A perspective-and-aspect-driven system of temporal anchoring in Portuguese: semantic reanalysis of the present perfect (PPC) and other Cross-Romance divergences

A Perspective-and-Aspect-Driven System of Temporal Anchoring in Portuguese Theoretical Analysis of the Present Perfect and Other Cross-Romance Divergences Shinya Makino Abstract: This paper proposes a feature-based analysis of the indicative and subjunctive tenses in Portuguese, grounded in four primitive anchoring-aspect features and their figurative extensions: [non-assertive], [VP∈past], [finished], and [assumptive]. In contrast to the tensedriven systems typical of other Romance languages, Portuguese employs a perspective-andaspect-driven system. Like those systems, it takes the speech time as the anchoring point. Portuguese, however, instead of establishing reference points such as [±past], divides the time axis into two intervals: one extending backward and excluding the speech time — the PAST — and the other extending forward and including it — the NON - PAST . Simultaneously, the speaker's temporal viewpoint is located either within the NON - PAST or within the PAST . From this viewpoint, the situation to be described is then evaluated as either finished or not finished. Its temporal anchoring is thus secondarily derived from the location of the temporal viewpoint and whether the situation is viewed as finished or not from that viewpoint. This structural opposition between the two systems accounts for a range of cross-Romance differences, including the uniquely Portuguese interpretation of the present perfect, the functional erosion of the future tense and the associated future-oriented use of the preterite, and the remarkable vitality of the preterite in this language. 2 0. Preface to the Preprint Edition This preprint (v2) builds on existing research while introducing a unified framework — grounded in feature-based structure, paradigmatic saturation, functional competition, and aspectual reinterpretation — for the semantic reanalysis of the Portuguese present perfect indicative (pretérito perfeito composto, PPC). Rather than attributing this change to morphosyntactic restructuring or lexical drift, the account locates its source in a deeper structural constraint. At the time PPC emerged, Portuguese's core temporal-aspectual system — defined by a 2×2 grid of [±VP∈past] and [±finished] — was already fully saturated by the present, preterite, imperfect, and pluperfect indicative. Given that the core system was completely saturated, any additional form would have risked creating redundancy. Originally, PPC₁ was specified as [−VP∈past, +finished, +resultative], directly overlapping with the simple preterite (pretérito perfeito simples, PPS: [−VP∈past, +finished]). In order to remain viable, PPC₁ set aside the feature [+finished] — which placed it in direct competition with the PPS — and, through a synecdochic reinterpretation of its result state ― which, importantly, the PPS could not express independently without adverbials (e.g., já, ainda não) or contextual/ situational cues, and which was the only oppositive feature between the two — it became PPC₂: PPC₁ → PPC₂ [−VP∈past, +finished, +resultative] ↓ [−VP∈past, +finished, +resultative] ↓ [−VP∈past, ✓ initiated, ✓ stative, −finished] Building on this structural reanalysis, the feature [ ✓ stative] came to coerce an iterative-stative interpretation of eventive situations. Through habitual use, this interpretation gave rise to a cognitive frame — "the initiation and continuation of an iterative state from a certain point in the past up to the speech time" — which, once associated with PPC₂, began to trigger an iterative-stative reading even for inherently stative predicates. When the context fails to align with this frame, however, the frame is neutralised, yielding instead a simple continuous-stative interpretation. 3 In sum, PPC evolves from PPC₁ ([−VP∈past, +finished, +resultative]) to PPC₂ ([−VP∈past, −finished, ✓ initiated, ✓ stative]), acquiring a unique profile among Romance perfects by ceasing to encode completed events with present relevance. While previous studies have offered valuable insights, this paper builds upon them in an attempt to provide a more integrated explanatory model. The author welcomes scholarly feedback and dialogue prior to peer-review submission. 4 1. Introduction and Theoretical Orientation Excluding the imperative mood, which does not exhibit any temporal opposition, Portuguese finite verb forms are classified into the following combinations of mood and so-called tense1: ( 1 ) Indicative: ･Present ･Preterite ･Imperfect ･Pluperfect ･Present Perfect ･Future ･Future Perfect ･Past Future ･Past Future Perfect Subjunctive: ･Present ･Preterite ･Imperfect ･Pluperfect ･ ＿ ･Future ･Future Perfect ･ ＿ ･ ＿ As shown in (1), the subjunctive mood has no form corresponding SEMANTICALLY to the present perfect indicative, and neither the past future subjunctive nor the past future perfect subjunctive exist. In this paper, we shall analyse the semantic oppositions among these forms, taking as our point of departure the opposition — assumed to be primitive — between the present and the preterite in the indicative mood. The analysis reveals that each tense form corresponds to a specific combination of the following four primitive functional features, with semantic extension through conceptual metaphor, synecdoche, and metonymy applied to them: mood [non-assertive], perspective [VP∈past], aspect [finished], and modality [assumptive], each marked either positively [+] or negatively [−]. Furthermore, it is shown that the negative value [−] may assume the neutral value [0] when licensed by a specific context or situation. The analysis also shows that two highly marked combinations — namely, [+non-assertive, +VP∈past, +finished, +assumptive] and [+non-assertive, +VP∈past, −finished, +assumptive], corresponding to the theoretically possible past future perfect subjunctive and past future subjunctive — are absent from the system. This corresponds to the cross-linguistically attested tendency to avoid overly marked grammatical forms. Additionally, provided that the temporal viewpoint is located exclusively on the speech time, an exceptional opposition marked by the 1 The corresponding Portuguese names for these moods-tenses are, in the indicative mood, Presente, Pretérito Perfeito Simples, Pretérito Imperfeito, Pretérito Mais-que-Perfeito, Pretérito Perfeito Composto, Futuro, and Futuro Composto do Indicativo, along with Condicional and Condicional Composto; and in the subjunctive mood, Presente, Pretérito Perfeito Composto, Pretérito Imperfeito, Pretérito Mais-que-Perfeito, Futuro, and Futuro Composto do Conjuntivo. It should be noted, however, that the English terminology for Portuguese tenses is notoriously inconsistent, with a single tense often being referred to by multiple names. For example, the Pretérito Perfeito Simples may be translated as Perfect, Past, Simple Past, Preterite, or Simple Preterite. 5 aspectual feature [initiated] and the Aktionsart feature [state] is observed only within the least marked area of the system — namely, [−non-assertive, −VP∈past, −finished, −assumptive] — and corresponds to the present indicative, marked as [0 initiated, 0 state], and the present perfect, marked as [ ✓ initiated, ✓ stative]. The combination of the four primitive features — [non-assertive], [VP∈past], [finished], and [assumptive] — each taking either a positive or a negative value, logically yields 16 finite verb forms. Of these, two highly marked forms are excluded, and one additional form is added, resulting in a total of 15 actual forms. These are exemplified in (2), using the verb falar in its third-person singular forms. The least marked area and the most marked one are shaded, and the areas not filled with actual forms are indicated with strikethrough: ( 2 ) [−finished] [+finished] [−finished] [+finished] [−VP ∈ past] [0 initiated , 0 stat ive ] fala falou falará terá falado [−non-assertive] [✓initiated, ✓stative] tem falado [+ VP ∈ past] falava tinha falado falaria teria falado [−VP ∈ past] fale tenha falado falar tiver falado [+non-assertive] [+ VP ∈ past] falasse tivesse falado ― ― [−assum ptive] [+assumptive] This system differs fundamentally, despite superficial similarities, from that of Reichenbach and its theoretical successor, the tense system proposed by Klein. In the latter frameworks, the temporal system of a language with absolute tense is wholly anchored in the speech time, and the temporal location of a situation — state or event — is determined with respect to this absolute point of reference. Even among the Romance languages, Spanish — along with French and Italian — is one such language. Its system of temporal anchoring first determines whether the situation to be described belongs to the actual world or to a non-actual one — a distinction encoded by the mood feature [±non-assertive]. Next, starting from the speech time, two reference times are established relative to it: the past, which precedes the speech time, and the non-past, which coincides with it. These two reference times are linguistically encoded by the absolute tense feature [±past]. For each of these reference times, two sub-reference times are then defined: the relative future, which is posterior to the reference time — marked as either [−past] or [+past] — and the relative non-future, which is simultaneous with it. This opposition is encoded by the relative tense feature [±posterior], which in Spanish is metonymically derived from the modal feature [±assumptive]. With reference to one of the four referencetime combinations thus established — namely, [−past, −assumptive], [−past, +assumptive], 6 [+past, −assumptive], and [+past, +assumptive] — the situation time is evaluated as being either prior to or simultaneous with it. This evaluation is encoded by the relative tense feature [±anterior]. Finally, only in the case of situations marked as [−non-assertive, +past, −assumptive, −anterior] — and, in literary written Spanish, also those marked as [−non-assertive, +past, −assumptive, +anterior] — does the speaker mark two alternative locations for the temporal viewpoint, depending on whether it is situated on the non-past reference time or on the past reference time. This opposition is encoded by the perspective feature [±VP∈non-past]. We shall refer to this type of temporal system as a tense-driven temporal system. The combination of the four major primitive features — [non-assertive], [past], [assumptive], and [anterior] — each taking either a positive or a negative value, logically yields 16 finite verb forms. Of these, two highly marked forms are excluded, and two additional forms — introduced by the feature [±VP∈non-past] — are added, resulting in a total of 16 actual forms that can be found in litterary written Spanish. These are exemplified in (3), using the verb hablar in its third-person singular forms. The least marked area and the most marked one are shaded; theoretically possible areas not filled with actual forms are indicated with strikethrough; and forms rarely used in modern standard spoken and written Spanish are shown in italics: ( 3 ) [−assumptive] [+assumptive] [ − past] hablo ha hablado hablará habr á hablado [−non-assertive] [+past] [− VP ∈ non - past] hablaba [− VP ∈ non - past] había hablado hablaría habría hablado [+ VP ∈ non - past] habló [+ VP ∈ non - past] hubo hablado [−past] hable haya hablado hablare hubiere hablado [+non-assertive] [+past] hablara hubiera hablado ― ― [ − anterior] [ + anterior] [ − anterior] [+anterior] (4) is identical to (3), except that the italicised forms rarely used in modern standard spoken and written Spanish have been excluded: ( 4 ) [−assumptive] [+assumptive] [−past] hablo ha hablado hablará habr á hablado [−non-assertive] [+past] [−VP ∈ non - past] hablaba había hablado hablaría habría hablado [+VP ∈ non - past] habló [−past] hable haya hablado ― ― [+non-assertive] [+past] hablara hubiera hablado ― ― [−anterior] [+anterior] [ − anterior] [+anterior] 7 In literary written French, on the other hand, the modal feature [assumptive] is replaced by the relative-tense feature [posterior], as shown in (5). Two forms in italics ― namely, [+nonassertive, +past, −posterior, ±anterior] ― are rarely used even in this register today: ( 5 ) [−posterior] [+posterior] [−past] il parle il a parlé il parlera il aura parlé [−non-assertive] [+past] [−VP ∈ non - past] il parlait [− VP ∈ non - past ] il avait parlé il parlerait il aurait parlé [+ VP ∈ non - past ] il parla [+ VP ∈ non - past ] il eut parlé [−past] il parle il ait parlé ― ― [+non-assertive] [+past] il parlât il e ût parlé ― ― [−anterior] [+anterior] [ − anterior] [+anterior] In modern spoken French, however, the system in (5) is drastically reduced to that in (6), due to the deletion of the highly marked two forms specified as [+VP∈non-past] and the neutralisation of the [±past] opposition in the subjunctive area marked as [+non-assertive]. Interestingly, in both Spanish and French, as in Portuguese, the most highly marked cell — in terms of feature values — remains unsaturated: ( 6 ) [−posterior] [+posterior] [−past] il parle il a parlé il parlera il aura parlé [−non-assertive] [+past] il parlait il avait parlé il parlerait il aurait parl é il parle il ait parlé ― ― [+non-assertive] [−anterior] [+anterior] [ − anterior] [+anterior] In contrast, Portuguese diverges fundamentally from the aforementioned tense-driven system in that its temporal anchoring operates in a different way. Here — as in Spanish and French — after the mood feature [±non-assertive] is marked to indicate whether the situation to be described belongs to the actual world or to a non-actual one, the speech time is taken as the primary anchoring point. Portuguese, however, instead of establishing reference points such as [±past] on the basis of temporal relations to the speech time, divides the time axis, with reference to that point, into two intervals: one extending backward and excluding the speech time ― the PAST ―and the other extending forward and including the speech time ― the NON - PAST . Simultaneously, the speaker's temporal viewpoint for describing a situation is located either within the NON - PAST — whose default value is the speech time — or within the PAST . From this located viewpoint, the situation to be described is then evaluated as either finished or not finished. In this system, what is crucial is whether the temporal viewpoint is located within 8 the non-past interval or within the past interval, and whether the situation is viewed as not finished or finished. This is because the location of the temporal viewpoint linguistically encodes the opposition between two distinct worlds — the NON - PAST and the PAST — in terms of the possibility or impossibility of human intervention in the world as such, and the distinction between whether the situation is confirmed as not finished or as finished likewise encodes this same type of possibility or impossibility. The temporal anchoring of a situation is secondarily derived from the position of the speaker's temporal viewpoint along the time axis and from whether the situation is viwed as finished or not from that viewpoint. Crucially, in this model, tense is not treated as a primitive category. Rather, it is analysed as a derived grammatical value determined by the interaction of these primitive anchoring-aspect features: [±VP∈past] and [±finished]. This perspective stands in contrast to traditional accounts, such as those of Reichenbach and Klein, which take tense — especially past, present, and future — as the foundational axis of temporal interpretation. This may be called a perspective-andaspect-driven temporal system. The differences between the temporal anchoring processes in these two temporal systems are summarised in (7): ( 7 ) Temporal anchoring process 1 st stage 2 nd stage 3 rd stage 4 th stage PT [±VP ∈ past] speaker's temporal viewpoint located [±finished] finishedness evaluated → situation time located [±assumptive] occurence possibility evaluated SP [±past] reference times defined [±assumptive]→[±posterior] sub-reference times defined [±anterior] situation time located [±VR ∈ non-past] viewpoint selected FR [±past] reference times defined [±posterior] sub-reference times defined [±anterior] situation time located [±VR ∈ non-past] viewpoint selected (PT, SP, and FR stand for Portuguese, Spanish, and French, respectively.) In Portuguese, the temporal anchoring of a situation is processed in two phases: first by applying the feature [±VP∈past], and then [±finished]. Once the situation has been localised in future time — through the combination of the feature value [−VP∈past, −finished] and contextual cues such as adverbial expressions — the feature [+assumptive] is subsequently used to express the speaker's evaluation of the likelihood of the future event's occurrence. As such, this feature does not serve a temporal anchoring function. By contrast, French and Spanish require three ― at most, four ― phases in the temporal anchoring of a situation: [±past] ⇒ [±posterior] ⇒ [±anterior] (⇒ [±VP∈non-past]), or alternatively, [±past] ⇒ [±assumptive] ⇒ 9 [±anterior] (⇒ [±VP∈non-past]). In both cases, the features [±posterior] or [±assumptive] are necessary for constructing a sub-reference time, which enables the localisation of the situation in future time. This fundamental opposition between these two systems helps clarify a number of differences observed between Portuguese and other major Romance languages, including Spanish and French. Why does the present perfect (pretérito perfeito composto) in Portuguese exhibit a temporal reference that is SUPERFICIALLY very different from that of the corresponding tense forms in the other languages? Why does the preterite use of the present perfect — widely recognised in French, and partly in Spanish — not exist in Portuguese? Why has the range of uses of the preterite expanded in Portuguese, while in French the simple past has long become obsolete in spoken language, with Spanish again occupying a transitional position? Why is it that the preterite (pretérito perfeito simples), when accompanied by the temporal adverbs já / ainda não, can be used to express future perfect meaning only in Portuguese? Why, in paralelle, has the system of future tenses in Portuguese undergone obvious functional erosion, while in French the temporal localisation of future situations predominantly relies on the future tenses, and Spanish appears to occupy an intermediate stage? Why does the choice of tense in counterfactual constructions diverge in Portuguese from the patterns found in the other Romance languages? Why is the opposition between the preterite and the imperfect maintained in the subjunctive in Portuguese, whereas it is neutralised in the subjunctive in Spanish? We shall account for such differences solely on the basis of (a) the inventory of primitive anchoring-aspect features ― PAAFs ― employed in each language; (b) the sequencing of those features within the anchoring process; and (c) the types of figurative extension applied to those features. 2. Primitive Semantic Oppositions among Finite Verb Forms in Portuguese The present indicative bears overt morphological marking only for person and number, lacking any overt markers of mood or temporal reference. This can be seen by comparing the firstperson plural form fala-0-mos of the verb falar in the present indicative with its counterparts in the imperfect indicative (falá-va-mos), future indicative (fala-re-mos), past future indicative (fala-ría-mos), present subjunctive (fal-e-mos), imperfect subjunctive (falá-sse-mos), and future subjunctive (fala-r-mos). While all of the latter forms bear overt markers of mood and temporal reference — namely, -va, -re, -ria, -e, -sse, and -r — the present indicative lacks such markers, 16 non-past interval — the default value of which is the speech time, serving as the reference point distinguishing the non-past interval from the past interval. (21) A Teresa é médica. "Teresa is a doctor." (22) A Teresa foi médica. "Teresa was a doctor." (23) A Teresa era médica quando a encontrei. "Teresa was a doctor when I met her." (24) A Teresa tinha sido médica quando a encontrei. "Teresa had been a doctor when I met her." In both (21) and (23), the state a Teresa ser médica is viewed as ongoing from the speaker's temporal viewpoint and is thus marked as [−finished]. In (21), however, that viewpoint is placed at a point within the NON - PAST interval — more specifically, at the speech time — and the finite verb é is therefore marked as [−VP∈past]. In contrast, in (23), the temporal viewpoint is located at a point within the past interval, namely the moment at which the event eu a encontrar occurred; the finite verb era is hence marked as [+VP∈past]. In both (22) and (24), on the other hand, the same state is viewed as no longer holding from the temporal viewpoint and is therefore marked as [+finished]. In (22), this viewpoint is placed at the speech time, and the finite verb foi is accordingly marked as [−VP∈past]. In (24), by contrast, the temporal viewpoint is located at the past point quando a encontrei, and the finite verb tinha sido is consequently marked as [+VP∈past]. On the basis of this observation, the present and preterite indicative forms are to be specified not only for [±finished] but also for [−VP∈past], as shown in (25). ( 25 ) [ − finish ed] [+finished] [−VP ∈ past] fala falou [+VP ∈ past] falava tinha falado Although previous research has touched on related issues, it has not explicitly identified that the entire assertive indicative paradigm of Portuguese — falo, falei, falava, tinha falado — can be exhaustively generated by the two binary features [±VP∈past] × [±finished]. This minimal 2×2 grid constitutes the foundational backbone of the system and explains why the present perfect — tem falado — could never integrate into this core paradigm (cf. Section 5.1). Historically speaking, until the rise of compound tenses, the Portuguese indicative fact-marking paradigm was exhaustively realised by the four simple forms falo, falei, falava, falara — precisely the 2×2 grid generated by [±VP∈past] × [±finished]. Due to sound changes, however, the thirdperson plural form falaram of the pluperfect indicative became homophonous with falarom of 17 the preterite indicative. To avoid this homophony, the former was replaced by the periphrastic form haviam/tinham falado. As this replacement extended to the entire paradigm of the pluperfect indicative, it is thought that falara was ultimately replaced by tinha falado. The replacement of falara by havia/tinha falado should thus not be understood as the result of a systemic reorganisation of the relations among terms within the system. Returning to the main point, the above feature specification — the preterite as [−VP∈past, +finished] and the imperfect as [+VP∈past, −finished] — directly accounts for the acceptability and naturalness of (26), the ungrammaticality of (27), and the markedness of (28). (26) A Teresa era médica quando a encontrei, e ainda o é / mas já não o é. "Teresa was a doctor when I met her, and she still is / but she no longer is." (27) *A Teresa foi médica, e ainda o é. "Teresa was a doctor (but no longer is), and she still is." (28) ?A Teresa foi médica, mas já não é. "Teresa was a doctor (but no longer is), but she no longer is." In (26), the imperfect era, marked as [+VP∈past, −finished], merely indicates that the state Teresa ser médica held at a certain point in the past — namely, the point referred to by quando a encontrei — without making any reference to what happened to that state afterward. Thus, it remains possible that the state ended before the speech time, but it is equally possible that it continued up to it. For this reason, e ainda o é "and she still is" and mas já não o é "but she no longer is" are both compatible continuations. In contrast, in (27) and (28), the preterite indicative foi, marked as [−VP∈past, +finished], means that the state Teresa ser médica is confirmed as already finished at the speech time. Therefore, e ainda o é "and she still is" cannot be conjoined without a logical contradiction. On the other hand, the continuation mas já não o é "but she no longer is" is logically equivalent to foi médica "she used to be a doctor," and is thus semantically redundant — a redundancy that accounts for the unnaturalness of (28). Earlier, we stated that the distinction [±finished] — that is, whether the finishedness of a situation is confirmed or not — is essentially significant for humans, since a situation whose finishedness is confirmed is one upon which no further intervention is possible, whereas a situation whose finishedness is not confirmed remains one open to intervention. Likewise, the distinction between the temporal intervals of the non-past and the past on which the temporal viewpoint for describing the situation is located — i.e., [±VP∈past] — is also fundamentally important, for the past represents a world in which one can no longer live and upon which no further intervention is possible, while the non-past represents a world in which one is currently 18 living or may still live, and which is thus open to potential intervention. In other words, the opposition between the preterite indicative and the present indicative may be understood as the linguistic encoding of this fundamental cognitive distinction between situations upon which intervention is no longer possible and those upon which it remains possible, as seen from a world in which such intervention is still viable. Returning to the the imperfect indicative and the pluperfect indicative, there are also cases in which they do not express [+VP∈past]. One such case is their use in the consequent clause of a counterfactual conditional, as in (29): (29) Se eu tivesse mais dinheiro, comprava / tinha comprado isto.5 "If I had more money, I would buy this." / "If I had had money, I would have bought this." For the speaker, both "the past world" and "the non-real world" are conceived as alternative worlds opposed to "the non-past world" and "the real world", to which the speaker and hearer actually belong, and which are open to human intervention. If the latter two are referred to more generically as "this world", then the former two constitute "another world", which can be accessed only through memory or imagination, and which is not open to human intervention. Thus, these two alternative worlds — the PAST and the NON - REAL — jointly constitute a higherorder category: another world or intervention impossibility, which is encoded morphologically by the imperfect and the pluperfect. In other words, the tense feature [+VP∈past] undergoes metaphorical extension6 through the schema7 of another world or intervention impossibility, which subsumes both the PAST and the NON - REAL , to yield [+VP∈non-real]. This figurative extension makes it possible to describe situations located in a non-real world from a cognitive viewpoint situated within that very world. Considered in this light, positing [+VP∈past] as the primitive value presents no theoretical difficulty. 5 In the Spanish equivalents Si yo tuviera más dinero, compraría esto and Si yo hubiera tenido más dinero, habría comprado esto, unlike in the Portuguese example (31), the tense of the subordinate clause varies depending on whether the counterfactual situation concerns the past or the non-past. As will be discussed later, this difference stems from a conceptual distinction between a perspective-and-aspectdriven temporal-anchoring system (Portuguese) and a tense-driven temporal-anchoring system (Spanish, French, and Italian). 6 Metaphor is the use of a linguistic form that conventionally refers to one entity or concept to refer to another, based on some perceived similarity between the two. For example, in the phrase the flower of the workplace, the word flower refers not to a botanical organ, but to a person, mediated by the shared feature of visually striking beauty (cf. Tsuji 2002, pp.16–17). 7 Schema is a commonality shared between two similar entities or concepts (cf. Tsuji 2002, p.124). 19 On the other hand, although the present indicative and the preterite indicative may be considered to stand in opposition to these forms in terms of a different temporal perspective — that is, as [−VP∈past] — there also exist cases in which it is difficult to interpret either of these two forms as bearing the feature [−VP∈past]. In the typical uses of the present indicative exemplified in (8)-(10) in section 2-1, the temporal viewpoint is located at the speech time, which falls within the non-past interval. However, in the use commonly referred to as the gnomic present, exemplified in (11), the possibility arises that no temporal perspective is present at all. Furthermore, in uses of the present indicative found in dictionary definitions — such as brilhante: que brilha "brilliant: that shines" — the temporal perspective, as with the gnomic present, either does not exist or, if it does, the viewpoint may be located at any point on the temporal axis. It is therefore not appropriate to interpret these uses as bearing the feature [−VP∈past]. The same applies to the preterite indicative. For instance, in definitions found in dictionaries — such as perdido: que se perdeu "lost: that was lost" — where the preterite is used, the temporal viewpoint may be located at any point on the temporal axis, and therefore can fall within either the past or the non-past interval. What this use of the preterite does, then, is indicate the finishedness of the situation as viewed from that arbitrary temporal viewpoint. On the basis of the above, it may be concluded that while the present indicative and the preterite indicative are, in their primitive senses, marked as [−VP∈past], it is appropriate to interpret them as bearing the value [0VP∈past] in special contexts or situations that require the arbitrary selection, ubiquity, or absence of a temporal perspective. Thus far, the analysis has proceeded without positing any tense features such as [±past], on the assumption that in this language, the temporal localisation of a situation is derivatively computed from the feature value combination of [±VP∈past] and [±finished]. For instance, in the case of the preterite indicative, which bears the features [−VP∈past] and [+finished] here, the situation as a whole is necessarily located in the past, i.e., prior to the speech time. On the other hand, in the case of the imperfect indicative, which bears the features [+VP∈past] and [−finished], at least the development of the situation up to the past point at which the temporal viewpoint is set is located in the past. It seems that this is a major characteristic of temporal localisation in Portuguese, and, as we shall see later, it constitutes a significant point of divergence from other Romance languages such as Spanish and French. Accordingly, in what follows, the analysis will likewise proceed without positing any independent tense features. 2.2 Assumption-Marking Tense Forms: Non-future Forms vs. Future Forms ― [±finished] × [±VP∈past] × [±assumptive] 20 In Section 2.1.2, we saw that the four basic tense forms used to present a situation as factual are exhaustively generated by the two binary features [±finished] × [±VP∈past]. In this section, we shall show that by superimposing the modal feature [±assumptive] onto this core system, the future tense forms are recursively derived. The present, preterite, imperfect, and pluperfect indicative forms, as seen in (30)–(33), all indicate that the situation described is taken by the speaker to be factual. (30) Ele agora está em Lisboa. "He is now in Lisbon." (31) A Ana ainda não chegou. Perdeu o autocarro. "Ana hasn't arrived yet. She missed the bus." (32) O Nuno tinha vinte anos quando o encontrei. "Nuno was twenty years old when I met him." (33) Quando elas chegaram ao porto, o barco já tinha saído. "When they arrived at the harbour, the boat had already left." By contrast, the future, future perfect, past future, and past future perfect indicative forms — as illustrated in (34)–(37) — typically express, particularly in formal written registers, a doxastic possibility on the part of the speaker (i.e., an assumption), while maintaining the same temporal reference as their factual counterparts (30)–(33). (34) Ele agora estará em Lisboa. "I believe he is in Lisbon now. (35) A Ana ainda não chegou. Terá perdido o autocarro. "Ana hasn't arrived yet. I believe she missed the bus." (36) O Nuno teria vinte anos quando o encontrei. "I believe Nuno was twenty years old when I met him." (37) Quando elas chegaram ao porto, o barco já teria saído. "When they arrived at the harbour, I believe the boat had already left." That is, while the present, preterite, imperfect, and pluperfect indicative forms — when used without modal expressions — typically present the situation referred to by the clause as an actual event or state, the future, future perfect, past future, and past future perfect indicative forms present the same situation, with the same temporal localisation, as being more likely the case, based on the speaker's estimation. Assumption is a type of modality whereby the situation referred to by the sentence is not asserted as an actual fact, but rather as something that is possibly the case. However, the kind of 21 possibility expressed by the future tense is not the same as that expressed by the modal verb poder in (38): (38) Ele agora pode estar em Lisboa. "He may be in Lisbon now." This is because the sentence in (38) can be paraphrased as Ele agora pode estar ou pode não estar em Lisboa "He may or may not be in Lisbon now," whereas (34) Ele agora estará em Lisboa "I believe he is in Lisbon now" cannot be paraphrased as *Ele agora estará ou não estará em Lisboa "*He will be or will not be in Lisbon now". This is due to the fact that, in (38), the speaker treats the possibility of the occurrence of ele agora estar em Lisboa and that of ele agora não estar em Lisboa as equally plausible, taking no stance on either. In contrast, in (34), the speaker places some degree of belief in the possibility of the occurrence of ele agora estar em Lisboa, while withholding belief from that of ele agora não estar em Lisboa. This asymmetry is further confirmed by the fact that (34) can be paraphrased as Creio que ele agora está em Lisboa "I believe he is in Lisbon now". In this way, assumption refers to the speaker's belief, to some degree, in the possibility of the occurrence of a given situation. In the present study, this type of possibility ― normally called doxastic possibility ―, as expressed by the future tense forms in the indicative mood, is posited as their primitive meaning and is represented as [+assumptive]. As we have seen, the future, future perfect, past future, and past future perfect indicative each establish an identical temporal anchoring to that of the present, preterite, imperfect, and pluperfect indicative, respectively. Therefore, apart from the opposition in [±assumptive], they share the same feature values for [VP∈past] and [finished]. The combination of the three primitive features established thus far, with each taking either a positive or negative value, both logically and actually yields eight finite verb forms. These are exemplified in (39), using the verb falar in its third person singular forms: ( 39 ) [ − finish ed] [+finished] [ − finish ed] [+finished] [−VP ∈ past] fala falou falará terá falado [+VP ∈ past] falava tinha falado falaria teria falado [−assumptive] [+assumptive] The feature [+assumptive], proposed above, also accounts for the use of the future indicative to express deontic modality, as exemplified in (40): (40) Não matarás. "You shall not kill." 22 In such cases, a metonymic extension8 — a type of figurative process — appears to occur: from "I believe you will not kill" (a belief serving as cause or justification) to "(therefore) you must not kill" (a command as expected effect or consequence). Here, a form that originally expresses a cause comes to stand for its result — a mechanism commonly attested in world languages. On the other hand, while the uses of the future tense forms as in (34)-(37) and (40) are generally referred to as modal, the following uses in (41) and (42) are commonly classified as temporal. (41) A Ana casará daqui a dois meses. "Ana will get married in two months." (42) O Jorge disse-nos que voltaria daí a uma semana. "Jorge told us that he would return in a week." However, when examples (41) and (42) are compared with their non-future indicative counterparts (43) and (44), it becomes clear that the temporal anchoring of the situation is, in fact, exactly the same. (43) A Ana casa daqui a dois meses. "Ana gets/is getting married in two months." (44) O Jorge disse-nos que voltava daí a uma semana. "Jorge told us that he was returning in a week." In both (41) and (43), casará and casa indicate a situation located in the future relative to the speech time — daqui a dois meses "in two months". Similarly, in both (42) and (43), voltaria and voltava refer to a situation located in the future relative to the PAST point at which the event o Jorge dizer-nos occurred — daí a uma semana "in a week". In Portuguese, therefore, the future-oriented reference ― i.e., temporal localisation of a situation in the future ― is not a function exclusive to the future tenses. As for the specification in (39), both the present indicative and the future indicative share the values [−VP∈past, −finished], opposed solely in [±assumptive]. Their common primitive function is thus, nowadays, to describe a situation whose finishedness cannot be confirmed from a temporal viewpoint located at some point within the non-past interval — typically, its default value, the speech time. Since a future situation, viewed from the speech time, has not even begun, its finishedness cannot be confirmed either. For this reason, such situations can be described not only by the future 8 Metonymy refers to the use of a form that refers to one entity or concept in order to refer to another, based on their contiguity in the external world or, more broadly, on their cognitive or conceptual association. For example, in the expression "to read Eça de Queirós", it is not the author himself but rather his works that are referred to by the name, based on the conceptual relationship between "author" and "work" (cf. Tsujii 2002, pp. 35–36). 23 indicative but also by the present indicative. In modern Portuguese, the widespread use of the present indicative to describe future situations — insofar as it satisfies the condition of [−assumptive]9 — is due precisely to its feature values [−VP∈past, −finished]. Crucially, this usage does not represent a semantic extension of the present tense, but rather constitutes its primitive function. In parallel, both the imperfect and the past future indicative share the values [+VP∈past, −finished], opposed only in [±assumptive]. Their common primitive function is therefore to describe a situation whose finishedness cannot be confirmed from a temporal viewpoint located at some point within the past interval. Since a past future situation, viewed from a given point in the past, has not even begun at that point, its finishedness cannot be confirmed either. This is why such situations can be described not only by the past future indicative but also by the imperfect indicative. The widespread use of the imperfect indicative to describe situations in the past future — as far as it meets the condition of [−assumptive]10 — is due precisely to its feature values [+VP∈past, −finished]. This usage does not represent a semantic extension of the past tense, but constitutes instead its primitive function. What is interesting in contrast is the use of the past future indicative in (45): (45) Ele começou como líder estudantil e, mais tarde, seria / *era primeiro ministro. "He started out as a student leader and would later become prime minister." In this case, unlike in (44), the use of the imperfect indicative is not possible. The fundamental difference between the past future situation expressed by seria in (45) and that expressed by voltava in the imperfect indicative in (44) lies in the fact that seria in (45) refers to a past future situation as a fact, whereas with voltava in (44), it remains unclear whether the past future situation is being presented as an actual fact. This is because (45) indicates that the state ele ser ministro actually came to hold later as a fact, whereas in (44), the event o Jorge voltar daí a uma semana may be no more than a verbal commitment on Jorge's part, and this does not pose 9 When the feature [−assumptive] cannot be maintained in the expression of a future situation, in place of the present indicative, the periphrastic construction ir in the present indicative + infinitive is generally used in contemporary spoken Portuguese. The essential function of this periphrastic form is to situate the event in future time. Consequently, it cannot situate the event in the present. By contrast, the future indicative, whitch is marked today as [−VP∈past, −finished, +assumptive], has as its essential function the assumptive indication of non-finished situations located in the present or the future. Since its function is not essentially temporal, it is not generally selected as the primary option in such contexts. 10 When the feature [−assumptive] cannot be maintained in the expression of a past future situation, in place of the imperfect indicative, the periphrastic construction ir in the imperfect indicative + infinitive is generally used in contemporary spoken Portuguese. In this register, the past future indicative is not generally selected as the primary option in such contexts, for the same reason as the future indicative. 24 any problem. Thus, the past future indicative in (45) serves not only to locate the situation in the future relative to a past point expressed by the first clause ele começou como líder estudantil, but also to explicitly mark its factuality. Both the past future indicative and the imperfect indicative share the modal feature [−non-assertive], whose original meaning is the indication of factuality — as will be explained in Section 2.4, the tense forms of the indicative mood contrast with those of the subjunctive mood in terms of the feature [±non-assertive]. The factuality of ele ser ministro in (45) can be indicated by the feature [−non-assertive], and what remains to be accounted for is the indication of futurity. This is achieved through a metonymic semantic extension whereby the feature [+assumptive], originally modal, is extended to express the temporal feature [+posterior] in the past future indicative. A situation located in the future relative to a given point in time cannot, by definition, be known as a fact at that point; it must necessarily be the object not of knowledge, but of inference — such as assumption or estimation. In other words, [+posterior] and [+assumptive] stand in a conditional–consequential relation: if a situation is located in the future (the condition), then it must necessarily be treated as an object of assumption (the consequence). It is on the basis of this relation that the feature [+assumptive] is considered to have undergone a metonymic semantic extension to [+posterior]. By contrast, the imperfect indicative, marked as [−assumptive], leaves no room for such a metonymic semantic extension. It therefore cannot indicate futurity in contexts such as (45). From the above discussion, it can be concluded that even in examples like (45), the original meaning of the past future indicative may be posited as [+assumptive]. By contrast, while the present, perfect, imperfect, and pluperfect indicative forms in (30)–(33) present situations as facts, there are cases in which interpreting them as [−assumptive] becomes problematic. For example, in the sentence Se calhar ele é brasileiro ("He's probably Brazilian"), the function of [+assumptive] is carried by se calhar. However, if the present indicative é were [−assumptive], the sentence would imply the contradictory proposition that ele ser brasileiro is both a fact and an assumption. On the other hand, if é is interpreted as [0assumptive], the sentence would simply mean that ele ser brasileiro is an assumption, which is not problematic. It is therefore reasonable to interpret the present, perfect, imperfect, and pluperfect indicative forms as being originally [−assumptive], but capable of taking a [0assumptive] value when accompanied by modal expressions such as adverbial phrases. 25 2.3 Non-Assertive Tense Forms: Indicative vs. Subjunctive ― [±finished] × [±VP∈past] × [±assumptive] × [±non-assertive] This section examines the semantic opposition between the indicative and subjunctive moods. We shall see that, by superimposing the mood feature [±non-assertive] onto the full indicative system — which, as shown in Section 2.2, is generated by [±finished] × [±VP∈past] × [±assumptive] — the subjunctive tense forms are recursively generated. (46) Ela hoje veio / *tenha vindo aqui. "She came here today." (47) Espero que ela hoje *veio / tenha vindo aqui. "I hope she has come here today." In (46), the preterite indicative veio is used, and the speaker is, by means of this verbal form, asserting the event ela hoje ter vindo aqui as a fact. In other words, the speaker makes a truthvalue judgment as to whether the situation described by the sentence is factual, and judges it to be true. Consequently, if it turns out that she has not actually come, the speaker would be regarded as having made a false statement. On the other hand, in (47), the occurrence of ela hoje ter vindo aqui represents merely the speaker's wish, and the speaker is not in a position to guarantee whether it is actually the case. That is, the speaker must suspend a truth-value judgment, and it is precisely for this reason that the preterite subjunctive tenha vindo is used. (46) and (47) are examples in which the indicative and the subjunctive cannot be interchanged. However, there are also cases in which both moods can be used, though with a clear difference in meaning. (48) A Maria quer casar com um espanhol que sabe falar português. "Maria wants to marry a Spanish man, who speaks Portuguese." (49) A Maria quer casar com um espanhol que saiba falar português. "Maria wants to marry a Spanish man who can speak Portuguese." In (48), um espanhol que sabe falar português refers to an actually existing individual. Since he exists, his ability — saber falar português — is taken as a fact, and the present indicative sabe is used in order to assert that factuality. In contrast, um espanhol que saiba falar português in (49) refers to someone whom Maria merely imagines as an ideal partner, and the speaker is unable to assert whether such a person actually exists. If the indicative sabe were used in this case, it would present such a person as actually existing, as in (48). The present subjunctive saiba is thus used in (49). As such, the indicative is the verbal form that explicitly expresses the speaker's judgment as to 32 In other words, in the past subjunctive, the aspectual opposition between the preterite and the imperfect indicative is neutralised, and must instead be recovered from context or situational cues. At the same time, the fact that both the preterite and the imperfect indicative correspond to the same past subjunctive form suggests that, apart from the opposition between indicative and subjunctive moods and the aspectual contrast between the preterite and the imperfect, these three forms are temporally equivalent in value. What this suggests is that the preterite indicative, the imperfect indicative, and the past subjunctive are all characterised by the same combination of absolute tense feature [+past] and relative tense feature [−anterior], which stands in contrast to the pluperfect indicative and the pluperfect subjunctive. Excluding counterfactual uses, all five forms — the preterite indicative, the imperfect indicative, the past subjunctive, the pluperfect indicative, and the pluperfect subjunctive — temporally locate situations with respect to a reference point situated prior to the speech time. This is the function of the absolute tense feature [+past]. On top of that, it is the pluperfect indicative and the pluperfect subjunctive that locate situations prior to the reference time, whereas it is the preterite indicative, the imperfect indicative, and the past subjunctive that locate situations simultaneous with, or temporally including, the reference time. This constitutes the opposition between the relative tense features [+anterior] and [−anterior]. The full set of correspondences can thus be accounted for in terms of indicative and subjunctive forms that share the values of the features [±past] and [±anterior]: [−past, −anterior]: present indicative / future indicative ⇔ present subjunctive [−past, +anterior]: present perfect indicative / future perfect indicative ⇔ present perfect subjunctive [+past, −anterior]: preterite indicative / imperfect indicative / conditional indicative ⇔ past subjunctive [+past, +anterior]: pluperfect indicative / conditional perfect indicative ⇔ pluperfect subjunctive That is, for example, since both the present indicative and the future indicative are [−past, −anterior], they correspond to the present subjunctive, which is likewise [−past, −anterior]. The opposition between the present and the future indicative — both of which share [−past, −anterior] — can be analysed either as an opposition in the relative tense feature [±future], or as an opposition in the modality feature [±assumptive], that is, [±doxastic possibility]. Since Spanish is not as strongly temporally driven as French with respect to its future tense, which analysis is more appropriate remains an open question. It is to be noted that "anteriority" does not necessarily imply "finishedness," but "finishedness" 33 (as the cause) necessarily entails "anteriority" (as the effect). Based on this latter entailment and causal relation, the feature [+anterior] can undergo a metonymic semantic extension to [+finished]. The fact that the present perfect, pluperfect, future perfect, and conditional perfect in the indicative, as well as the present perfect and pluperfect in the subjunctive, can express finishedness is accounted for by this metonymic extension. Why, then, are these forms not defined from the outset as [+finished]? The reason is that doing so would require, by opposition, the past subjunctive to be defined as [−finished]. However, the past subjunctive corresponds not only to the imperfect indicative, which has a [−finished] value, but also to the preterite indicative, which carries a [+finished] value. If the past subjunctive were defined as [−finished], the [+finished] value of the preterite indicative would become inexplicable. The opposition between the preterite and the imperfect indicative cannot be captured by the absolute tense feature [past] and the relative tense feature [anterior] alone, since both are characterised as [+past, −anterior]. Moreover, since [+anterior] was argued to undergo a metonymic extension to [+finished] in the present perfect, pluperfect, future perfect, conditional perfect, present perfect subjunctive, and pluperfect subjunctive, It follows that the preterite indicative — marked as [−anterior] — cannot be extended to encode [+finished]., either. This is where the notion of the speaker's temporal perspective comes into play. The imperfect indicative is a tense form in which the speaker's temporal viewpoint is placed on the reference time for event anchoring, i.e., [+past]. By contrast, the preterite indicative places the speaker's viewpoint not on the reference time [+past], but on [−past] — more precisely, on the default value of [−past], namely the speech time. The aspectual distinction between the "finished" aspect of the preterite and the "non-finished" aspect of the imperfect is derived from this difference in the location of the speaker's viewpoint. In the imperfect, the temporal viewpoint overlaps with the situation time, making it impossible for the speaker to confirm whether the situation has ended. In the preterite, on the other hand, the speaker's viewpoint is external to the situation time, so the ending of the situation is potentially observable. Thus, the preterite is assigned a [+finished] value not because it necessarily encodes factual completion, but because it stands in minimal semantic contrast with the imperfect, which encodes [−finished], i.e., non-verifiability of completion. In other words, aspect here is a derived value. It is important to note, however, that while the speaker's viewpoint in the preterite is external to the situation and thus makes the confirmation of completion possible, this only means that the speaker is in a position to confirm it — the situation may, in fact, remain incomplete. The ability of the preterite to express a finished aspect depends on its opposition with the imperfect: the latter, by virtue of the internality of the 34 speaker's viewpoint, must be interpreted as [−finished], and the former, as its minimal contrast, receives a [+finished] interpretation. To represent this oppositional structure, the feature [±VP∈non-past] is introduced. In contrast, the past subjunctive is unspecified for this feature, which is why it can be used in both preteritelike and imperfect-like contexts. Moreover, the contrast between the preterite and the imperfect is unique within the system: they are the only two forms that share the same values for [±nonassertive], [±past], and [±anterior]. Consequently, no other forms in the system are specified for [±VP∈non-past], since there is no formal contrast to warrant such specification. In systems where temporal location is determined on the basis of the relation to the speech time as a reference point, the distinction between simultaneity with that point ("present") and nonsimultaneity ("past" and "future") is a fundamental one. The reason is straightforward: the temporal relation to the reference point differs. At the same time, "present" and "future" stand in opposition to "past" as a unified category — that of "non-past." The reason for this opposition is equally clear: the past is characterised by the non-variability of the events that belong to it, whereas the non-past is characterised by the variability of its constituent events. The two categories are thus ontologically distinct. This constitutes the theoretical foundation for introducing the opposition between "past" and "non-past" into the Portuguese perspective feature [±VP∈past]. Moreover, in Portuguese, this opposition is formally manifested in the contrast between the viewpoints of the imperfect/pluperfect and those of the present/preterite. In this way, while "past" is equivalent to "non-present" and thus formally symmetrical with "future," it is not equivalent to "future" insofar as the latter is grouped together with "present" to form the category "non-past," which stands in opposition to "past." This compels us to posit that the structuring of time with reference to the speech time necessarily involves a fundamental opposition between "past" and "non-past (= present + future)." That is, even in the structuring of absolute tense, it may be more appropriate not to assume an immediate three-way opposition of "past / present / future" from the outset, but rather to posit two successive binary oppositions: "past" vs "non-past," and, within "non-past," "present" vs "future." Indeed, Japanese exhibits only the opposition between the "past" -ta and the "non-past" -ru. In English, too, it is the presence or absence of –ed that marks the opposition between "past" and "non-past," and the "non-past" category is further divided into "present" and "future" through the presence or absence of the modal auxiliary will. English also allows the internal structure of the "past" category to be further articulated by the use of modals such as would — the past form of will. In the case of Spanish as well, the decision to characterise absolute tense primarily in terms of the 35 feature [±past], rather than positing a three-way opposition including an independent "future" value, is based on the same line of reasoning. On the basis of this, the reason for defining future tense forms not in terms of a tense feature [+posterior], but rather through the modality feature [+assumptive] ― namely, [+doxastically possible] ― lies in the fact that the semantic feature that characterises them cannot be unambiguously classified as a "pure tense feature", unlike [±past] or [±anterior]. In Spanish, the function of the future tense is not simply to locate a situation in future time. Rather, it expresses the speaker's belief that a future situation is likely to occur — a representation of the speaker's cognitive belief system. It is precisely this function that allows the future tense, originally used for expressions of assumption, to extend naturally to expressions of intention or command. For this reason, it is difficult to treat what the Spanish future tense expresses as being purely a matter of the tense feature [+posterior]. Its core meaning lies in doxastic possibility — that is, in the representation of a belief-based prediction. (Doxastic possibility refers to a possibility that holds within the speaker's belief system; that is, it expresses the speaker's belief in the possibility of the occurrence of a situation. It corresponds not to epistemic "may", but to epistemic "will". To speak about future situations entirely in terms of "may" is ultimately meaningless, as "may" implies a 50/50 possibility — i.e., the speaker conveys nothing definite about the future. To speak about the future is, rather, to speak about what one believes is likely to happen.) From this viewpoint, the oppositions in Spanish between present indicative and future indicative, between preterite/imperfect and past future, and between pluperfect and past future perfect, can be seen — just as in Portuguese — as expressing [±doxastically possible] or [±assumptive], features interpretable both as temporal and modal. The same applies to the opposition between the indicative and subjunctive moods: this, too, expresses a modal contrast — specifically, [±suspension-of-truth-value-judgment/commitment], or, equivalently, [±nonassertive]. 5. Theoretical Applications On the basis of the structural differences between the perspective-and-aspect-driven temporal anchoring system and the tense-driven one, we shall account for some striking contrasts between Portuguese and other major Romance languages ― especially Spanish and French. 36 5-1. The Systemically Driven Semantic Shift of the Portuguese Present Perfect Whereas the present perfect (pretérito perfecto compuesto) in Spanish and the compound past (passé composé) in French might express the original present perfect value(?) — namely, the termination of a situation prior to the speech time and the continuation of the result state up to that moment (i.e., resultative or experiential readings) — the Portuguese present perfect (pretérito perfeito composto: ter in the present indicative + the unmarked form of the past participle — masculine singular), which shares the same historical origin, appears today to have a markedly different temporal value. In contemporary Portuguese, it denotes the initiation of a situation, its continuation or repetition, and its non-termination at the speech time. This shift is not arbitrary, but follows from the fact that Portuguese employs a perspective-and-aspect-driven temporal system, in contrast to the tense-driven systems of its Romance counterparts. The Portuguese present perfect originally took the form of the possessive verb haver followed by a noun phrase denoting the possessed object and a transitive past participle in agreement with that noun phrase in gender and number (ex He um livro lido). Semantically, it conveyed the meaning "to have something done" — analytically decomposable as: (i) the termination of an action or event prior to the speech time; (ii) the initiation and continuation up to the speech time of the result state that begins simultaneously with that termination; (iii) the notion of a possessed object; and (iv) a possessive relation. In Spanish, French, and Italian, which use the auxiliaries haber, avoir, and avere respectively, the semantic origin is the same across all these languages. However, due to syntactic restructuring — on the surface, the rightward movement of the noun phrase and the loss of gender and number agreement on the past participle — the semantic components of (iv) "possession" and (iii) "possessed object" were eliminated from the original construction. Structurally and semantically, this change involved the noun phrase shifting from being the complement of haver to the complement of the past participle. This shift entailed the loss of its interpretation as referring to a possessed object, as well as the loss of haver's function as a possessive verb. At the same time, the noun phrase's relocation to the complement position of the past participle gave rise to a verb phrase headed by the participle, thereby resulting in the disappearance of gender and number agreement. As the past participle came to appear immediately after haver, a new phrase structure emerged in which haver selects as its complement the entire verb phrase headed by the past participle. As a result, the newly reanalysed structure haver + past participle in the masculine singular form + noun phrase came to express the meaning of (i) the termination of a situation prior to the speech time, and (ii) the 37 initiation and continuation without interruption12 — up to the speech time — of the result state that begins simultaneously with that termination. Subsequently, "In the 15th century, ter begins to occur as an expression of unmarked possession, thus competing with haver as a lexical verb. Haver becomes marginal in the 16th century and is definitively obsolete from the 17th century onward.11 In parallel with its lexical development, ter begins to substitute haver in the perfect periphrasis." (cf. Olbertz 2018: 486). In Spanish, French, and Italian, on the other hand, a similar process led to the development of the pretérito perfecto compuesto, passé composé, and passato prossimo, respectively. However, since these languages are tense-driven temporal systems (to be confirmed with caution), such constructions came to be specified for the features [−past, +anterior] or [−past, +finished], and attained stability within the system as semantically distinct counterparts to the preterite, simple past, or remote past, which are marked as [+past, −anterior, +VP∈non-past]. These forms have retained their original meaning and continue to express it to this day. Portuguese operates, on the other hand, within a perspective-and-aspect-driven temporal system, and the situation is therefore completely different. As discussed in Section 2.1, in the domain of fact-stating forms marked as [−non-assertive, −assumptive], the only primitive anchoring-aspect features used for temporal anchoring are [±VP∈past] and [±finished]. The four possible combinations of these features — [−VP∈past, −finished], [−VP∈past, +finished], [+VP∈past, −finished], and [+VP∈past, +finished] — are exhaustively and exclusively realised by the present indicative, the preterite indicative, the imperfect indicative, and the simple pluperfect indicative, respectively. This results in a structurally closed system, leaving no room for additional forms. It is this structural closure that triggered the semantic shift of the present perfect indicative. Let us now examine this development in detail. As noted earlier, the original temporal value of the present perfect in the indicative mood was the combination of (i) the termination of a situation prior to the speech time and (ii) the initiation and continuation up to the speech time of a result state beginning simultaneously with that termination. Within this system, however, the preterite indicative, marked as [−VP∈past, +finished], had long functioned as the highly stable form responsible for expressing the termination of a situation prior to the speech time. Moreover, this tense was capable — depending on context and especially in combination with temporal adverbials such as já, nunca, 12 Since the termination point of the situation precedes the speech time, there exists a temporal interval between these two points. Moreover, because the termination of the situation coincides with the initiation of the result state, there is also an interval between the beginning of this state and the speech time — and this interval corresponds to the duration or continuation of the result state. 38 or ainda não — of expressing not only the termination of the situation itself, but also the initiation, continuation without interruption, and non-termination at the speech time of its result state. Within such a system, the newly formed construction ter in the present indicative + past participle had no room to compete with the preterite indicative, which was already capable of fulfilling the same temporal function. As a result, although the present perfect indicative originally bore the prototypical present perfect value — [−VP∈past, +finished, +resultative] — its temporal value underwent a shift. It came to be reinterpreted as a form marked by [−VP∈past, −finished, ✓ initiated, ✓ stative], and survived in the system in opposition to the preterite indicative, marked as [−VP∈past, +finished], and the present indicative, marked as [−VP∈past, −finished]. This process may be hypothesised as follows. As has already been stated on several occasions, the original temporal interpretation of the present perfect indicative comprised: (i) the termination of a situation prior to the speech time, and (ii) the initiation — simultaneous with (i) — of a result state characterised by (a) its beginning prior to the speech time, (b) its continuation up to that point, and (c) its non-termination at the speech time. In a perspectiveand-aspect-driven system, this configuration corresponds to the specification [−VP∈past, +finished, +resultative], where [+resultative] is a provisional feature encoding the onset, continuation, and ongoing nature of the result state relative to the speech time. Of the two components, (i), marked as [−VP∈past, +finished] — and more specifically [+finished] — entered into direct competition with the same value as that expressed by the preterite indicative, which already occupied a stable and dominant position within the [±VP∈past] × [±finished] system. Consequently, the present perfect came to abandon the expression of this value [+finished] in favour of the preterite indicative. The other component, (ii) the initiation — simultaneous with (i) — of a result state characterised by its beginning prior to the speech time, its continuation up to that point, and its non-termination at the speech time, represents the distinctive value of the present perfect, namely [+resultative]. However, as a temporal structure inherently presupposing (i), this component (ii) could not be used in isolation. As a result, through a process of synecdochic extension, the interpretation shifted to the more general meaning of the initiation — simultaneous with (i) — of a result state characterised by its beginning prior to the speech time, its continuation up to that point, and its non-termination at the speech time, and this reinterpretation ― namely, [+VP∈past, −finished, ✓ iniciated, ✓ stative] ― came to be applied to the description of a wide range of situations13. 13 In connection with this semantic shift, the past participle in the present perfect indicative also came to 39 The fact that the present perfect indicative still expresses the value [ ✓ stative] finds its origin in this historical development. Consequently, it is naturally compatible with predicates whose lexical aspect is a state, and straightforwardly presents them — depending, however, on temporal adverbials, as will be discussed later — as situations characterised by an initial point, temporal duration, and the absence of a terminal point at the speech time. By contrast, in the case of non-stative predicates such as processes, achievements, accomplishments, and semelfactives — that is, events in general — the interpretation involves first reanalysing them as iterative states: repeated events construed as states. They are then presented as [−VP∈past, −finished, ✓ initiated, ✓ stative]. Thus, even in the absence of temporal adverbials, the interpretation of present perfect forms as indicating iteration in the case of eventive predicates derives from the aspectual operations described above. However, a difficult problem arises in the case of stative predicates: (66) Esta manhã tem chovido muito. "This morning it has been raining a lot." (continuity) (67) Tem chovido muito. "It has been raining a lot repeatedly." (repetition) (68) Ultimamente tem chovido muito. "Lately it has been raining a lot repeatedly." (repetition) In (66), the rain that began to fall this morning has continued to fall and is still falling at the speech time, which is itself still located within the same morning. This continuity interpretation appears to derive directly from the feature combination [−VP∈past, −finished, ✓ initiated, ✓ stative]. In (67) and (68), however, this continuous reading is no longer available. Instead, the situations are interpreted as iterative states composed of an unspecified number of finished states. Where does this reading come from? We hypothesise that the iterative stative interpretation of eventive situations, which is coerced by the [ ✓ stative] feature inherent in the present perfect indicative, had been habitually performed; as a result, a temporal and aspectual cognitive frame — that of "the initiation and continuation of an iterative state from a certain point in the PAST up to the speech time" — is established and comes to be associated with the present perfect indicative, thereby inducing an iterative stative interpretation even for stative situations. denote not its original meanings of "termination" or "completion," but rather "partial realisation". The past participle's original function — to express the termination of a situation as viewed from a particular point in time — underwent a metonymic shift to the complete realisation of the situation. This, in turn, was metaphorically extended, via the schema of "realisation", to denote the partial realisation of the situation. 40 As for stative situations, the feature combination [−VP∈past, −finished, ✓ initiated, ✓ stative] gives rise to an interpretation in which a state is understood as having begun prior to the speech time, continued up to that point, and remained unfinished at the speech time. This specification, however, does not determine whether the state is interpreted as iterative or continuous. At this point, the temporal and aspectual cognitive frame described above intervenes — namely, the continuation of an iterative state from a certain point in the past up to the speech time. In (67), which lacks any temporal adverbial, the iterative reading derives from this frame. In (68), which includes a temporal frame compatible with it, the iterative reading likewise arises. In (66), by contrast, the temporal frame is incompatible with this cognitive frame; the frame is thus neutralised, and the continuous reading prevails. Additionally, the present perfect indicative is incompatible with temporal adverbials that explicitly quantify the number of iterations. This is because such quantification presupposes a time interval that is closed at both ends: counting events requires not only a defined starting point but also a defined endpoint. The present perfect, by its very nature, denotes a time interval that is open at the endpoint, thereby resulting in a semantic mismatch. Besides the commonly discussed uses mentioned above, the present perfect indicative also exhibits an interesting additional usage. This usage may be described as the temporal anchoring of a composite state, composed of two sequential states, by means of the present perfect. Araújo (2003: 232–233) characterises it as follows: "L'événement est terminé mais il est mis en relation avec un autre qui le suit. Dans ce cas le pretérito perfeito composto insiste sur la continuité ou la répétition de cet événement (à la différence du pretérito perfeito simples)". On p. 233 of the same work, Araújo presents the following two examples: (69) O doente tem estado muito agitado mas agora acalmou. "The patient has been very agitated, but he has now calmed down." (70) Têm tomado várias decisões importantes, mas atualmente evitam assumir novas responsa-bilidades. "They have made several important decisions, but at present they avoid taking on new responsibilities.") At first glance, the fact that the first situation can be viewed as finished from the speech time might seem to contradict the earlier characterisation, but in fact it does not. In (69), the state₁ o doente estar muito agitado, together with the subsequent state₂ o doente estar calmo, forms a single composite state₃ indicated by the feature [✓state]: state₃=⟨state₁∪state⟩. The initiation of this composite state₃ — which coincides with that of state₁ — is denoted by the feature 41 [✓initiated], and its non-termination at the speech time — which coincides with that of state₂ — is denoted by the feature combination [−VP∈past, −finished]. The transition from state₁ to state₂ within the composite state₃ is encoded lexically by acalmar. Analytically, the sequence can be represented as follows: initiation of state₁ ⇒ continuation of state₁ ⇒ transition from state₁ to state₂ ⇒ non-termination of state₂ at the speech time. The present perfect's feature combination [−VP∈past, −finished, ✓initiated, ✓state], applied to the whole composite situation₃, is morphologically realised on the first predicate. What is crucial here is that the combination of [−finished] with [−VP∈past] indicates that the situation as a whole is not finished at the speech time. It does not mean that state₁ itself remains unfinished — [−finished] does not denote the termination status of the state₁ referred to by the predicate estar muito agitado on which it is realised. Even if state₁ has come to an end — that is, even though a transition to state₂ has occurred — the feature combination still clearly marks the composite state₃ as ongoing at the speech time. In (70), too, the repeated action tomar decisões construed as the state₁ and the subsequent absence of the action assumir novas responsabilidades ― that is, the subsequent state₂ ― together constitute a single composite state₃. Analytically, the sequence of states may be represented as: initiation of the state₁ tomar decisões ⇒ continuation of state₁ ⇒ transition from state₁ to state₂ não assumir novas responsabilidades ⇒ non-termination of state₂ at the speech time.Accordingly, the situation is still being described in terms of [−VP∈past, −finished, ✓iniciated, ✓state]. There are other cases that are subject to complex analysis. In contexts such as live news reporting, it may appear that the present perfect indicative is incompatible with an achievement predicate like registar um aumento de casos nos últimos dias, as in (71): (71) Temos registado um aumento de casos nos últimos dias. "We have registered an increase in cases in recent days." However, the present perfect indicative as a form functions here by reinterpreting the event registar as an iterative state — one that was initiated prior to the speech time and continues up to it. The phrase um aumento de casos nos últimos dias, on the other hand, serves to focus on one portion of that ongoing state — that is, it isolates one segment of the whole. Still, since "a part of a state" is itself a state, the focused portion remains compatible with the stative value of the present perfect. As has been said in Section 2.5, when the temporal viewpoint is not located exclusively on the speech time, the present perfect indicative may have the same temporal value as the preterite indicative — namely, [+VP∈past, +finished] — as exemplified in (58). This type of viewpoint 48 (79) *Cuando llegues, él yá murió. "*When you arrive, he already died." (80) *Cuando llegues, él yá ha muerto. "*When you arrive, he has already died." (81) Cuando llegues, él (ya) habrá muerto. "When you arrive, he will (already) have died." French behaves in a similar manner: both (83) and (84) are ungrammatical. (85) represents the corresponding form in the future perfect indicative. (82) *Quand tu arriveras, il mourut déjà. "*When you arrive, he already died." (83) *Quand tu arriveras, il est déjà mort. "*When you arrive, he has already died." (84) Quand tu arriveras, il sera (déjà) mort. "When you arrive, he will (already) have died." In their tense-driven temporal anchoring systems, the Spanish preterite indicative and the French simple past are marked, respectively, as [+past, −assumptive, −anterior, −VP∈past] and [+past, −posterior, −anterior, −VP∈past]. These forms describe past situations [+past] as finished [+finished] relative to the speech time ― the latter value being derived from [−VP∈past] ―and therefore cannot be used to refer to future situations. Similarly, the Spanish present perfect and the French compound past are marked as [−past, −assumptive, +anterior] and [−past, −posterior, +anterior], respectively. These forms describe situations as finished [+finished], relative to the speech time [−past], with [+finished] derived metonymically in this case from [+anterior], and likewise cannot be used to refer to future situations. It should be noted, however, that in Spanish, if the feature [assumptive] becomes morphologically or functionally eroded, and if the temporal interval marked by [−past] extends rightward (i.e., into the near future), then it becomes theoretically possible for the present perfect to take on a future perfect reading. Before that, however, it is also theoretically possible for a construction involving the verb ir in the present indicative + a + perfect infinitive to be established for the expression of a future perfect reading — as is already the case in Brazilian Portuguese, where ir in the present indicative followed by a perfect infinitive functions as a future perfect. For example: Quando você chegar, ele já vai ter morrido ("When you arrive, he will already have died"). 5.5 Cross-Linguistic Variation in Tense Selection in Counterfactual Constructions The opposition between a perspective-and-aspect-driven temporal anchoring system and a tense-driven one also manifests itself in the temporal marking of counterfactual constructions. 49 Consider the factual statement "Since I did not have money, I did not buy that car" and its corresponding counterfactual: "If I had had money, I would have bought it". (85) Como eu não tinha dinheiro, não comprei aquele carro. (factual) (86) Se eu tivesse dinheiro, tinha comprado aquele carro. (counterfactual) (87) ?Se eu tivesse tido dinheiro, tinha comprado aquele carro. (counterfactual) In Portuguese, the counterfactual corresponding to the factual statement in (85) is given in (86), while (87) is typically judged as unnatural. The contrast lies in the choice of verb form in the subordinate clause: in this case, the simple form tivesse is preferred in parallel with the simple form tinha in the factual statement, whereas the compound form tivesse tido is often perceived as pragmatically infelicitous. In this language, temporal anchoring proceeds in the order [±VP∈past] ⇒ [±finished] (cf. (7) in Section 1), and anchoring cannot be effected unless both of these features are activated. In counterfactual constructions, however, a metaphorical extension from [+VP∈past] to [+VP∈non-real] occurs via the schema another world or intervention impossibility (cf. (30) in Section 2.1.2). As a result, it becomes impossible to distinguish whether the temporal viewpoint within the NON - REAL world is situated in the PAST or in the NON - PAST . Moreover, even if one were to attempt to rely on the figurative extension of other features, such an approach would prove unworkable, since none of those features possesses the internal structure of "locating a viewpoint somewhere" in the way that [±VP∈past] does. Accordingly, whether the viewpoint is situated in the past or in the non-past must be inferred from context or pragmatic cues. By contrast, the presence or absence of termination can still be explicitly encoded by the feature [±finished]. On this basis, temporal anchoring in counterfactual constructions proceeds in the sequence shown in (88): (88) [+VP∈non-real] (←[+VP∈past]) ⇒ [±VP∈past] (determined by context or discourse cues) ⇒ [±finished] Returning to examples (85)–(87), the form tinha in the factual sentence (85) is specified as [−non-assertive, +VP∈past, −finished,]. In order to convert this into a hypothetical statement ―that is, one involving the suspension of truth-value judgment ― within the NON - REAL domain, two operations are required, as shown in (89): first, the mood feature [−non-assertive] of the indicative must be changed to [+non-assertive] to yield the subjunctive, and second, a metaphorical extension from [+VP∈past] to [+VP∈non-real] must be applied. (89) [−non-assertive, +VP∈past, −finished,] → [+non-assertive, +VP∈non-real, −finished] 50 When the verb ter bears this set of feature values, its surface realisation is the simple form tivesse as found in the counterfactual construction (86). In this case, the placement of the temporal viewpoint in the PAST — that is, the value [+VP∈past] — must be inferred from context or situational cues. By contrast, the compound form tivesse tido in (87) is marked as [+non-assertive, +VP∈non-real, +finished], and it is likely that speakers find it infelicitous because the value [+finished] constitutes a shift from the [−finished] value of tinha in the factual sentence (85). By contrast, in Spanish, the counterfactual sentence corresponding to the factual one in (90) is (92), whereas (91) is judged to be ungrammatical. In this case — in contrast to Portuguese — the verb form that corresponds to the simple form tenía in the factual sentence (90) is the compound form hubiera tenido in (92), rather than the simple form tuviera in (91). What does this imply? (90) Como yo no tenía dinero, no compré aquel coche. (factual) (91) *Si yo tuviera dinero, habría comprado aquel coche. (counterfactual) (92) Si yo hubiera tenido dinero, habría comprado aquel coche. (counterfactual) In this language, temporal anchoring proceeds in the sequence [±past] ⇒ [±posterior] (← [±assumptive]) ⇒ [±anterior] (cf. (7) in Section 1), and anchoring cannot begin unless the absolute tense feature [±past] is first activated. In Spanish, however, a metaphorical extension from [+past] to [+non-real] occurs in counterfactual constructions via the schema another world or intervention impossibility. As a result, within the NON - REAL world thus construed, it is no longer possible to use the feature [±past] to indicate whether the situation precedes the speech time ― i.e., [+past] ― or not ― i.e., [−past]. At the same time, since [±past] constitutes the first stage of temporal anchoring, it must still be realised morphologically. To satisfy this requirement, a metaphorical extension arises from [±posterior] to [±past] via the schema of abstract anteriority — that is, anteriority to some temporal point, whether to the speech time ([+past]) or to a reference time marked as [±past, ±posterior]. This figurative extension makes it possible for temporal anchoring to proceed. Accordingly, temporal anchoring in Spanish counterfactual constructions proceeds in the sequence shown in (93): (93) [+non-real] (← [+past]) ⇒ [±past] (← [±posterior]) Returning to examples (90)–(92), the form tenía in the factual sentence (90) is specified as [−non-assertive, +past, −assumptive, −anterior, −VP∈non-past]. To convert this into the 51 expression of a hypothetical situation situated in the NON - REAL domain, it must be mapped onto a form that can bear the feature values [+non-assertive, +non-real, +past, −anterior] in counterfactual constructions ― in the domain marked as [+non-assertive], the oppositions [±assumptive] and [±VP∈non-past] are neutralised. The only form that satisfies these conditions is hubiera tenido in (92), marked as [+non-assertive, +past, +anterior]. This is because, as shown in (93), this form allows for the feature extension in (94): (94) hubiera tenido [+non-assertive, +past, +anterior] → [+non-assertive, +non-real, +past,] The right-hand side feature set [+non-real, +past] arises through a semantic extension from the left-hand side specification [+past, +anterior]. In this extension, [anterior] is metaphorically extended to [past], which makes it impossible to preserve the original [−anterior] value associated with tenía; the interpretation is thus left to context and discourse. By contrast, in the case of tuviera in (91), marked as [+past, −anterior, +non-assertive], a different extension occurs, as shown in (95): (95) tuviera [+non-assertive,+past, −anterior,] → [+non-assertive, +non-real, −past] This form is consequently ruled out, since it cannot encode the [+past] value that was present in tenía. These differences in tense selection in counterfactual constructions between Portuguese and Spanish arise entirely from the structural opposition between the two temporal anchoring systems. The Spanish-type treatment also holds in other languages that employ a tense-driven temporal anchoring system, such as French, Italian, and English, as shown in (96)-(97). (96) Comme je n'avais pas d'argent, je n'ai pas acheté cette voiture-là. (factual) (97) *Si j'avais de l'argent, j'aurais acheté cette voiture-là. (counterfactual) (98) Si j'avais eu de l'argent, j'aurais acheté cette voiture-là. (counterfactual) 5.6 The Maintenance of the Preterite–Imperfect opposition in the Subjunctive in Portuguese and its neutralisation in Spanish In contemporary standard spoken and written Spanish, the speaker's temporal perspective — that is, the feature [±VP∈non-past] — is morphologically encoded only in the opposition between the preterite indicative and the imperfect indicative, as shown by the two shaded cells in (99) = (4). This feature is no longer encoded by the pluperfect indicative (e.g. había hablado), which once formed a morphological contrast with the now nearly obsolete preterite anterior (e.g. 52 hubo hablado). ( 99 ) [−assumptive] [+assumptive] [−past] hablo ha hablado hablará habr á hablado [−non-assertive] [+past] [−VP ∈ non - past] hablaba había hablado hablaría habría hablado [+VP ∈ non - past] habló [−past] hable haya hablado ― ― [+non-assertive] [+past] hablara hubiera hablado ― ― [−anterior] [+anterior] [ − anterior] [+anterior] Apart from these two forms, there exists no pair of finite verb forms that are identical in all values of the features [±non-assertive], [±past], [±assumptive], and [±anterior]. Accordingly, the speaker's temporal perspective is not morphologically encoded elsewhere in the system; its interpretation is context-situation-dependent — or possibly determined by some default setting (a possibility that merits further investigation). Although both the imperfect indicative and the preterite indicative are marked as [−non-assertive, +past, −assumptive, −anterior], they differ in terms of the speaker's viewpoint. In the imperfect indicative, the viewpoint is located at the reference time, which is specified as [+past, −assumptive] — effectively [+past] — and since the situation time is [−anterior], the speaker places their viewpoint within the domain in which the past situation held, and describes it from within that temporal space. By contrast, in the preterite indicative, the speaker's viewpoint is located differently. Given the structure of reference time in Spanish, it can only be placed at the moment of speech, which is marked as [−past]. The speaker thus describes the past situation from the present — that is, from a viewpoint external to the temporal domain in which the situation occurred. In contemporary standard Spanish, then, the preterite indicative is the only tense that explicitly marks the speaker's temporal perspective as located at the speech time — that is, in the present — while situating the described situation in the past. As such, it possesses a fundamentally distinct feature structure from other tense forms used to describe past situations. This separation between the speaker's perspective and the situation time appears to be the key to explaining the preterite's implication of not only finishedness but also perfectivity, verifiability, and factual presentation — that is, objectivity. The fact that the French simple past, which shares the same feature configuration, disappeared early from spoken usage may be seen as the logical consequence of a 'cold', objective mode of past temporalisation. As already explained in Section 4, the opposition of perspectives [±VP∈non-past] is defined only within the domain of the indicative forms marked as [−non-assertive, +past, −assumptive, 53 −anterior]; it is neutralised in the domain of the subjunctive forms marked as [+non-assertive, +past, −assumptive, −anterior]. As a result, in the subjunctive mood, the preterite with [+VP∈non-past] and the imperfect with [−VP∈non-past] do not contrast. The reason for this neutralisation is straightforward: as previously stated, the perspective feature [VP∈±past] is a one that becomes active only at the fourth stage of the temporal anchoring process described in (100) = (7). ( 100 ) Temporal anchoring process 1 st stage 2 nd stage 3 rd stage 4 th stage PT [±VP ∈ past] speaker's temporal viewpoint located [±finished] finishedness evaluated → situation time located [±assumptive] occurence possibility evaluated SP [±past] reference times defined [±assumptive]→[±posterior] sub-reference times defined [±anterior] situation time located [±VR ∈ non-past] viewpoint selected FR [±past] reference times defined [±posterior] sub-reference times defined [±anterior] situation time located [±VR ∈ non-past] viewpoint selected (PT, SP, and FR stand for Portuguese, Spanish, and French, respectively.) Moreover, in contemporary usage, within the indicative mood marked as [−non-assertive], the only contrast generated by this feature inside the domain [−non-assertive, +past, −assumptive, −anterior] is the one between the preterite and the imperfect. A highly marked feature such as this, which entails a high maintenance cost, cannot operate within the domain of the subjunctive — already marked as [+non-assertive] — in comparison to the indicative. By contrast, Portuguese operates with a perspective-and-aspect-driven temporal anchoring system, in which, as shown in (100), temporal anchoring occurs only after the process [VP∈±past] ⇒ [±finished] has been completed. Accordingly, unless both the viewpoint feature [VP∈±past] and the aspectual feature [±finished] are activated, the temporal anchoring of the situation cannot be achieved. For this reason, even in the subjunctive mood — marked as [+nonassertive] — the expression of both [VP∈±past] and [±finished] is obligatory, just as it is in the indicative mood marked as [−non-assertive]. This obligatory expression is formally manifested in the opposition between the preterite subjunctive [+non-assertive, −VP∈past, +finished] (e.g. tenha falado) and the imperfect subjunctive [+non-assertive, +VP∈past, +finished] (eg. falasse). 54 6. Conclusion This study has accounted for various cross-linguistic differences in temporal anchoring among Portuguese and other Romance languages — namely, Spanish and French — by means of three analytical components: ・set of primitive anchoring-aspect features, ・the ordered sequence in which these features apply, and ・the metaphorical semantic extensions to which they are subject. Temporal anchoring process 1 st stage 2 nd stage 3 rd stage 4 th stage PT [±VP ∈ past] speaker's temporal viewpoint located [±finished] finishedness evaluated → situation time located [±assumptive] occurence possibility evaluated SP [±past] reference times defined [±assumptive]→[±posterior] sub-reference times defined [±anterior] situation time located [±VR ∈ non-past] viewpoint selected FR [±past] reference times defined [±posterior] sub-reference times defined [±anterior] situation time located [±VR ∈ non-past] viewpoint selected PT, SP, and FR stand for Portuguese, Spanish, and French, respectively. These differences include the decline of future tense forms in Portuguese, the aoristic shift of the present perfect in Spanish, the elimination of the simple past in French, and the divergent tense selection patterns in counterfactual constructions across these languages. Among these phenomena, particular emphasis has been placed on the semantic reanalysis of the Portuguese present perfect indicative, which represents a distinctive contribution of this study. This study has also shown that, depending on the minimal set of features employed for temporal anchoring, at least two distinct systems of temporal anchoring exist within the Romance languages: a tense-driven system (as found in Spanish and French), and a perspective-and-aspect-driven system (as found in Portuguese). In the former, temporal anchoring is achieved by constructing a complex reference time with respect to the speech time and locating the situation to be described relative to that reference point, following a sequence such as: [±past] ⇒ [±posterior] ⇒ [±anterior] (⇒[±VP∈non−past]). 55 In the latter, temporal anchoring is achieved by dividing the time axis into two intervals with respect to the speech time and determining whether the situation to be described is viewed as finished or not finished from a viewpoint situated within either interval: [±VP∈past] ⇒ [±finished]. The decline of future tense forms in Portuguese may be attributed to this system: since temporal anchoring terminates at the level of [±VP∈past]⇒[±finished], the morphological future tenses do not serve a necessary anchoring function and have therefore undergone functional obsolescence. to be continued… 56 Reference (incomplete) Alarcos Llorach, E. (1980). Estudios de gramática funcional del español (3ª ed.). Madrid: Editorial Gredos. Balla-Johnson, W. R. (2022). The Portuguese pluperfect: Development and the effect of auxiliary ter generalization. Journal of Portuguese Linguistics, 21, 1–40. Bybee, J., Perkins, R., & Pagliuca, W. (1994). The evolution of grammar: Tense, aspect, and modality in the languages of the world. Chicago: The University of Chicago Press. Bybee, J., & Fleischman, S. (Eds.). (1995). Modality in grammar and discourse (Typological Studies in Language). Amsterdam: John Benjamins Publishing Company. Câmara Jr., J. M. (1972). The Portuguese language. Chicago: University of Chicago Press. Campos, M. H. C. (1997). Tempo, aspecto e modalidade: Estudos de linguística portuguesa (Colecção Linguística, Vol. 6). Porto: Porto Editora. Carreira, M. H. A. (2003). Pratique de portugais de A à Z. Paris: Hatier. Carreira, M. H. A. (2004). Modalités et verbe en portugais. Revue Belge de Philologie et d'Histoire, 82, 691–702. Comrie, B. (1981). Aspect: An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press. Comrie, B. (1985). Tense. Cambridge: Cambridge University Press. Fleischman, S. (1988). The future in thought and language: Diachronic evidence from Romance (Cambridge Studies in Linguistics). Cambridge: Cambridge University Press. Klein, W. (1994). Time in language. London: Routledge. Mateus, M. H. M., Brito, A. M., Duarte, I. S., & Faria, I. H. (2003). Gramática da língua portuguesa (5ª ed., rev. e aumentada). Lisboa: Editorial Caminho. Moens, M., & Steedman, M. (1988). Temporal ontology and temporal reference. Computational Linguistics, 14, 15–28. Olbertz, H. (2018). The perfect in (Brazilian) Portuguese: A Functional Discourse Grammar view. Open Linguistics, 4(1), 478–508. Oliveira, F. (1985). O futuro em português: alguns aspectos temporais e/ou modais. In Actas do 57 1.º Encontro da Associação Portuguesa de Linguística (pp. 353–357). Lisboa: APL. Oliveira, F. (2000). Some issues about the Portuguese modals dever and poder. Belgian Journal of Linguistics, 14, 115–129. Oliveira, F., & Leal, A. (2012). Sobre a iteração do Pretérito Perfeito Composto em Português Europeu. Linguística – Revista de Estudos Linguísticos da Universidade do Porto, 7, 65–88. Oliveira, F., & Leal, A. (2018). Sobre a natureza homogénea do Pretérito Perfeito Composto em Português Europeu. Linguística – Revista de Estudos Linguísticos da Universidade do Porto, 13, 57–78. Palmer, F. R. (2001). Mood and modality (2nd ed.). Cambridge: Cambridge University Press. Perini, M. A. (2002). Modern Portuguese: A reference grammar. New Haven, CT: Yale University Press. Reichenbach, H. (1947). Elements of symbolic logic. New York: The Free Press.SANTOS, Santo, Diana. (2003). Português para estrangeiros: Gramática básica para alunos que já falem e escrevam a nossa língua. Retrieved from: http://www.linguateca.pt/Diana/download/portugisisk.html Sten, H. (1973). L'emploi des temps en portugais moderne. København: Munksgaard. Teyssier, P. (1985). Manuel de langue portugaise: Portugal-Brésil (2e éd., revue et corrigée). Paris: Klincksieck. Vendler, Z. (1967). Linguistics in philosophy. Ithaca, NY: Cornell University Press.