The Lexicon of Transitive Verbs of Motion and the Asymmetry Between Goal and Source PPs

,


Introduction
The aim of this paper is the empirical verification of Nam"s hypothesis (2004) on the asymmetry between source and goal PPs, which was confirmed by the results of the experiments carried out by Lakusta andLandau (2005, 2012). In particular, Lakusta and Landau claim that the prominence of the goal with respect to the source, i.e. the goal bias, at the cognitive level may also be related to the possible presence in the lexicon of a higher number of verbs that require goal PPs rather than source PPs.
The classification and analysis of about 500 Italian transitive verbs of motion allowed me to verify the prominence of goal patterns in the lexicon, as hypothesized in Lakusta and Landau"s research. The Appendix presents the six classes into which the Italian transitive verbs of motion are classified according to the principles of the Lexicon-grammar framework, as developed by Gross (1975Gross ( , 1995 (Note 1).
Empirical analysis of the data allowed me (a) to numerically weigh the verbs that require the source and/or goal PPs, and (b) to highlight the syntactic properties that distinguish goal and source verbs. For example, transitive-intransitive pronominal alternation has a high applicability with caused motion verbs that take a goal PP and a low applicability with those that take a source PP. Furthermore, subject alternation is systematically applied only to goal verbs following a certain syntactic pattern. This paper is organized as follows. Section 2 presents some of the more influential hypotheses on the representation of spatial expressions and the relationship between language and thought. I present Nam"s hypothesis on the asymmetry between goal and source PPs, and Lakusta and Landau"s claims on the goal bias. Section 3 provides an overview of the motion verb patterns, while sections 4 to 6 present a lexical-syntactic analysis of Italian transitive verbs of motion. Section 7 describes the results of the quantitative analysis carried out on Italian, while section 8 discusses the results of the diagnostics applied to the Italian data. Section 9 presents some conclusions.

Linguistic and Psycholinguistic Hypotheses
"Spatial cognition is central to human thinking, and spatial language is thus an important area of study, as it may reveal fundamental properties of human thought" (Levinson 2009). The literature devoted to spatial expressions is rich precisely because spatial cognition, despite being a central aspect of human thought, shows cross-linguistic variation.
S-framed languages encode manner in the verb and path in a satellite to the verb. V-framed languages encode path in the verb and manner in an adjunct clause.
The conceptual structure of motion events is universal while the linguistic structures are cross-linguistically variable. Jackendoff (1983) analyzes the universal conceptual representation of spatial expressions and, at the same time, accounts for the linguistic differences between English and Spanish (Jackendoff 1990: 223). There is a long-standing debate on cross-linguistic variation and the relationship between language and thought. Levinson (1996Levinson ( , 2009) and Evans and Levinson (2009) claim that linguistic patterns shape the conceptual patterns of speakers and, therefore, conceptual representations are not all universal. Slobin (1996Slobin ( , 1998 argues that "language may have an effect on the process of thinking for speaking" (Gennari et al. 2002) and highlights the critical aspects of Talmy"s typological hypothesis. In the context of lexical semantics, Levin and Rappaport (2019) and Beavers et al. (2010) propose a way of overcoming Talmy"s typology by developing new hypotheses on the relationship between linguistic variation and conceptual representation.
Extensive experimental cognitive research has been carried out on the relationship between the conceptual representations and linguistic patterns of spatial events. A number of scholars (cf., among others, Gennari et al. 2002, Papafragou et al. 2002, 2006, Papafragou and Selimis 2010) discuss Talmy"s and Slobin"s hypotheses. On the other hand, Lakusta (2005), Lakusta andLandau (2005, 2012), Lakusta et al. (2007, Lakusta and DiFabrizio (2017), Johanson et al. (2019) test the linguistic hypothesis proposed by Filip (2003), Nam (2004Nam ( , 2012, and Markovskaya (2006) on the asymmetry between source and goal locatives. Georgakopoulos (2018) shows the prevalence of Goal paths in ancient Greek. Nam (2004) argues in favor of syntactic and semantic asymmetry between goal PPs and sources PPs. He claims that source PPs are generated under the higher VP and are, therefore, adjuncts. Goal PPs behave like internal arguments and are generated under the lower VP. I will only take some of Nam"s syntactic evidence into account; for further details the reader is referred to the original paper. The syntactic contrast between goals and sources is shown by topicalization ((1)-(6) are from Nam 2004): the source PP in (1) can move to the front, whereas the goal PP in (2) resists such a move. "This suggests that the Goal PP behaves more like a true complement of the verb send than the Source PP does" (Nam 2004: 11, par. 2.4):

Asymmetry Between Source and Goal: The Linguistic Evidence
(1) From Los Angeles John sent the letter to Chicago (2) ??To Chicago John sent the letter from Los Angeles Furthermore, the PP from the library in (3) can move over the durative adverbial for ten minutes, but the goal PP to the library in (4) is not allowed to move over the time-frame adverbial in ten minutes. Therefore "a Source PP is more ready to scramble with a temporal/aspect PP" (Nam 2004: 11, par. 2.4): (3) a. He ran from the library for ten minutes b. He ran for ten minutes from the library International Journal of Linguistics ISSN 1948-5425 2020 www.macrothink.org/ijl 84 (4) a. He ran to the library in ten minutes.
b. ??He ran in ten minutes to the library. Further evidence is shown by locative alternation, as in (5) and (6): (5) a. John sprayed paint on the wall b. John sprayed the wall with paint (6) a. The woman embroidered flowers on the jacket b. The woman embroidered the jacket with flowers Nam (2004: 13, par. 2.5) claims that "locative PPs involved in such alternations are mostly Goal-type locatives [...] we claim that V-modifiers like Goal-type PPs allow locative alternation while VP-adjuncts like Source [...] hardly do". Gehrke (2007: 86, par.1) argues that there is no such semantic and syntactic asymmetry and states "that the apparent differences between goals and sources derive from other factors". For reasons of space, I will not address Gehrke"s discussion, and I refer the reader to the cited papers for further details.

Asymmetry Between Source and Goal: Cognitive Evidence
Despite Gehrke's criticism, the asymmetry between source and goal PPs is one of the most widely studied topics in cognitive experimental research, for example Lakusta (2005), Lakusta andLandau (2005, 2012), Regier and Zeng (2007), Papafragou (2010), Lakusta and DiFabrizio (2017), Johanson et al. (2019), and Do et al. (2019). Landau (2010: 73) states that studies on spatial language can provide insights into "how the non-linguistic spatial systems shared by all species support the human ability to acquire and use language". As Landau (2010: 74-75) reports, there are two main hypotheses on spatial language. The author calls them the Space First and the Language First hypotheses. According to the Space First hypothesis, spatial language is learned early and easily because it reflects the representation already present in the non-linguistic spatial system prior to language learning. On the other hand, the Language First hypothesis states that spatial language reflects representations created by exposure to spatial expressions in one"s native language. She also states that the latter hypothesis is "appealing because we know there are significant cross-linguistic differences in the way that languages express spatial relationships [...]" but it is not convincing since "it ignores the large attested similarities in the spatial concepts encoded by languages world-wide". Landau andLakusta (2009), andLandau (2010) propose a third hypothesis where spatial language is a hybrid of spatial representation and language. I will not here discuss the details of the hybrid hypothesis, and I refer the reader to the cited papers for further details.
A great deal of experimental research investigating the relationship between spatial language and non-linguistic representations has been carried out by examining paths. Consider the sentences below (drawn from Landau 2010: 78): The PP to the bottom of the maze in (7) encodes the goal path, whereas the PP from the top of the maze encodes the source path in (8), and in (9) both source and goal paths are expressed. Landau and Zukowski (2003), Lakusta and Landau (2005), Landau (2010), and Lakusta and Landau (2012) carried out experiments on paths, and the results pointed out an asymmetry between source and goal in the way events are described. There is a strong bias among children and adults to avoid expressing the source alone: the authors call it the goal bias (Landau 2010: 81, 83). The goal pattern also holds for such events as a transfer of possession (The book went/passed from Jane to John) or a change of state (The bear's ears changed/went from green to red) (Note 2). Landau (2010: 89) explores the cause of this asymmetry and states that "the strong and absolute tendency for language to encode goal paths in preference to source paths may be rooted in the patterns of language to which the child is exposed. Specifically, there may be more verbs that take goal path expressions, and these may be represented in the language the child hears at a much higher frequency than verbs that take source path expressions". Similarly, Lakusta and Landau (2012: 538) raises the following questions "What is less clear in the case of goals and sources is whether there is a distinction between goal and source PPs at the grammatical level (as there is between the syntactic subject and object). Do goal and source PPs (e.g., into x, out of x, to x, from x) differ in their grammatical status? For example, are goals arguments and sources adjuncts of manner of motion verbs (e.g., ""run"")? Do goal PPs show up as core syntactic arguments more than source PPs (e.g., are there more verbs of attachment, such as ""put"", than detachment verbs, such as ""remove"")? If so, then the alignment of goal and source elements at the conceptual, pragmatic, and grammatical level may explain the mapping of goals and sources in language". It is conceivable that the goal path is more prominent than the source path and also that agents are more prominent than patients when they are subjects (see Fillmore"s thematic hierarchy (1968)) (Note 3). The prominence of goals is also evident at a pragmatic level: a goal event presupposes a source, but the opposite is not true. Lakusta and Landau (2012: 517, 538) states that if there are more verbs of attachment, such as put, than detachment verbs, such as remove, it is conceivable "that the path asymmetries observed in language stem both from the properties of non-linguistic event representations and constraints internal to language". The analysis and classification of Italian transitive verbs of motion I present in this paper will make it possible to investigate the question raised by Landau and Lakusta (2012): Are there more verbs of putting than verbs of removing?
I will examine the syntactic patterns of Italian transitive motion verbs and check the frequency of the verbs that present (a) only PPs encoding the goal path, (b) only PPs encoding the source path, or (c) PPs encoding both source and goal paths. In the case of verb pattern (c), the analysis will also focus on verifying Nam"s hypothesis on syntactic asymmetry between source and goal PPs. Even though the cited cognitive and linguistic research considers only goal/source PPs, I will analyze also those syntactic patterns where goals and sources are expressed by a direct object or by the noun in the denominal location verbs. This will provide a more complete picture on the expressions of goals and sources in the lexicon of Italian.

The Argument Structure of Verbs of Motion
Transitive verbs of motion can take only a direct object, as in (10), or a direct object and one or two prepositional phrases, as in (11) Unaccusative verbs of motion can take a prepositional argument, as in (14), or two prepositional arguments, as in (15). Unergative verbs of motion may take a prepositional adjunct, as in (16): The train pro-comes from Milan (Note 5) "The train comes from Milan" (15) Maria arrivò a casa dall" università Maria arrived at home from=the university "Maria came home from university" International Journal of Linguistics ISSN 1948-5425 2020 (16) Maria camminò nel parco (Note 6) Maria walked in=the park "Maria walked in the park" Some verbs may have both a directed motion use and a stative use, such as attraversare (cross) in (17) Locative complements are traditionally associated to PPs, although they can occur as the head of a direct object. Their interpretation depends on the verb they occur with. As stated by Guillet and Leclè re (1992: 8) and Gross (1995) (Note 7), the toponym Roma denotes "the Romans" in (21), since the sentence sconfiggere un luogo (to defeat a place) is not licensed. The toponym Roma denotes a place in both (22) and (23), however only attraversare (cross) is a motion verb, though both sentences attraversare/fotografare un luogo (cross/photograph a place) are admitted: The Wh-question introduced by (Preposition) Dove (where) is traditionally applied to distinguish locative complements. However, question (24) is inappropriate, even though International Journal of Linguistics ISSN 1948-5425 2020 attraversare (cross) is a motion verb in (22). It is the Wh-question introduced by Che cosa (What), as in (25), that is correct: (24) -*Dove attraversò Gianni?

Where crossed
Gianni?
What/Which place crossed Gianni?
"What place did Gianni cross?" -Roma In this paper I will examine only transitive caused motion verbs. I will analyze unaccusative motion verbs in further research.

Transitive Caused Motion Verbs
Caused motion verbs that take two complements enter two different patterns: In (26) the direct object denotes the located entity while the prepositional complement denotes the place/location. In (27) the direct object denotes the place/location while the prepositional complement denotes the located entity. In both sentence patterns the subject has the role of the causal Agent.
In (26) the direct object denotes the locatum, while the prepositional complement denotes the location. On the other hand, in (27) the direct object denotes the location, while the prepositional complement denotes the locatum (Note 8). Verbs like mettere (put) have a locatum-location pattern, while verbs like riempire (fill) have a location-locatum pattern.

The Locatum-Location Pattern (Class A)
Locatum-location verbs may take one PP that performs the thematic role of source in (28) and goal in (29), or two PPs denoting the source and the goal in (30): International Journal of Linguistics ISSN 1948-5425 2020 (28) Maria estrasse la pistola dalla fondina Maria extracted the gun from=the holster "Maria pulled the gun out of the holster" (29) Maria inserì la pistola nella fondina Maria inserted the gun in=the holster "Maria put the gun into the holster" (30) Lei trasportò la merce da Milano a Torino "She transported the goods from Milan to Turin" The locatum-location pattern includes 179 verbs: of which 118 take a goal PP, 29 take a source PP, and 32 take both a source and a goal PP. The data show that the majority of locatum-location verbs have a goal pattern (66%), while source verbs are the least frequent (16%). Verbs included in this class may enter transitive-intransitive pronominal alternation. As will be shown below, this property has a greater or a lesser applicability in relation to the type of PP. Before proceeding, I will give some details on transitive-intransitive alternation.
In Italian, a causative transitive sentence may alternate with an intransitive pronominal sentence that can be considered either a reflexive variant, as in (31b) and (33b), or an anticausative variant, as in (32b)  b. I rifiuti si dispersero nell" ambiente (con il vento/*con Gianni) The waste SI spread in=the environment (with the wind/ with Gianni) "The waste spread into the environment with the wind/*with Gianni" Three types of hypotheses have been proposed in the literature: (a) causative alternation and reflexive constructions can both be explained by means of the same reflexivization process (Chierchia 2004), (b) reflexive constructions can be analyzed according to the same mechanism as causative alternation (Kayne 1975), and (c) the two constructions are independent of each other (Schä fer and Vivanco 2013). The on-going debate shows that the two types of si-constructions are morphosyntactically similar (Note 11).
Given their similarities, and since the object of this paper does not concern the analysis of pronominal forms, I will not distinguish between anticausative and reflexive variants, but I will consider them all generically as intransitive pronominal variants.

The Location-Locatum Pattern (Class B)
The location-locatum pattern includes 29 verbs whose direct object may denote the goal in (40) and the source in (41) These verbs may have a "holistic" reading (Anderson 1971): they imply that the reference object occupies (partly or entirely) the place denoted by the direct object. This reading is absent for locatum-location verbs.
Goal verbs (22 out of 29) are more numerous than source verbs (7 out of 29), but in this case transitive-intransitive alternation applies equally to half the goal verbs (about 11 out of 22 verbs) and half the source verbs (about 4 out of 7 verbs), as in (42)  The stadium SI cleared of=the fans "The stadium was cleared of fans" The data show that transitive-intransitive alternation is much less frequent when the source is expressed by a prepositional phrase (as in locatum-location verbs) rather than a direct object (as in location-locatum verbs). In other words, this type of alternation does not produce any differences in the frequency rate when goal and source are expressed by a direct object. ISSN 1948-5425 2020 The property that characterizes location-locatum verbs is subject alternation (Levin, 1993: 79 andsubsequent, Levin andRappaport 1998). This type of alternation does not involve any change in transitivity: the noun of the prepositional phrase in (44) and (46) appears in subject position in (45) and (47), while the subject of (44) and (46) is obligatorily omitted in (45)  Subject alternation is systematically applicable to all goal verbs, while only 2 source verbs participate in such alternation. Table 2 summarizes the data:

Locative Alternation (Class C)
In Italian, as in other languages, locative alternation is a property of some caused motion verbs: a limited number of them enter both the locatum-location pattern and the location-locatum pattern, such as caricare (load), cospargere (sprinkle), disseminare (disseminate), spalmare (spread), spruzzare (spray), spolverare (dust), sgombrare (clear), as in the sentences below: (48) L" esplosione disseminò tutta l" area di rottami The explosion spread whole the area of debris "The explosion spread the whole area with debris" (49) L" esplosione disseminò i rottami su tutta l" area The explosion spread the debris on whole the area Unlike other languages, Italian is lexically poor in verbs that enter the locative alternation, with only 4 out of 14 alternating verbs taking a source PP (Note 12). The data confirm Nam"s claim about locative alternation. Unlike location-locatum verbs, none of these verbs allows subject alternation, while transitive-intransitive pronominal alternation is allowed by almost all source verbs.

Denominal Verbs: Locatum and Location Verbs (Classes D and E)
One-complement transitive verbs include two types of denominal verbs: locatum and location verbs.
In the examples below, imburrare (butter) and sminare (demine) are verbs whose base noun denotes the located entity, while the direct object denotes the goal in (52) and the source in (53) (Note 13): (52) Maria imburrò la teglia "Maria buttered the pan" (53) Gli artificieri sminarono tutta la zona The sappers demined whole the area "The sappers demined the whole area" They are called denominal locatum verbs (Class D) and are similar to location-locatum verbs (section 4.2), as in the examples below: (54) Maria ricoprì / spalmò la teglia di/con il burro Maria covered/ smeared the pan of/with the butter "Maria covered/smeared the pan with butter" (55) Gli artificieri liberarono la zona dalle mine International Journal of Linguistics ISSN 1948-5425 2020 The sappers cleared the area from=the mines "The sappers cleared the area of the mines" These two different verb patterns show similarities and differences. In both cases the subject is a causal agent and the direct object has the role of source/goal with a holistic reading. Denominal locatum verbs contain a noun corresponding to the located entity, as in (52) and (53). For location-locatum verbs the located entity is expressed by the noun of the PP, as in (54) and (55).
Both types of verbs permit a PP introduced by di/con/da "of/with", where the noun has to be either a count noun in the plural or a mass noun. The count noun can be in the singular iff it denotes an entity that has an "extension". However, the PP is an argument with location-locatum verbs while it is an adjunct with denominal locatum verbs. In order to be permitted, the noun of the adjunct PP has to specify the nature of the base noun more precisely (Labelle 2000: 228, note 7, Jackendoff 1990, as in (56) ISSN 1948-5425 2020 www.macrothink.org/ijl 96 The slope SI in-snow "The slope was covered with snow" Section 4.2 pointed out that location-locatum verbs systematically participate in subject alternation only in the case of goal verbs, as in (59). Denominal locatum verbs show the same syntactic behavior, but only in 64 goal verbs out of 89, as in (60) The lower frequency of subject alternation may be due to the fact that the locatum is incorporated in the verb. The data is summarized in Table 4: Denominal location verbs (Class E) contain a noun denoting the source/goal, while the direct object denotes the located entity. The verbal prefix distinguishes the source from the goal (see also Note 13): it indicates the goal in (61), the source in (62) ISSN 1948-5425 2020 "Maria took the cake out of the oven" (63) Maria rinvasò la pianta "Maria repotted the plant" Denominal location verbs are similar to caused motion verbs whose pattern is locatum-location (section 4.1), as (64) Both types of verbs take a prepositional phrase that is an argument for locatum-location verbs, but an adjunct for denominal location verbs, because in the latter case the noun of the prepositional phrase is already contained in the verb. Moreover, the adjunct has to further specify the Place, otherwise the output sentence is unacceptable, as in (67) ISSN 1948-5425 2020 www.macrothink.org/ijl 98 b. Il vino si rovesciò sulla tovaglia The wine SI spilled on=the tablecloth "The wine spilled on the tablecloth" Denominal location verbs participate in transitive-intransitive pronominal alternation, only when they denote a goal as in (69), although the frequency rate is much lower than in locatum-location verbs, i.e. 12 out of 59 goal verbs. Again, the data show that this type of alternation is widespread when the goal is expressed by a prepositional phrase: (69) a. Il capitano imbarcò i passeggeri "The captain boarded the passengers" b. I passeggeri si imbarcarono The passengers SI boarded "The passengers boarded" No source verbs participate in Subject alternation either, while only 4 out of 59 goal verbs do. The data is summarized in Table 5:

Non-denominal One-Complement Transitive Verbs (Class F)
In the Italian lexicon there are some non-denominal transitive verbs whose direct object denotes the source in (70) The pilot over=flew Paris "The pilot flew over Paris" ISSN 1948-5425 2020 The verbs lasciare (leave) and raggiungere (reach), besides the direct object, may take a prepositional phrase denoting the goal in (74) and the source in (75) ISSN 1948-5425 2020 (83) Maria passò oltre il ponte "Maria crossed over the bridge" (84) Maria circumnavigò l" isola "Maria circumnavigated the island" (85) Maria navigò intorno all" isola Maria sailed around at=the island"

International Journal of Linguistics
"Maria sailed around the island" The verb circumnavigare (circumnavigate) in (84) denotes a motion starting and finishing in the same place.
English also shows that verbs like enter (The bear entered the cave) and walk into (The bear walked into the cave) are considered equivalent: the meaning of into is incorporated into the meaning of enter (Jackendoff 1983, Zwarts 2006. Jackendoff (1983: 183, par. 9.5) states: "In order to deal with the more general case, we must ask how a conceptual structure can be carved up into lexical items. The verb "enter" serves as a good preliminary example. "The dog entered the room" can be paraphrased by "The dog went into the room." Both sentences have the semantic structure [...] in which "the dog" is theme and "the room" is the reference object of the path. However, this structure is lexicalized differently in the two cases. [...]. In other words, the verb "enter" itself lexicalizes the path-and place-functions instead of leaving them to be overtly expressed by a preposition".
This class is very heterogeneous and includes 41 verbs: 14 of which denote a median place, 12 denote a goal, 8 indicate a passage, 4 denote the source, and only 3 denote both the source and the goal. Nevertheless, goal verbs are much more frequent than source verbs.
Transitive-intransitive alternation is permitted in few entries: 4 out of 12 goal verbs, and only 1 out of 4 source verbs. This rate confirms that verbs of this class are similar to unaccusative motion verbs, where alternation occurs very rarely. Table 6 summarizes the data:  ISSN 1948-5425 2020 Table 7 reports the overall quantitative data related to the transitive motion verbs examined in this paper. Considering only those verbs that denote either the source path or the goal path, it can be observed that goal verbs account for 310 out of 426, i.e. 72%. These data confirm the prominence of the goal pattern in the lexicon.

Data and Results
The low number of Italian source verbs participating in locative alternation confirms Nam's observation that this type of alternation concerns primarily goal verbs. The data is summarized in Table 7: The syntactic analysis of transitive motion verbs highlights that there is also an asymmetry between goal and source with regard to transitive-intransitive pronominal alternation and subject alternation. The former applies more frequently to goal verbs, while the latter applies systematically only to goal verbs.
As far as transitive-intransitive alternation is concerned, analysis of the data reveals that locatum-location goal verbs (mettere "put") and denominal location verbs (infornare "bake") are the majority that accept this property. On the other hand, location-locatum verbs (riempire "fill") and denominal locatum verbs (imburrare "butter") show that the source/goal pattern does not affect the frequency rate of transitive-intransitive pronominal alternation. Therefore, if the goal/source is expressed by a direct object, no asymmetry occurs.

Asymmetry Between Goal and Source: Evidence From Italian
As already stated in section 2.1, Nam (2004) points out that topicalization in sentences (86) and (87) ISSN 1948-5425 2020 "Luca transported the goods from Rome to Milan" Topicalization produces the acceptable sentences (88) The application of further diagnostics, such as the pro-form replacement test lo+fare (in English this pro-form corresponds to do so), confirms that source and goal PPs behave differently. If a phrase must necessarily be deleted when replacing part of the verbal phrase with the pro-form, that phrase is an argument, otherwise it is an adjunct (see Schütze, 1995: 105 for pro-forms in English), as in the examples below: (92) Lia mise gli abiti nella valigia stamattina, e Eva lo fece ieri Lia put the clothes in=the suitcase this=morning, and Eva it did yesterday "Lia put the clothes into the suitcase this morning, and Eva did so yesterday" (93) *Lia mise gli abiti nella valigia, e Eva lo fece nell" armadio Lia put the clothes in=the suitcase, and Eva it did in=the closet "*Lia put the clothes into the suitcase, and Eva did so into the closet" Let us apply the pro-form lo+fare to sentences (86)-(87). When replacing only the verb and the goal PP, the pro-form produces the dubious sentences (94)-(95). When replacing the verb and the source PP, the pro-form produces the ungrammatical sentences (96)-(97): In the sentences examined above, the Agent shares only the source with the located entity, as in (86), or both the source and the goal, as in (87). If the Agent shares neither the source nor the goal with the located entity, as in (98), the diagnostics gives us partially different results: (98) Luca spostò i libri dalla mensola sulla scrivania Luca moved the books from=the shelf on=the desk "Luca moved the books from the shelf onto the desk" Topicalization produces a grammatical sentence when moving the source PP to the front, as in (99), while the goal PP resists such a move in (100) ISSN 1948-5425 2020 On the other hand, the pro-form lo+fare produces ungrammatical sentences, either applied to the goal PP as in (101) or to the source PP as in (102): (101) *Luca spostò i libri dalla mensola sulla scrivania, e Lia lo fece dalla scatola Luca moved the books from=the shelf on=the desk, and Lia it did from=the box "*Luca moved the books from the shelf onto the desk, and Lia did so from the box" (102) *Luca spostò i libri dalla mensola sulla scrivania, e Lia lo fece sulla sedia Luca moved the books from=the shelf on=the desk, and Lia it did on=the chair "*Luca moved the books from the shelf onto the desk, and Lia did so onto the chair" The sharing of source/goal between the Agent and the located entity affects the syntactic behavior of caused-motion verbs. Given these results, the syntactic status of source and goal PPs deserves further analysis, also taking into account the role of locative PPs in unaccusative motion verbs.

Conclusions
The asymmetry between goals and sources has been asserted from a linguistic perspective in Nam (2004) and from a cognitive perspective in Lakusta andLandau (2005, 2012). In particular, Landau (2010) hypothesized that the goal bias may be also related to the higher number of goal verbs in the lexicon, with the result that children are exposed to this type of verbs more frequently. Lakusta and Landau (2012: 538) states that if there are more attachment verbs, such as put, than detachment verbs, such as remove "then the alignment of goal and source elements at the conceptual, pragmatic, and grammatical level may explain the mapping of goals and sources in language".
In order to verify the quantitative presence of goal and source patterns in the lexicon, I classified and analyzed 486 Italian transitive verbs of motion and detected 6 classes of verbs.
In the Italian lexicon, out of 486 transitive motion verbs 310 show the goal pattern, 116 show the source pattern, and 38 shows the source-goal pattern ( Table 7). The data confirm the prominence of goal patterns. The availability of cross-linguistic data on this topic would provide further evidence to the fact that the goal bias may be related to the lexicon.
I have shown that transitive-intransitive pronominal alternation is much more frequent with goal verbs especially when the goal is expressed by a prepositional phrase. Further research needs to be carried out in order to determine whether this behavior is motivated or due to idiosyncrasies in the lexicon. Analysis of the data also showed that subject alternation is systematically applicable only to goal verbs.
I have checked the syntactic asymmetry in Italian by applying not only topicalization (as in Nam 2004) but also pro-form. The results confirm Nam"s hypothesis on syntactic asymmetry between goal and source PPs. Nam claims that goal PPs are arguments while source PPs are not, however the syntactic status of such prepositional phrases (whether or not they are both arguments) deserves further analysis taking into account unaccusative motion verbs as well as manner-of-motion verbs. This is an important point at a cognitive level as Lakusta and Landau (2012: 538) points out: "Do goal and source PPs (e.g., into x, out of x, to x, from x) differ in their grammatical status? For example, are goals arguments and sources adjuncts of manner of motion verbs (e.g., ""run"")?
Moreover, the diagnostics highlight that the sharing of source/goal paths between the Agent and the located entity affects equally the grammaticality of the pro-form lo+fare whether applied to source or goal PPs. All these issues deserve future research in order to formulate an explanation.