linguistic meanings without knowing about the intentions of other people. But understanding a language involves more than understanding linguistic meanings, it involves understanding what speakers want to say. Understanding metaphors and irony, for example, involve more than understanding literal linguistic meanings; and metaphors and irony are precisely the kind of language patterns that normally escape the capacities of understanding of children that don’t have, or don’t have yet, a developed theory of mind. So it seems that at least some uses of language require, not only that the speaker and hearer share a context, which may be enough to understand the literal meaning of words and sentences, but also that both know about the intentions and thoughts of the other.

The need to understand the intentions of speakers in order to understand what they are saying seems to be present even in cases in which the speaker just wants to communicate some information. Suppose for example that I am giving you instructions to pick up some papers from my office, and I tell you: “The key is a bit tricky but I really need those papers.” What do you need in order to be able to understand this sentence? Millikan would say that the context should be enough. Both you and I know that I am asking you to pick my papers and that for that you need to enter into my office. We both know that normally to enter locked places you need a key, and suppose you also know that my office is locked, and that sometimes keys are tricky. So far so good. But for the sentence to move you to action you need to interpret it as saying that, no matter how difficult it is for you to get into my office, I need you to pick up those papers. However, the sentence doesn’t say that explicitly, and knowing the literal meaning of the words that compose it is not enough to really understand what I am trying to convey. To interpret my sentence as implying that no matter how difficult it is for you to get into my office, I need you to pick up those papers, you have to know something about my intentions. You have to know that I don’t intend to complain about the trickiness of the key and that I am not just informing you about my needs. You need to know that I intend you to pick up the papers and that I am informing you that it may be difficult to enter into my office because the key is tricky. We can see that it is not that easy to explain how hearers understand what speakers say without attributing to hearers some understanding of what speakers intend to convey. Here, as with conventions, intentions seem to play a role.