An “epoch in Historical Linguistics and Indo-European scholarship”: in memoriam Romano Lazzeroni (1930–2020)

1 Verb system and argument coding in the ancient Indo-European languages: at the morphosyntax/semantics interface

The complex system of relationships between the functional dimensions and both the morphological and the syntactic dimensions of language was at the center of Romano Lazzeroni’s interests for more than 60 years. In one of his first papers (Lazzeroni 1956), that dates back to 1956, 11 years earlier than the seminal work by Zeno Vendler on verb classes (Vendler 1967) and 20 years earlier than the well-known book on aspect by Bernard Comrie (Comrie 1976), he investigated the relationship between verbal aspect and negation in ancient Greek, by clearly appreciating the distinction between the point of view on the event and the semantic properties of the verb lexeme – something that the theoretical tools available at that time made much more subtle than today. His qualitative and quantitative analyses of the totality of the texts by Herodotus, Sophocles, Aristophanes, Plato, Aeschines, Demosthenes, Isocrates and Thucydides led to the conclusion that in Ancient Greek, like other languages, negation neutralizes the telic/atelic distinction and specifically interacts with the perfective vs. imperfective viewpoint, but Ancient Greek is distinctive in that this condition is represented not only at the lexical but also at the morphological level, in relation to the use of certain grammatical categories to encode aspectual notions. The relationship between negation and telos of the denoted event has been largely investigated, in more recent years, from different perspectives (cf. Hopper and Thompson 1980, among many others), in the organization, change and acquisition of various languages, from Slavic (Lindstedt 1995; Slabakova 2001: 92–94; Willis 2013) to Romance (Manzini and Savoia 2011: 80–158; Romagno 2015), from Ancient Greek (Romagno 2002, 2005: 27–38) to Chinese (Song 2018).

In his 1956 study, preceded only by a short note by Vendryes (by which “il semble qu’il y aurait contradiction à nier ce qui est en même temps affirmé comme accompli” [it seems that there would be a contradiction in denying what is at the same time affirmed as accomplished]: Vendryes 1946: 11) and a few sporadic comments by Brunel (1939: 81, 98, 139, in which there is no clear distinction between the notions of concreteness, perfectivity and telicity), Romano Lazzeroni was the first to remark on the need for analyzing the alternation between affirmed and negated event not only in relation to distinctions depending on verb semantics, but also in relation to the morphological categories that represent complementary strategies to encode those distinctions. In Lazzeroni’s work, the ability to combine different perspectives and methods constantly goes together with special attention to the systemic nature of language: this distinctive sign of his breakthrough research is the result of an original combination of deep and extensive knowledge, freedom of thought and rigour in action.

The 1956 paper on aspect and negation in ancient Greek, together with the related study published in 1957 (Lazzeroni 1957), opened up a new era in Indo-European Linguistics, by indicating a novel direction, followed by a growing number of scholars, both in Italy and abroad, and still recognized as indispensable to achieving one of the most fundamental goals of Historical Linguistics and, especially, Indo-European Linguistics: addressing historical questions guided by contemporary theoretical advances and experimental achievements. Along this course and with special attention to verb categories and argument coding, Romano Lazzeroni reconstructed a complex and coherent system of correspondences between the ancient Indo-European languages, in which the rationale for each element is established by itself and, at the same time, is necessary to account for the other elements. This system is mainly – but not exclusively – based on Greek and Indo-Iranian (in particular, Vedic and Avestan) data: this relies on a principle of homogeneity of Greek and Indo-Iranian verb and argument coding systems, that does not apply, to the same degree, to other ancient Indo-European languages.^[1]

In a long series of studies^[2], inaugurated by the 1977 masterpiece “Fra glottogonia e storia: ingiuntivo, aumento e lingua poetica indoeuropea” – which reveals, even in its title, Romano’s sharp, insightful and unconventional trait – he has shown that in an archaic phase of the Greek and Indo-Iranian verb system (or, rather, in the Indo-European tradition passed on into the Greek and Indo-Iranian verb system), the relationship between verb lexeme and tense does not have a grammatical expression (cf. Kuryłowicz 1932, 1964; Stang 1932; Thurneysen 1885). The system, instead, is based on the distinction between state and process representation. The morphological encoding of tense is realized through the gradual grammaticalization of two deictic particles: -i, which marked proximity (e.g., Sanskrit bharati , as opposed to the older form bharat) and, in a part of the Indo-European world that includes Greek, Indo-Iranian and Armenian, *e- (> Indo-Iranian a-, the so-called “augment”), which probably was a marker of distal deixis (e.g., Sanskrit a-bharat coexisted – and gradually replaced – bharat, as opposed to bharati ). The acquisition of the morphological expression of tense gave rise to a complex series of changes that obscured the original situation. However, traces of that situation are present in the older stages of the ancient Indo-European languages and did not escape Romano Lazzeroni’s genius: as many of his studies have demonstrated, language change always entails a gradual process in which old and new elements often coexist (Lazzeroni 2015a: 8).

Romano Lazzeroni’s account of the Indo-European verb system allowed him to solve a large number of enigmas in Indo-European studies, in a unitary model. The first enigma links the three elements, apparently unrelated, that constitute the title of the 1977 masterpiece (see also Lazzeroni 1980, 1982, 1984a, 1984b, 2017b). The term “injunctive” was coined by Brugmann (1880) to refer to those unaugmented verb forms with primitive endings (incongruously called “secondary”) that are more stably attested in Vedic and Avestan than in Homer (Hoffmann 1967). As regards its meaning components, the injunctive shows a plurality of functions, including both modal and non-modal values, that have long appeared incompatible: in fact, it takes past tense endings, but may have present meaning; it does not take the augment, but may have past meaning; it does not have modal markers, but it may have modal values. By rigorously placing the diachronic perspective at the center of his analysis, Romano Lazzeroni reconstructed the sequence of events that accounts for the comparative data, both formally and functionally:

1. the active injunctive encoded the representation of “tenseless” (above time) processes;

2. when the type *bhereti (Skr. bharati ) was formed, with actual present meaning, the injunctive *bheret (Skr. bharat) became the expression of non-actuality and, consequently, of both the indefinite or “general” present (thus maintaining the original value of the old indicative) and the past; moreover, it took modal values, as opposed to the new indicative, marked with -i. Vedic and Avestan (in which the augment is almost exclusively absent) show traces of this phase;

3. the injunctive, as an expression of the past, was then marked with the augment: the augmented forms differed from the corresponding unaugmented forms, as they encoded past meaning, as opposed to timeless and modal meanings. This phase largely corresponds to the situation attested in the R̥gveda;

4. the unaugmented injunctive, as an expression of the indefinite or “general” present, disappeared under the pressure of the new present forms, marked with -i. The present/past distinction included two functional units, one of which had two formal variants: e.g., bharati (present) vs. a-bharat/bharat (past). This phase corresponds to the Homeric and – with a few differences – Atharvavedic situations;

5. the type bharat (without augment) disappeared and, consequently, the functional distinction became one between the present bharati and the past a-bharat. This phase corresponds to the Classical Sanskrit, Old Persian and Classical Greek data.

In conclusion, the results of Romano Lazzeroni’s research clearly show that the free use of the augment in Homer cannot be attributed to the Indo-European poetic language (on which see Wackernagel 1926: 211–222), but is a consequence of the survival of the injunctive in the conservative tradition of poetry (Lazzeroni 1977: 30). The Mycenaean tablets – whose content is far from poetic – show that the augment may also be absent in prose.

Many other elements define the solidity of Romano’s system, in which “tout se tient” [everything holds together], across languages, and fundamental pieces of the Indo-European tradition transferred in the historical languages are revealed and incontrovertibly explained.

1. He identified the rationale, diachronically based on both formal and functional grounds, for both the distribution of the different present classes and their relationship with the aorist stem: in languages such as Greek and Indo-Iranian which distinguish between present/imperfect and aorist stem, the polymorphism is a consequence of the acquisition of a morphological expression of tense (Lazzeroni 2011c: 49). To mention only some of Romano’s insightful works on this topic, I refer to his papers on root formations (Lazzeroni 1980, 1983a, 1985b, 1993a, 1993b, 2008), the role of *-ye-/-yo- (Indo-Iranian -ya-) in the expression of the anticausative (Lazzeroni 2002a, 2004, 2017a), the strategies to encode causative and transitive (Lazzeroni 2009, 2017c; see also 2011b), the solution to Kuiper’s enigma (Lazzeroni 2008).

2. He reversed the perspective on the supposed archaism of the Hittite verbal system, by clarifying the sequence of events that culminated in the formation of the monothematic paradigm, in which present and past have the same stem (e.g., present: etmi ‘[I] eat’ < *ed-mi, past: etun): the lack of distinction between imperfect and aorist in Hittite does not originate from a Pre-Indo-European or Indo-Hittite phase, but is the consequence of the relatively recent acquisition of the morphological expression of tense (Lazzeroni 2011c: 51). Basically, every root injunctive produced a present and there were no residual injunctives to form the aorist. In this scenario, the mystery of the Hittite -hi conjugation (on which see Di Giovine 1996; Jasanoff 2003; Rose 2006, among others) can be unraveled: Romano Lazzeroni showed how the -hi formations are related to the so-called “resultative” perfect, that is, the more recent perfect that encodes the state of the object, unlike the old perfect, that expressed the representation of the subject’s state (Lazzeroni 2011c).

3. He clarified the foundations for a full understanding of the development of the Indo-European middle: it originally was a derivational category (in Joan Bybee’s sense: Bybee 1985), but progressively became an inflectional one and encoded new values: “in un sistema fondato sull’opposizione dei tempi erano contenute le premesse perché le due diatesi, unite dal tratto della temporalità, diventassero, tendenzialmente, simmetriche” [a system based on the opposition of tenses contained the grounds for the two diatheses, united by the temporality trait, to become basically symmetrical] (Romano Lazzeroni 1990: 22).

4. He provided fundamental contributions to the understanding of the relationship between middle and perfect (Lazzeroni 1990, 2014; cf. Neu 1968, on Hittite), which, in Romano’s model, constitute the two building blocks of the Indo-European verbal system. The injunctive encoded the representation of process, whereas the old perfect encoded the representation of state (Lazzeroni 1977, 1990, 2014). Both perfect and middle – whose relationship is also manifested on formal grounds: see the two archaic series of endings (cf. Kuryłowicz 1932; Stang 1932; Watkins 1969) – expressed the notion of the subject’s state; however, their mutually exclusive distribution, attested in Vedic (Di Giovine 1990–1996) and Greek (Romagno 2005), can be accounted for in relation to the grammatical expression of different noetic contents (Lazzeroni 1990): the middle, prototypically unaccusative, may include dynamic semantic components; the perfect, instead, prototypically stative, excludes any dynamic semantic components and represents the state as a property or condition of the subject (Romagno 2005, 2014, 2021). The perfect, in fact, remained longer unaffected by the grammaticalization of tense.

5. He provided the definitive answer to the long-debated question of the third diathesis (cf. Jasanoff 1978; Oettinger 1976), which has been recognized not only in Indo-Iranian (Kümmel 1996), but also in the Hittite “type -a middle” (Watkins 1969: 84): Romano Lazzeroni has shown that the so-called “stative” endings do not belong to a third diathesis distinct from the middle, but correspond to the original endings of the middle, that pertained at a phase antecedent to the integration of the middle into the tense system and the consequent remodeling of its endings on the basis of the active present endings. The so-called “stative” endings, in fact, correspond to the functional prototype of the category, which resisted the innovation longer (Lazzeroni 1990, 1993b, 2014, 2015a).

These are only some of the enigmas of the Indo-European verb system that have been solved through the exceptional intelligence of Romano Lazzeroni. We could discuss many other fundamental achievements in this field for which we are thankful to Romano’s research. I would add only a further remark on the results of a growing number of independent studies on different Indo-European languages, that are robustly consistent with Lazzeroni’s model of the Indo-European verbal system. To mention only some of these important contributions, refer to Stephanie Jamison’s (1983) and Leonid Kulikov’s (2012a, 2012b) researches on Sanskrit, and to Helena Kurzová’s researches on Greek, Latin and other ancient Indo-European languages (1993, 1999, 2011, 2014).

Romano Lazzeroni’s ground-breaking research also provided fundamental contributions to the understanding of argument coding strategies in the ancient Indo-European languages: in particular, he opened up new avenues to understanding the alternation between canonical and non-canonical marking of core arguments (cf. Aikhenvald et al. 2001) in relation to semantic roles, syntactic functions and the properties of the referent of the subject/object nouns. His studies, that combined a rigorous analysis of Indo-European data with theoretical and typological investigations, shed new light on a series of apparently unrelated phenomena that can be explained on the basis of a “universal” (both typological and cognitive) principle (cf. Dixon 1994): languages do not show typological “pureness”; the phenomena that do not follow the dominant type can be accounted for in relation to the unbalance between conceptual-semantic representations and systemic constraints (cf. Cennamo 2001; Romagno 2011; among others). Romano Lazzeroni’s studies (see, in particular, Lazzeroni 1995a, 2002b, 2002c, 2013a, 2015b, 2017d) showed that also in the ancient Indo-European languages, the argument coding system – whose principles have long been debated (cf. Cuzzolin 1998; Dardano 2013; Drinka 1999; Pooth et al. 2019; Rumsey 1987; Uhlenbeck 1901; Villar 1984) – is governed by mechanisms of interaction (and, sometimes, conflict) between semantic roles, syntactic functions and referential properties (in particular, animacy and individuation). I will mention here only his brilliant contributions on the metaplasm of gender, the origin of the thematic neuter (the so-called “weak” neuter), and the non-canonical marking of subject (see, in particular, Lazzeroni 2002b, 2002c). Romano’s in-depth investigations into Sanskrit, Greek, Iranian, Hittite, Latin and Romance varieties have shown that:

1. the selection of gender (animate vs. inanimate) is related to the semantic role of the argument that has the subject or object function, respectively;

2. the accusative subject represents a semantically based marking, independently of gender selection: the semantically non-canonical subject (that is, the undergoer subject) takes the canonical marker of the object; the accusative, in fact, is the typical case of the object, whose referent is prototypically inanimate and unagentive, but not of the subject, whose referent is prototypically animate and agentive;

3. only thematic neuter nouns frequently have an allomorph of animate gender that belongs to the same inflectional class. The presence of variants with animate gender corresponding to the athematic (the so-called “strong”) neuter nouns, instead, always entails metaplasm of inflectional class (e.g., Greek ὄναρ/ὄνειρος, Sanskrit svar/sūraḥ, etc.). It is worth remarking that athematic neuter nouns represent the oldest members of the neuter category;

4. the paradigm of weak neuter nouns – unlike the strong neuter – differs from the paradigm of weak (i.e., thematic) animate nouns only in the nominative (in Sanskrit, in fact, the weak neuter nouns have an animate vocative);

5. in conclusion, weak neuter nouns correspond to a group of old thematic animate nouns that became neuter when their accusative was generalized as a marker of the (semantically non-canonical) subject: “il nominativo neutro tematico è omofono dell’accusativo tematico animato perché è l’accusativo tematico animato” [the neuter thematic nominative is homophonous with the animate thematic accusative, because it is the animate thematic accusative] (Lazzeroni 2002b: 317).

2 Cultural reconstruction between linguistics and philology

Romano Lazzeroni’s studies on the Indo-European verb and argument coding systems benefitted from his love for Vedic. He is one of the most eminent scholars in Old Indic Linguistics and, in particular, in Vedic and Sanskrit Linguistics. As remarked in the citation for the so-called “Gonda Prize”^[3] awarded to Romano in 1999, his contributions show an extraordinary ability to interpret the Vedic texts – which appear, as is well known, impenetrable in many respects – and to reconstruct their cultural background, by combining rigorous philological analyses with both theoretical investigations and typological observations. The studies collected in the volume La cultura indoeuropea (Lazzeroni 1998) – which has also been translated in Portuguese (Lazzeroni 1999b) – together with many other works by Romano Lazzeroni, testify to the impact of his scholarship on Oriental Studies. I will mention only two of those masterpieces: his study of the organization of living and non-living entities and his study of the relationship between nectar and ambrosia and the Indo-European representation of death.

In the first study, Romano Lazzeroni reconstructed a coherent system of conceptual representations in which movement is the distinctive feature of life and feet are the symbol of movement. He showed how the correspondence between Vedic taxonomy and Iguvine taxonomy that is manifested in the long-debated formula “bipeds and quadrupeds” can be accounted for only in a conceptual-semantic space based on the representation of the living world as “something that moves, that goes” (jágat-, Pre-Vedic term!), as opposed to the non-living (inanimate) world, which is represented as “something that stays” (sthā-). Both in Vedic and in Classical Sanskrit, only human beings, but not birds, are defined as “bipeds”: birds, in fact, use wings, but not feet, to move. On the contrary, the exclusion of birds from the biped category does not belong to the Greek and Roman cultures. However, the comparison between different ancient Indo-European languages reveals that the formula “bipeds and quadrupeds” belongs to the Indo-European tradition. Romano Lazzeroni, via his brilliant and rigorous analysis of the Vedic texts, reconstructed a complex cosmogonic system in which the principles underlying the representation and organization of living and non-living entities are clearly explained, and the rationale for the above-mentioned formula is provided. Therefore, it became clear that the whole conceptual-semantic system that includes this formula belongs to the Indo-European tradition. Relics of this system, which in the Vedic texts is coherently structured, survived in various areas: adapted (in Iranian), reinterpreted (in Greek), fossilized (in Umbrian). In conclusion, Romano Lazzeroni demonstrated that the formula “bipeds and quadrupeds”, which is attested also in the Iguvine Tablets, as a part of a prayer (in the dative plural: dupursus peturpursus), is not a relic of the Indo-European poetic language, but the expression of a socio-culturally determined interpretation of the world that was transferred into poetry.

Romano’s investigation into the Indo-European representation of death unraveled the mystery of the two types of food of the gods that guaranteed eternal life, as they were complementary antidotes to death. In the Homeric poems, nectar (νέκταρ) and ambrosia (ἀμβροσίη) occur together and do not differ either in function or in the way they are taken. By combining a linguistic and a philological perspective, Romano Lazzeroni, once more, provided an insightful interpretation of the historical memory transferred into the texts, and reconstructed a complex system of correspondences between different ancient Indo-European languages (in particular, Sanskrit, Greek and Latin), that is based on common noetic matrices. The etymology of Greek νέκταρ reveals the descriptive value of something that “crosses, passes over” (and, therefore, wins) “death” (νεκ-, cf. Latin nex, Greek νέκυς, plus*tr̥/*ter(ə)), but does not provide a reason for this representation. Romano has shown that the reason can be identified by reconsidering Greek data in relation to data coming from the Vedic texts and the Latin tradition. The Vedic world distinguishes natural death from accidental (premature) death: the former is a consequence of age, whereas the latter is caused by any other factors (in the Atharvaveda, the non-natural types of death are a hundred: cf. II, 28, 1). Only premature death represents an evil to be warded off and overcome; victory over premature death is represented as an action of “going across” it (tarati). At the other side of the Indo-European world, in Archaic and Classical Latin, nex is the marked term to refer to death, as it denotes premature death, as opposed to mors, which denotes the natural death or the generic notion of death. Therefore, in the Greek word νέκταρ, the lexeme that in Latin denotes premature death is combined with the verb base *ter(ə), which in Vedic refers to victory over this type of death. It is, then, clear that the conceptual-semantic representation underlying this expression is Indo-European. The correspondence remains to be explained between beating death and “crossing” death. Romano Lazzeroni’s acute analysis of Vedic texts provided a brilliant explanation for this correspondence as well. In Vedic culture, bad things, including death, which is the worst evil, are represented as narrow things (aṁhas-); winning over death (and other evils) is represented as going across narrows. Furthermore, the Greek term for ambrosia includes the term for natural death: ἀμβροσίη < *n̥mr̥t-. It is, then, clear why the two types of food of the gods, nectar and ambrosia, occur together: one is the antidote to premature death, while the other is the antidote to natural death. In conclusion, the Greek names for the food of the gods reveal the binary Indo-European notion of death.

In these and other studies for which we are thankful to Romano Lazzeroni’s genius, the typology of the Indo-European sign established by Walter Belardi (Belardi 1985, 1990, 1993) became a fruitful tool for linguistic and philological interpretation and cultural reconstruction.

The mention of the “Gonda Prize” gives me the opportunity to recall the attitude that Romano had towards prizes and awards. He received a huge number of distinguished honours, prizes and awards, and was Founder, Member, President and Honorary Member of many prestigious national and international scientific societies, including – to mention only some – the Accademia dei Lincei, the Italian Society for Historical Linguistics (Società Italiana di Glottologia), the Italian Association for Sanskrit Studies, the Cambridge Academy of Europe, the International Institute for Etruscan and Italic Studies, the International Institute for Advanced Asian Studies. Romano considered any honour or award as a token of responsibility towards his students and colleagues and of commitment to scientific research: he destroyed any traces of the citation for the “Gonda Prize” and discreetly donated the money to the Linguistics Branch of the Pisa University Library, which still has one of the most important European collections of Oriental Studies and, in particular, of Indian studies. Romano’s spirit was incompatible with any kinds of self-celebration: when the Gold Medal for Meritorious Scientists in Education, Culture and Art was awarded to him by the Presidency of the Italian Republic, he neither participated in the award ceremony nor shared the information with anyone; many of his colleagues and friends only read about it in the national newspapers.

3 Conceptual representations and linguistic categories in synchrony and diachrony

Romano’s works display special attention to the linguistic organization of conceptual and empirical data and, consequently, to the relationship between graded noetic categories and discrete formal categories. He was the first who introduced the prototype theory and the family resemblance model (Berlin and Kay 1969; Rosch and Mervis 1975; Wittgenstein 1953) into Indo-European Linguistics (cf. Lazzeroni 1990, 1995b), thus revolutionizing – once more – the perspective on old and unsettled questions and identifying new problems and novel solutions.

One of the central and long-lasting debates about the ancient Indo-European languages (dating back to Pāṇini’s work) is the diversity of usage types and functions of the middle category. The aporia whereby middle markers appear to be associated with a variety of meanings has been solved by Romano Lazzeroni (Lazzeroni 1990), who proposed to change the perspective on voice (and, in particular, on middle voice) and consider it as a “natural category” (in Eleanor Rosch’s terms: Rosch 1973, 1978; cf. Taylor 1989), in which it is possible to identify a prototype, corresponding to the most central members, and various peripheral members, linked to the prototype via a similarity relationship. This idea not only solved some systemic aporias that appeared to be irreducible (such as the lack of diathetic distinctions in the imperative: the agentivity required by the imperative is incompatible with the middle prototype; the middle imperative is the product of a process that made the two diatheses symmetrical), but also opened up new avenues of research on the Indo-European voice system, that continue to provide the most robust and consistent results and to indicate the most promising developments.

The use of the prototype and family resemblance model in investigations into the grammatical gender of the ancient Indo-European languages (see, in particular, Lazzeroni 1992a, 1993a) allowed Romano to clarify fundamental aspects of a category that has been defined as “one of the still unsolved puzzles of linguistic science” (István 1959: 1). His 1993 suggestion is still crucially valid: “molto del genere grammaticale resta enigmatico. Ma se rifletteremo sui principi costitutivi delle categorie linguistiche e sulle relazioni cognitive espresse dalle categorie grammaticali, forse il genere grammaticale ci apparirà meno fantastico della classificazione degli animali nell’Emporio celeste di Borges” [much of grammatical gender remains enigmatic. But if we reflect on the constitutive principles of linguistic categories and on the cognitive relationships expressed by grammatical categories, perhaps grammatical gender will appear less fantastic than the classification of animals in Borges’ Celestial Emporium] (Lazzeroni 1993a: 16).

We could discuss many other brilliant studies that Romano Lazzeroni conducted on the relationship between conceptual categories and linguistic categorization. I will mention only two further works on ancient Indo-European languages: his work on the vexata quaestio of the agent nouns in *-tér and -tor (Lazzeroni 1992b) and his work on action nouns (Lazzeroni 1997, 2012a). In the former, Romano explained the alternation between the two series of derivatives in relation to the different degrees of animacy (Silverstein 1976) and individuation (Timberlake 1977) of their referent. In the second, that comprises two studies (Lazzeroni 1997, 2012a), he showed that: 1. the differences in morphological and syntactic behavior between action nouns and agent nouns can be explained within a gradient of nouniness vs. verbiness; 2. the degree of prototypicality of each of the two categories depends on the distinct parameters defining the gradient, but interacts with the structural constraints of the system (see, in particular, Lazzeroni 2012a). The results of these studies – like many if not all of Romano’s studies – have important implications that go largely beyond the specific cases under investigation and can be related to two fundamental principles that are both theoretical and methodological: 1. the hierarchies underlying the linguistic encoding of noetic categories should be interpreted as representations that are not only multifactorial but also multidimensional; 2. the classification of a given category on the basis of certain features or dimensions depends on the functional principles governing the system in which that category is encoded, either morphologically or syntactically.

The definition of prototype includes both the notion of frequency (and the related notion of cognitive salience) and the notion of markedness (and the related notion of extension: “the unmarked term […] may have or may acquire greater reference potential (extension) than its marked opposite” [Andersen 2001b: 48]). The crucial role of markedness in language (and beyond) is strictly related to the notion of local markedness: that is, the degree of markedness of a given element can be measured (only) in relation to a given category or dimension or context. Local markedness determines so-called “markedness agreement” (in Henning Andersen’s terms: Andersen 1980, 2001a, 2001b), which – as Romano Lazzeroni showed in many studies – critically operates in language organization and change. When innovations correspond to the creation of a category, the linguistic change first affects the less marked, more prototypical elements, and progressively reaches the more marked, less prototypical ones; the opposite pattern characterizes innovations that can be interpreted as category merging or loss (Andersen 1990, 2001a, 2001b; Timberlake 1977). Romano Lazzeroni’s research on many ancient and modern Indo-European languages showed how this principle governs a large number of different phenomena: e.g., the Indo-European grammaticalization of tense, that started from the indicative, as it is less marked in relation to tense distinctions, compared to the subjunctive and optative, that encode the non-factual and counterfactual modality, respectively (cf. Lazzeroni 1977, 1984a, 1984b, 2016b, 2018a, 2018b, 2019b); the progressive loss of the locative from Latin to Romance (cf. Lazzeroni 2005); the syncretism of the genitive and ablative in ancient Greek and Sanskrit (cf. Lazzeroni 2005); the merging of the auxiliaries in the compound tenses of various ancient and modern Indo-European languages (cf. Lazzeroni 2013b); the progressive loss of the dual in ancient Greek (cf. Lazzeroni 1960), which reveals how a universal cognitive principle underlies a typological “universal” (see Greenberg 1963); to mention only some of the categorization processes, on which Romano shed new light by clearly explaining their functional architecture in both synchrony and diachrony.

The important contribution of Andersen’s markedness theory to the understanding of the principles and patterns of language change has been repeatedly noted by Romano Lazzeroni, who called Henning Andersen “one of the greatest linguists of the last few centuries”. The identical expression was used by Henning Andersen to refer to Romano Lazzeroni: the reciprocal esteem between these two giants of Linguistics dates back many years, even if they met in person only recently, during the 22nd International Conference on Historical Linguistics, held in Naples in 2015.

4 Principles and patterns of language change

Language change is the common superordinate theme connecting all the lines of research discussed so far. To the best of our knowledge, Romano Lazzeroni’s studies represent the most significant contribution to the elaboration of a holistic theory of language change, which is one of the main goals in language sciences.

As regards the hendiadic – and certainly not dichotomous – relationship between language organization and change (which Romano, with his inimitable style, used to remark the importance of from the first lecture of any Historical Linguistics class for undergrads, and on which Eugenio Coseriu wrote definitive pages: Coseriu 1952, 1958, 1966, 1994), it will be enough here to note that André Martinet, one of the godfathers of structuralism, chose a diachronic point of view for the summa of his theoretical investigations (Économie des changements phonétiques: Martinet 1955): the dynamic perspective is probably the most appropriate to clarify the principles that govern language systems, as innovations yield critical phases in which systems have to reorganize.

Romano Lazzeroni constantly emphasized the need for an explanation of language change that includes both the formal and the functional dimensions of language in a unitary and coherent model, and showed that to achieve this goal it is necessary to look at languages as both natural and cultural products (Lazzeroni 1987b, 2005): understanding the principles and mechanisms of language organization, change, acquisition and loss requires us to disentangle socio-historically determined from “universal”, neurobiologically determined factors.

Within a tridimensional representation of language change (on which see Belardi 1978; Labov 2001; Trudgill 1983; Weinreich et al. 1968, among others), Romano Lazzeroni’s researches on Indo-Iranian, Greek, Aegean-Anatolian, Italic, Latin and Romance (Lazzeroni 1967, 1968, 1969, 1972, 1984b, 1991a, 1999a, 2006, 2007, 2012b) showed how variation and change (including those that are due to contact: Lazzeroni 1983b; cf. Mancini 2008, 2013) involve not only time and space, but also the sociocultural stratification of speech communities, and demonstrated that, if it is true that the observation of ongoing changes in modern languages is especially fruitful, it is also true that, if investigations into past languages exploit the perspectives and methods refined by studying present languages, linguists can reach the other side of Labov’s purpose, that is, the understanding of the present by explaining the past: “capire il presente spiegando il passato” (Lazzeroni 1991b: 217).

Lazzeroni’s model stands at the intersection of three planes, with an increasing degree of abstractedness and a decreasing degree of variability (see, in particular, Lazzeroni 2005, 2015c, 2016b, 2018a, 2018b, 2019b, 2019c, 2020):

1. the plane of the relationship between linguistic phenomena and historical accidents: the actualization of language systems is historically determined and, therefore, language systems depend on sociocultural variables;

2. the plane of the relationship between conceptual representations and linguistic categories: languages are cognitive tools that convey an interpreted reality and, therefore, testify to the way in which speakers represent and organize the world in a given chronotope;

3. the plane of the relationship between language systems and the neurocognitive mechanisms underlying their functional organization: languages are the most typical product of the human brain (the “miracle” of language is the distinctive feature of humans: Maffei 2018) and, therefore, depend on the principles that govern brain activity.

Within this model, Romano Lazzeroni’s research, which benefitted from his penetrating overall view and his extraordinary ability to rapidly recognize the pattern of relationships between different phenomena, identified the constants (the “universals”) that define the principles of language change (the same – as he showed – that underlie language acquisition: Lazzeroni 2005) and allow him to explain and foresee (but never predict!) its pathways: as Romano used to remark during his lectures, the patterns of language change can be foreseen, but are never predictable, because they also depend on the numerosity and variability of historical accidents. Lazzeroni’s “universals” can be summarized in the following:

1. the tension between discrete forms and noetic continua: human memory limitations require to encode into discrete units a variety of meanings that are organized in gradient (“fuzzy”) categories, whose boundaries are blurred and overlapping. The studies on the processes of category creation, merging and loss mentioned above provided a clear explanation of some possible patterns of language change;

2. the tension between declarative memory and procedural memory: the former supports the storage of data and information, whereas the latter supports the processing of connections and rules that entail the automatic production and comprehension of linguistic units. A principle of economy of the memory system underlies the functional organization and change of languages. Speakers, in fact, use strategies that operate at a superordinate level relative to the single units, and exploit procedural memory in preference to declarative memory (cf. Lazzeroni 1992a, 2000b, 2015c, 2016b, 2019b, 2020).

Romano Lazzeroni’s research showed how the principle of economy of the memory system can account for a large number of (apparently unrelated) linguistic changes that occur in different Indo-European and non-Indo-European languages, and provided a definitive solution to crucial questions in Historical Linguistics, from the so-called “symmetrical” changes (first reported but not explained by Hermann Paul: Paul 1880; see also Giannelli 1973, among many others) to Sapirian drift (Sapir 1921), from the evolutionary interpretation of language change (cf. Croft 2000) to the long-debated issue of the extension of the scope of application of a given rule (cf. Kiparsky 1968). Romano has clearly explained the internal structure of the “symmetric” patterns of change: the innovation first affects a single member of a given category and progressively rises to the superordinate category level, thus involving all the category members. This allows the speakers to memorize only one “new” rule for all the constituents of the category involved in the change.

In Vedic, for instance, when both Proto-Indo-European *e and *o became a (by following a symmetrical pattern: the change should have first affected the anterior vowel, as shown by the Aryan loan words in Finnic), the -ă inflectional class (< PIE *-o) was associated with the masculine gender, and the -ā class (< PIE *-ā) with the feminine. In this phase, the quantity of the vowel was a gender marker only for -a nouns. Throughout the history of the Sanskrit language, the grammatical gender of all vowel classes of nouns has been progressively redistributed on the basis of the quantity of the vowel. In the first phase, if a stem ended in -ă, the noun was masculine; whereas if a stem ended in -ā, the noun was feminine; in the final phase, attested in Middle Indic, all the nouns in short vowels were masculine, whereas all the nouns in long vowels were feminine (Lazzeroni 1992a).

In this and many other changes, of which Romano Lazzeroni explained the pattern and identified the underlying principles, the change in the scope of application of the rule corresponds to the transfer of the rule to a more abstract level, in order to favor the automatic production (and comprehension) of the correct forms: that is, the speakers do not need to memorize the new forms one by one, but only the rule that generates them; in this way, they are able to produce all the correct forms only by applying the rule to a class of units (e.g., the vowel nouns).

Furthermore, as Romano’s innovative studies have demonstrated, the principle of economy of the memory system underlies the numerous cases of changes of the “paradigm structure conditions” (in Wolfgang Wurzel’s terms: Wurzel 1984, 1989; cf. Carstairs 1987). In Latin, when -os became -us (lupos > lupus), there were conditions in place for the merging of the second and the fourth inflectional classes. At a first glance, one could think that a complete merger occurred. However, Romano Lazzeroni uncovered a completely different process (Lazzeroni 2000b). The fourth class survived in Romance, in a certain number of relics: e.g., Central/Southern Italian: la mano – le mano (“hand”), la fico – le fico (“fig”), la peco – le peco (“sheep”), etc.; Calabria and Campania varieties: la nuoro – le nuoro (“daughter in law”), etc. (cf. Rohlfs 1949–1954). All these nouns are feminine, and so are the nouns that passed from other classes to the fourth (e.g., la suoro – le suoro “sister”). The pattern by which the speakers preserved the two inflectional classes is, then, clear: despite the merging of the two nominatives – which originally functioned as paradigm structure conditions – the two classes have been preserved by changing their distinctive feature, that is, the ending of the nominative was replaced with the grammatical gender. All the masculine nouns of the fourth class, in fact, went to the second; conversely, all the feminine nouns of the second class went to the fourth, which kept its own original feminine members and took “new” feminine nouns from other classes. The mixed paradigms of senatus and domus show the ongoing process. As is clear, the scope of application of the rule that assigned nouns to their inflectional class did not widen; rather, the features that identified the paradigms were replaced, in order to preserve the mechanism that allows speakers to automatically assign each noun to its class, on the basis of a feature that operates at the category level and, therefore, is shared by all members of the inflectional class.

The principle of economy of the memory system operates not only in morphology and in phonetics – as shown above – but also in syntax: it will be enough to mention here the so-called “solidarity principle”, dynamically investigated by Hawkins (1979, 1983, but already present in one of Greenberg’s (1963) “universals”.

In conclusion, Romano Lazzeroni clearly demonstrated how Sapir’s drift (Sapir 1921), Kiparsky’s principle of maximization (Kiparsky 1968) and Croft’s evolutionary theory refer to epiphenomena of functional constraints that have a universal value, as they rely on cognitive and neurobiological principles.

The crucial role of memory can be observed also in the organization of so-called “irregular” forms, that is, the forms that need to be memorized one by one: “non sarà un caso” – Romano Lazzeroni wrote (Lazzeroni 2019b: 178; cf. Lazzeroni 2005: 14–16) – “che le forme irregolari – e in particolar modo le più irregolari quali sono le forme suppletive – appartengano invariabilmente e in ogni lingua alle parole con l’indice di frequenza più alto” [it is no coincidence that the irregular forms – and especially the most irregular such as suppletive forms – invariably and in every language belong to the words with the highest frequency index]: frequency is a powerful support to memory. When the frequency level decreases, like in language acquisition and loss, in fact, automatic processing prevails, going even beyond etymological borders; for instance, “incorrect” forms such as the following have been reported: Italian *ando – andiamo, from andare ‘go’ (modelled on the regular paradigm amo – amiamo, from amare “love”); gallo (masculine) –*galla (feminine) ‘rooster’ (modelled on the type gatto (masculine) – gatta (feminine) ‘cat’, etc.; (cf. Antinucci and Miller 1976; Giacalone Ramat 1983, 2003; Lazzeroni 2005, 2020; Sasse 1992).

Furthermore, the principle by which speakers privilege processing (supported by procedural memory), in preference to storage (supported by declarative memory) also operates in the organization of irregular or sub-regular paradigms of words with a high or very high degree of frequency: we refer, for instance, to the long-debated issue of the partition of the paradigm in ancient and modern Indo-European languages, on which Romano Lazzeroni, once more, wrote memorable pages (Lazzeroni 2015c).

To summarize and conclude, Romano Lazzeroni’s investigations into various ancient and modern Indo-European languages clearly showed how speakers organize (or re-organize) irregular and sub-regular forms – be they innovations or residuals – in a “schema” (in the sense used by Bybee and Slobin 1982), on the basis of common features: this mechanism helps automatic processing and, therefore, reduces memory load. The schema entails the use of rules that operate at the category level and not at the unit level, and whose more or less probabilistic vs. invariable character may change over time. As shown by Romano Lazzeroni, the diachronic perspective is also the most appropriate to appreciate the “scalar character of (morphological) irregularity” (Ramat 1985).

5 The teaching and mentoring of Romano Lazzeroni

In the final section of this modest token of gratitude to a giant of science and big-hearted mentor, I would like to provide a small portrait of Romano Lazzeroni as a brilliant and unique teacher.

In Romano’s activity, scientific research and higher education have always been inseparable. This tight link is based on both his strong belief that education is impossible if it is not animated – inspired, encouraged, renewed – by the spirit of research and his profound desire to share his path with younger scholars (who – in Romano’s world – also included undergraduate students). He used to say that “one of the greatest joys in life is appreciating how young eyes can shine as the mind runs a new adventure”. In Romano’s idea of academia, there were no temporal, environmental, spatial or national borders, no ideological barriers, no separation between areas of knowledge, including both humanities and science.

The Department of Linguistics of the University of Pisa was constantly animated by his presence: when conferences or other important commitments did not take him outside Pisa, Romano was at work every day, all day long, until very late, and always available – and even delighted – to welcome students and colleagues, independently of their background or academic position. No one left Romano’s office without a precious suggestion or an illuminating answer. Having a conversation with him was always an entertaining experience: his profound and sharp mind was accompanied by an extraordinary sense of humor, that made even the most complex phenomenon or the most critical situation clear and light.

Romano Lazzeroni was everyone’s teacher, probably because he considered anyone – and, in particular, the younger – a teacher. Remembering a well-known avant-propos by Meillet, he used to joke about his activity as a teacher by saying that scholars grow up by themselves. His example, instead, profoundly and clearly shows that scholars grow up thanks to the generosity of their mentors and that the greatness of a mentor is measured by his ability to set his students free to make their own choice and find their own way. Romano constantly advised his students to follow the teachings and research perspectives and methods of other scientists and professors as well; he never considered his students as property but, rather, as someone to respect and support.

One of the foundational principles of his teaching relied on his in-depth awareness of the limits of our activity, as both researchers and educators. He strongly believed that the main progress in scientific research does not come from (or with) authority, but from being part of a community, in which each piece of research opens up the way for further research, thus contributing to a potentially unlimited process (Lazzeroni 2019a). Romano’s legacy will always be part of this process.