2 Russell's (Reference Russell1905) theory of definite descriptions

Consider the definite description in sentence (3):

1. (3) The house was sold.

According to Russell's (Reference Russell1905) analysis, its logical translation would be (4) (ignoring the semantics of the past tense):

1. (4) Ǝx(H(x) & ~Ǝy(H(y) & x≠y) & S(x))

   i.e. there is an x which is a house and there is no y such that y is a house and non-identical to x and x was sold

Example (3) accordingly makes three claims:

1. (5)

   1. (a) Existence: there is a house

   2. (b) Uniqueness: there is only one house

   3. (c) Predication: this entity was sold

If sentence (3) is true, then each of the italicised sentences in (5a), (5b) and (5c) will be true, and hence (3) entails the conjunction of the italicised sentences (5a)-(5c).

The crucial distinction between (3) and the corresponding indefinite description (6) lies in the uniqueness claim:

1. (6) A house was sold.

2. (7) Ǝx (H(x) and S(x))

   i.e. there is an x which is a house and x was sold

The truth of sentence (6) requires that there should be at least one house that was sold. It is logically compatible with there being more than one such, or with one only, and so (6) is logically neutral to uniqueness and does not actually contradict it. Existence and predication entailments are shared between (3) and (6) and hence (3) entails (6) but is not entailed by it.

For any account of natural language Russell's theory of definite descriptions raises the question: what exactly does it mean for an entity to be unique? There are millions of houses out there so how do speaker and hearer co-operate and coordinate their references so as to understand a given unique one on a given occasion, in the manner of Grice (Reference Grice, Cole and Morgan1975)? The answer given in Hawkins (Reference Hawkins1978, Reference Hawkins1991) appeals to the pragmatic structuring of the universe of discourse and to the existence of pragmatic sets (P-sets) within which uniqueness is achieved in everyday discourse. Hawkins (Reference Hawkins1991) also explains why indefinite descriptions are neutral to uniqueness on some occasions, but contrast with the on other occasions and ‘implicate’ non-uniqueness (a senator can refer to one of the 100 senators of the US senate, but a president cannot refer to the unique president of the USA).

Consider now the definite article + modifier combinations of the present article, starting with (8), (9) and (10):

1. (8) The same house was sold.

2. (9) The identical house was sold.

3. (10) The similar house was sold.

Compare (8) and (9) first. The interesting, and at first apparently contradictory, point to note about the identical house in (9) is that it makes an existence and a uniqueness claim about the house in question in accordance with Russell's semantics in (4) and (5), i.e. that there is some house x and there is no house y non-identical to x. But at the same time we have seen that the semantics of identical does assert the existence of some other house y non-identical to x (recall example (2) with the identical ice hole)! The apparent contradiction is resolved by appealing to a pragmatically more fine-tuned universe of discourse within which the semantics of the identical house is interpreted, for example the ‘previous discourse set’ shared by a given speaker–hearer pair and containing entities that they have talked about (see Hawkins Reference Hawkins1991). Appropriate usage of the identical house can be achieved if this set contains just one house with the property of being identical to a second one whose existence is entailed, i.e. the uniqueness of the one is achieved by being the only one within the relevant set that has the property of being identical to some other house.

For the same house we have seen, in John and Mary live in the same house, that there does not have to be actual referential distinctness and that John and Mary could both be living in a single existing and unique house. This is the interpretation for same that appears to merely emphasise Russell's semantics for the whereby there is a house x and there is no other house y non-identical to x. The more explicit and logically entailed paraphrase for this is The house that John lives in is the (same) house that Mary lives in (the paraphrase goes through both with and without same) which compares and equates two definite descriptions, the house that John lives in and the house that Mary lives in, each of which requires the.

Let {Hj} stand for the set of properties associated with house that John lives in, and let {Hm} stand for the set associated with house that Mary lives in, and let us represent the house that John lives in informally as ‘the x ({Hj} x)’ and the house that Mary lives in as ‘the y ({Hm} y)’. The former achieves its uniqueness in the usual way by entailing that there is no other y distinct from x of which {Hj} holds, i.e. there is only one house that John lives in. The latter achieves its uniqueness also by entailing that there is no other x apart from y of which {Hm} holds, i.e. there is only one house that Mary lives in. Now, the crucial additional claim made by same is, in the more usual and straightforward interpretation for John and Mary live in the same house, that x = y, i.e. the unique x of which {Hj} holds is in fact identical to the unique y of which {Hm} holds. There is an equivalence between these two definite descriptions that figure in the more explicit and logical paraphrase for this sentence, i.e. between ‘the x ({Hj} x)’ and ‘the y ({Hm} y)’, resulting in the possible use of same.

But the equivalence between ‘the x ({Hj} x)’ and ‘the y ({Hm} y)’ does not have to involve the actual identity of x and y for the same to be used appropriately, as we have seen. John and Mary live in the same house could mean that John and Mary live in distinct referential tokens of a single house model, and this interpretive possibility is also present in the logically equivalent paraphrase The house that John lives in is the (same) house that Mary lives in as well. This meaning involving distinct referents will be pragmatically preferred in a sentence such as The house that John lives in was sold and so was the same house that Mary lives in since there is a much more direct description in the event that only one referential token is intended, namely The house that John and Mary live in was sold, and its absence in . . . and so was the same house that Mary lives in will lead to the inference that the referential tokens are different. The meaning of same is closer to that of identical when the comparison does involve distinct referential tokens, as we have seen. More generally, the form of equivalence between definite descriptions in logical paraphrases that are entailed by appropriate uses of the same is an equivalence in pragmatic referential possibilities, not a full logical or semantic equivalence, as we shall illustrate in more detail in the next section. It is also an equivalence that requires only a partial overlap between the definite descriptions being compared, ‘the x ({Hj} x)’ and ‘the y ({Hm} y)’, either at the level of the properties that figure in the definite description, {Hj} and {Hm} etc, or at the level of the referents x and y.

The semantic and pragmatic analysis of the similar house in (10) proceeds as for the identical house in (9). There is an existing and unique house x within some pragmatically defined set and the definite description achieves its uniqueness by being the only x in that set with partial property matching to some other referent y whose existence is entailed and also possibly within the same pragmatic set, resulting in appropriate sequences such as The house that John owns was sold and so was the similar house that Mary owns.

Notice finally in this section that sentences corresponding to (8)–(10) with the indefinite article, namely (11)–(13), involve ungrammaticality in the case of *a same, as we have mentioned, and a different semantics and pragmatic interpretation for an identical and a similar compared with their definite counterparts:

1. (11) *A same house was sold.

2. (12) An identical house was sold.

3. (13) A similar house was sold.

The indefinite article is incompatible with same, we claim, because the semantics of same involves a comparison and a pragmatic equivalence between two definite descriptions, each of which already requires the, e.g. The house that John lives in is the (same) house that Mary lives in. The product of this comparison and equivalence and its reduction to John and Mary live in the same house cannot be any less definite and less uniquely referring in the single NP, the same house, than it is in the two compared definite descriptions the house that John lives in and the house that Mary lives in that are being claimed to be pragmatically equivalent (see the further discussion of this in the next section). For identical house and similar house the description can be either definite as in (9)–(10), or indefinite as in (12)–(13), depending on whether there is a unique direct referent with full or partial property matching to the indirect referent(s). If no uniqueness in the manner of Russell's analysis is claimed for the direct referent, then the indefinite article will be used for identical and similar as in (12)–(13), and it will be asserted that there is at least one house (the direct referent) that is fully or partially matching in properties to the indirect referent(s).

3 Same and the pragmatic equivalence of definite descriptions

Same is inherently relational and comparative. So are identical and similar. They link and compare one entity with another. But they do so in different ways. Our proposal is that the comparison with same is between two logically entailed definite descriptions, with an assertion of ‘pragmatic equivalence’ between them. This pragmatic equivalence can result from the fact that the two definite descriptions have identical reference tokens (a single house token referred to by both, for example) or they may have different reference tokens but a sufficient sharing of semantic properties combined with real-world pragmatic knowledge (of models of house, for example, with their distinctive and criterial features) to justify being called the same. The comparison made with identical and similar is, we claim, more straightforward and less abstract and applies to linguistic and real-world entities directly, not to logically entailed definite descriptions of these entities. The difference between identical and similar is then one of full versus partial property matching between these entities. The greater variability in the interpretation of the same + Noun, which we shall now illustrate, is a consequence of this more abstract and linguistically specified comparison between descriptions of entities, as opposed to the direct comparison between entities themselves which is characteristic of identical and similar.

One consequence of this form of comparison with the same is that it allows for considerable ambiguity with respect to the unique entity that it refers to. This entity can be realised as one individual token, in the more usual interpretation of John and Mary live in the same house. Consider also the more usual interpretation of Mary was wearing the same dress as yesterday. Most plausibly yesterday's dress token worn by Mary was also the one worn subsequently. Alternatively there may be different house or dress tokens in these sentences, in their less usual interpretations. In the example John has the same nose as his father (example provided by Ekkehard König) the normal interpretation will be that we are talking about two distinct noses.

This ambiguity has sometimes been described in terms of token versus type meanings (and was so described in Hawkins Reference Hawkins1978). The notion of a type is not straightforward, however. It is often loosely defined in the literature, and its very existence has been fiercely debated (see e.g. Wetzel Reference Wetzel2009; Kearns Reference Kearns and Wetzel2010; Bromberger Reference Bromberger and Bromberger1992). Kearns (Reference Kearns and Wetzel2010) observes that once we list all the facts about each individual token (for example, in the case of US grizzly bears, how each one eats and behaves and what it looks like, etc.) and the generalizations about each of them, evoking the notion of a unifying type (i.e. the US grizzly bear) seems superfluous.

In the present context it is not helpful to talk about a ‘type’ meaning for the same in John has the same nose as his father, or in John and Mary live in the same house meaning the same model of house. This is in part because the same does not oblige us to consider all relevant tokens and compare their properties, which is what we have to do for identical, checking that all properties are shared. Moreover, if we try to impose a ‘type’ analysis on distinct tokens we soon encounter theoretical and descriptive problems involving the very definition of the type, what features define it, what properties the entities that belong to it must possess, how it is delimited and how type membership is determined. These are the kinds of problems for which Langacker (Reference Langacker1991, Reference Langacker, Pecher and Zwaan2005) and Breban (Reference Breban2010, Reference Breban2011) have proposed an alternative analysis in terms of referential ‘generalization’, an insight that we believe can be incorporated here for examples such as John has the same nose as his father.

What the same invites us to compare is two logically entailed definite descriptions, for example the house that John lives in, ‘the x ({Hj} x)’, and the house that Mary lives in, ‘the y ({Hm} y)’. The claim made by same is that there is at least some equivalence between these two that is pragmatically sufficient for the comparison and equivalence to be made. In the extreme case x=y and all the properties of {Hj} will be identical to those of {Hm}. This is full logical equivalence between definite descriptions and we see it realised when one and the same house token is involved. But when there are different house tokens (x≠y) there must then be some equivalence between entities at the level of their properties, i.e. between {Hm} and {Hj}, and just how much equivalence there needs to be at the property level, in order to describe relevant items as the same, seems to be pragmatically highly variable and context-dependent. That is why we claim that the kind of equivalence between definite descriptions that is required, in general, for the appropriate use of the same, may be full-bodied logical equivalence at the one end between referential tokens and their properties, but only a much looser and pragmatically sanctioned equivalence at the other, between some of the properties of {Hj} and {Hm}.Footnote ²

Consider some further examples that illustrate this variability in the interpretive possibilities for the same and that justify our position that the kind of equivalence that is required between logically entailed definite descriptions is pragmatic in nature, i.e. an equivalence deemed sufficient in context to justify the claim of ‘sameness’ based on some sharing at least of referential and/or semantic and/or real world properties, not a stricter form of logical equivalence between definite descriptions.

Consider the following:

1. (14) John and Mary saw the same white rhino.

This example can be interpreted in many different ways depending on the situation being described. The speaker may be referring to just one token (a singular entity ‘white rhino _i ’ that appeared either once or twice and that both John and Mary saw, together or on separate occasions) or he could be referring to more than one token of a common white rhino species. If it happens to be clear that there is only one such animal in the relevant context (in a zoo for example), then this single token will be linked to the two individuals, John and Mary, who may have seen it on one or more than one occasion. If this is not clear, on a safari for example, then the interpretation will be quite vague with respect to the number of white rhino tokens seen by these two individuals. In fact, neither the speaker nor John nor Mary may have a clue whether there was one or more than white rhino token that they saw.Footnote ³ The crucial point is that the referential status of the same white rhino may be irrelevant or unknowable in a given real-world situation, and this example highlights the flexibility and variability that has to exist pragmatically with respect to the possible referents of definite descriptions entailed by the same (the white rhino that John saw was the same white rhino that Mary saw). When our world knowledge is more constrained and there is just one white rhino (in the zoo), this indeterminacy is much less.

With respect to the descriptive content of the definite descriptions entailed by the same, in the event that referential tokens are shared (the house that John lives in is the same house that Mary lives in) then the real-world properties of {Hj} and {Hm} will, of course, be identical too. In the event that referential tokens are not shared (John has the same nose as his father) then the normal expectation may also be that the properties of John's nose {Nj} are identical to those of his father {Nf}, and this may even be an implicature in the sense of Grice (Reference Grice, Cole and Morgan1975), Sperber & Wilson (Reference Sperber and Wilson1995) and Levinson (Reference Levinson2000), requiring cancellation in different contexts in the event that the implicature is not intended, as seen in the following examples involving jackets:

1. (15)

   1. (a) John and Bill were wearing the same jacket but for the buttons.

   2. (b) John and Bill were wearing the same jacket but for the colour.

   3. (c) ?John and Bill were wearing the same jacket but for the style.

   4. (d) *This jacket and this dress are the same.

Cancellation of the implicature (‘all properties of the coreferential tokens are shared with same’) is possible in (15a) and even in (15b). Two jacket tokens worn by two individuals (or by one individual at different times or by one or more than individual at different times and places) can qualify as the same, whether or not the buttons are property for property identical. Even the colour can vary, as in (15b). Example (15c) is less felicitous, however, since the style of a jacket is a more inherent part of it, and the absence of this match renders it less plausible to view one of these entities as pragmatically equivalent to the other. (15d) is completely unacceptable because two entities that are as different as a jacket and a dress cannot be judged to be the same.

An extreme case illustrating the role of real-world knowledge in sanctioning an acceptable equivalence between two definite descriptions at the property level is the following. Every massive California redwood tree begins its life cycle as a tiny seed. Comparing a token of each, the fully grown tree and the seed, it would be possible to say: These are the same (tree); thereby establishing the link across times between the fully grown redwood and the seed that is, despite appearances, the same tree. The entities in question are known to change radically over time, but their very different appearances at different stages are not sufficient to block the pragmatic equivalence of two definite descriptions (the redwood tree that is fully grown is the same tree as the seed) describing exemplars at extreme ends of the life cycle with minimal property overlap but a form of identity provided by real-world knowledge. Notice, by the way, that we could not say that The California redwood and the tiny seed are identical (nor even that they are similar, see section 5), since hardly any of their properties match. But we can say that they are the same (tree), because of the time and life cycle link that is known to connect the definite descriptions describing these tokens.

Returning to jackets and the like, notice that whereas different colours may not be sufficient to block pragmatic equivalence between definite descriptions for these items, they may do so for others. Compare (16a) and (16b):

1. (16)

   1. (a) ?Their eyes were the same except for the colour.

   2. (b) The two women were the same except for the colour of their eyes.

In (16a) colour is an important and salient property of any pair of eyes, and one pair would not normally be deemed equivalent to another if that feature contrasts. Two women as a whole, however, possess many more properties than their respective eye colours, and so the comparison of one woman token with another can overlook this particular feature and establish sufficient equivalence between them at the level of their properties, despite their different eye colours. In the more limited context of eyes alone colour is too important a difference to render an equivalence plausible.

Notice finally in this section the interesting compound form selfsame in Modern English, which through the addition of the reflexive and intensifier form self to same results in preferred readings for the same that stress referential identity and co-reference rather than distinct referential tokens: John and Mary live in the selfsame house has this preferred interpretation. So does John and Mary live in one and the same house, where the addition of one also encourages the single referent interpretation. Interestingly, the -self of Modern English himself, herself, etc, was first used in the form of an adjective modifier of nouns and written self or sylf in Old English with the meaning of Modern English ‘same’, as in se sylfa God ‘the same God’ (from König & Siemund Reference König, Siemund, Frajzngier and Curl1999: 57; see also Farr Reference Farr1905: 18). The cognate form selb- in Modern German still translates same in Modern English (see König & Siemund Reference König, Siemund, Frajzngier and Curl1999 and Gast Reference Gast2006: 3). For selfsame in English, see more generally König & Siemund (Reference König and Siemund2000) and Keenan (Reference Keenan, Minkova and Stockwell2002) on the origins of reflexive pronouns in English and on their related intensifier uses (The Queen _i herself _i was there), and for other lexical overlaps across languages between items expressing identity between referential tokens see König & Siemund (Reference König, Siemund, Comrie, Gil and Haspelmath2005).

4 Identical

Semantically same and identical seem at first to be close (recall section 1). Their historical origins are different. Same is inherited in English from Common Germanic (via Norse; see Faarlund & Emonds Reference Faarlund and Emonds2014), whereas identical came from French (see Davidse et al. Reference Davidse, Breban and Van linden2008: 479).

Identical occurs with both a and the, as we have seen, though the use of the latter is pragmatically restricted to cases where there have been previous mentions, as in (17a) and also (18b) below.

1. (17)

   1. (a) Jackie was wearing an identical dress to the one that I was wearing. The identical dress caused a lot of trouble at the party!

   2. (b) Jackie was wearing a similar dress to the one I was wearing. The similar dress caused a lot of trouble at the party!

We mentioned in section 1 that a central difference between the same and identical lies in the number of entities referred to, namely reference to at least two for identical vs one or more for the same. For expressions with identical the identity is actually a predication (of full property-for-property matching) applied directly to these two distinct entities. For the same the comparison is between two logically entailed definite descriptions that describe the single or plural referential tokens in question. The co-occurrence restrictions of articles provide independent syntactic evidence for saying that identical does not in fact overlap semantically with same: identical pairs better with similar since they can both co-occur with a.

Identical and similar share a fundamental semantic feature: they predicate their description, namely full versus partial property matching respectively, of at least two separate entities. When they occur within referential expressions they maintain this plurality of reference. The meaning of identical can be described as universal quantification over all the properties of these distinct entities (i.e. all their properties are shared). The meaning of similar involves existential quantification (some of their properties are shared). Plurality of entities is inherent in the meaning of phrases containing identical and similar, whereas same occurs in NPs that can have a single referent. This contrast can be seen again graphically in the following minimal pair:

1. (18)

   1. (a) I have just seen the same twin (that I saw yesterday).

   2. (b) I have just seen the identical twin (to the one I saw yesterday).

(18a) refers to ‘the twin _i ’ that I have just seen and links this entity to the coreferential ‘the twin _i ’ that I saw yesterday. (18b), by contrast, links ‘the twin _i ’ that I have just seen to the other, referentially distinct twin ‘the twin _j ’, that I saw yesterday. The use of identical, which is appropriate and commonplace when referring to twins, necessarily involves a referential plurality and a reference to two quite distinct individuals; the same twin involves coreference here to a unique one. Notice, interestingly, that the identical twin in (18b) is appropriate as a first-mention definite NP referring back to the other twin (Yesterday I saw one twin: Today I saw the identical twin). This further confirms the plurality of reference inherent in identical: (18b) refers directly to ‘the twin _i ’ that I have just seen while at the same time asserting the existence of another ‘twin _j ’ and it acquires its anaphoric definiteness by reference to this other, previously mentioned ‘twin _j ’. Since identical, like similar, involves reference to a plurality, these modifiers can combine with a to refer non-uniquely to just one of a plural set (an identical twin like an identical coat, also a similar coat [to the one Mary was wearing]). In a parallel way a prince can refer to one of the princes of England, a senator to one of the US senators, a window to one of the house's windows, and so on (Hawkins Reference Hawkins1978, Reference Hawkins1991).

Consider now some further examples that highlight this difference between identical and the same:

1. (19)

   1. (a) Jane was wearing an identical jacket to the one Mary wore yesterday; in fact, it was the same [one]!

   2. (b) Jane was wearing an identical jacket, but it was not the same [one that Mary wore yesterday].

   3. (c) Jane was wearing the same jacket, not an identical one.

   4. (d) Jane was wearing the same jacket that Mary wore yesterday, or rather, an identical one.

In (19a) the speaker first believes that there are two jacket tokens and then realises that there was just a single one and so corrects the reference to the same, after first using an identical. Example (19b) clearly signals that the jacket in question was a separate token from the token in an earlier reference (to the jacket that Mary wore the day before). Example (19c) emphasises the singularity of the jacket token, i.e. there was only one jacket, not two that were identical. In (19d) we have the same situation as in (19b), but in reverse. The speaker first thinks there was only one token, and then realises that there were actually two and decides to correct himself.

When the pragmatic interpretation of the same involves distinct referential tokens, however, the contrast with identical is much less and both (20a) and (b) can be used:

1. (20)

   1. (a) Mary and Jane wore an identical jacket.

   2. (b) Mary and Jane wore the same jacket.

In (20a) two distinct jacket tokens are being referred to, and it is asserted that they match one another in all their properties, hence they are identical. (20b) with the same has the more usual interpretation that there are again two jacket tokens and not just one, as there might be in a circus act with two people in one jacket, and it achieves this similar referential effect through the comparison and pragmatic equivalence between logically entailed definite descriptions, the jacket that Mary wore is the same jacket that Jane wore. It is commonplace in everyday language use for one and the same situation to be describable in different ways (e.g. the candle is on the candle holder vs the candle holder is under the candle) and this is what is going on in (20). A single situation or event can be conceptualised and lexicalised in different ways. The manner in which similar reference is achieved in (20a) and (b) is also different, as we have seen.

Notice that even though identical involves the matching of all properties across different tokens in the normal case, a certain latitude in its descriptive meaning is also permitted. The following are appropriate uses of identical even when the sharing of all properties is explicitly denied:

1. (21)

   1. (a) Jackie was wearing an identical dress to mine, except for the colour.

   2. (b) [pointing to a pair of gloves] I bought him an identical pair of gloves, just a bigger size.

Sentences (21a) and (b) are understood as qualifications of, or exceptions to, the full property-for-property matching claim, but not as contradictions. Universally quantified sentences with all also permit such exceptions without apparent contradiction, as in All the boys were having a good time, except for Charlie. The exception qualifies and does not contradict the universal claim made about all the boys, just as (21a) and (b) qualify and do not contradict the claim that all properties are shared between the two dresses and two pairs of gloves. What we wish to highlight here is simply the difference between identical and same with respect to the number of entities referred to, a necessary plurality for identical versus a possibly single one with same, as well as the manner in which the comparison is made, through direct reference to entities on the one hand versus a comparison of definite descriptions on the other. The plurality of reference tokens required for identical makes this item closer to similar than to same, and this is reflected in their parallel article co-occurrence possibilities contrasting with the/*a same. On the other hand the pragmatic equivalence between definite descriptions logically entailed by the use of the same may involve a sharing of all referential and semantic properties between definite descriptions, or it may involve only a partial sharing of these properties (recall the same white rhino and the same redwood tree examples above), which means that same can overlap semantically with both identical and similar on different occasions of use.

5 Similar

Similar is usually preceded by the indefinite article, except when it is previously mentioned and the context allows the use of the definite article (recall (17b)), as was the case for identical (in (17a)). Grammatically, similar patterns like identical, therefore. Semantically, it differs from identical in that the concept of similarity involves the sharing of only some criterial properties and not of all. For example:

1. (22)

   1. (a) Jackie was wearing a similar dress to the one I was wearing.

   2. (b) Bill bought a similar car to ours.

   3. (c) He saw a similar tree to the one we saw. The similar tree was by the orange house down the road.

In examples (22a–c) we have different tokens that share a number of properties (some but not all), by virtue of which they can be referred to as similar to one another. This near-identity together with the potential variability with regard to which properties are shared and which are not allows for an infinity of similarity descriptions and hence for the indefinite article (unless prior reference is available, as in 22c). The same, by contrast, does not have a focus on property-matching between distinct individuals but rather on the pragmatic equivalence between definite descriptions that are logically entailed by the sentence containing the same.

The reason why the is uncomfortable with similar is because the sharing of only some criterial properties makes the existence of other, similar entities inevitable, in general, which conflicts with the uniqueness of definiteness. When the nature and number of other similar entities can be brought under pragmatic control, as in (22c), the definite article becomes possible with similar. The interpretation of the definite description in (22c) is relativised to a set of just two trees and because there are just two entities in the pragmatic set under consideration, each becomes unique compared to the other (see also Hawkins Reference Hawkins1978: 250). This conflict between the and similar is also present between the and identical. In contrast to identical and similar, the uniqueness of definiteness maps perfectly onto the meaning of same, since same involves a form of double definiteness and pragmatic equivalence between two definite descriptions that are logically entailed by a sentence containing the same. Since each compared definite description is uniquely referring and requires the, the product of the comparison, the same house etc. can be no less definite and uniquely referring.

Notice finally in this section that the adjective different is in many ways the negative counterpart to similar in English, predicating of two or more entities that ‘at least some properties are not shared between them’, as in John's jacket is different from Bill's.