1 DILLo—Database Italiano del Lessico per Logopedisti (‘Italian lexical database for speech pathologists’)

DILLo—Database Italiano del Lessico per Logopedisti (‘Italian Lexical Database for Speech-Language Pathologists’) is an online resource created through collaboration between linguists, speech pathologists, and computer scientists. It is specifically intended to facilitate the development of speech and language therapy treatment tasks. Its main goal is to provide clinicians with word lists selected based on phonological, phonotactic, and morphological properties, and their frequencies. These cues are crucial for administering appropriate stimuli to the patients based on their rehabilitation needs. Therefore, DILLo is a lexical database that allows Speech-Language Pathologists (SLPs) to obtain a rich, tailored, and varied selection of suitable stimuli through a quick and easy search.

DILLo's applications include the treatment of Developmental Communication Disorders (specifically, organic or functional SSDs—Speech Sound Disorders, such as articulation disorders, phonological disorders, and CAS—Childhood Apraxia of Speech; cf. APA, 2013; Bowen, 2023), SpLDs—Specific Learning Disabilities (i.e., dyslexia, dysgraphia, and dysorthography), aphasia, dysarthria, dysphonia, and the auditory training that follows cochlear implantations.

Similar resources, intended to collect words coded for (psycho)linguistic properties, already exist for other languages. Examples are the MRC Psycholinguistic DatabaseFootnote 1 (Coltheart, 1981; Wilson, 1988) for English and Norwegian WordsFootnote 2 (Lind et al., 2015), or the Croatian Psycholinguistic DatabaseFootnote 3 (Peti-Stantić et al., 2021). Nonetheless, analogous databases cannot automatically be transferred to other languages, especially if their targeted users and scopes are different. Thus, DILLo fills a gap in the context of Italian clinical linguistics and enriches the tools for speech-language treatments.

DILLo includes 7690 Italian words orthographically transcribed in Standard Italian, which may be consulted using three different query options. Each word has been manually coded and verified by a team of trained linguists to ensure its accuracy.

This paper is organized as follows: Sect. 2 provides an overview of the linguistic information on which DILLo was designed and, as a result, included in the tool. Section 3 instead illustrates DILLo’s interface and search options to explain how to consult the database. Finally, Sect. 4 presents future developments and research directions for the tool.

2 Coded features of the words in DILLo

Sections 2.1–2.5 describe and motivate the various linguistic and paralinguistic features that DILLo's words have been coded for: phonological transcription and phonotactic structure, word frequency, regional variants, parts of speech, and ARASAAC pictograms.

2.1 Phonological transcription and phonotactic structure

DILLo contains the phonological transcription and phonotactic structure of all included words, thus offering the possibility to search for specific phonemes, or words with different syllabic lengths and stress types.

The inclusion of a phonological transcription is mainly due to the common notion of some phonemes being more complex than others, consequently leading to different ease of production. Epiphenomena of this complexity are the earlier acquisition of simpler phonemes by children (Gayraud et al., 2018; Kager et al., 2004; Stoel-Gammon, 1985) and their better preservation after brain damage (Buchwald, 2009; Galluzzi et al., 2015; Marquardt et al., 1979; Romani & Galluzzi, 2005; Wolk, 1986).

The articulatory complexity of Italian phonemes—and thus their relevance for SSDs—can be based on their acquisition trajectories. Phonemes acquired later in life can be considered more complex than those acquired earlier. For instance, the study by Zanobini et al. (2012) showed that children aged 36–42 months were able to produce voiceless plosives, bilabial and alveolar nasals (/m/ and /n/), and the alveolar lateral approximant /l/. More than 80% of the participants had an inventory encompassing at least 16 of the 23 consonants, as they successfully articulated voiced plosives (excluding /g/), all fricatives, the unvoiced postalveolar affricate /ʧ/, the alveolar trill /r/, and the approximants /j/ and /w/. A minority of the children exhibited production of the voiced postalveolar affricate /ʤ/, the voiced plosive /g/, the voiceless postalveolar fricative /ʃ/, the lateral palatal /ʎ/, or the nasal /ɲ/. Notably, none of the children included the affricates /ts/ and /dz/ in their consonant inventories. Another indirect measure of phonological complexity can be derived from information regarding which sounds more frequently or intensively undergo simplification, as this need to simplify suggests they are more complex than the others. Both typical and atypical simplification strategies (and timing) have been extensively described in the literature (Chilosi et al., 2014; Sabbadini et al., 2000). Nevertheless, phonetic and phonological development are not independent, as they are strictly intertwined together with lexical development (Vihman, 2017; Zamuner & Thiessen, 2018). For specific information regarding the order of acquisition of phonemes and phonological processes in Italian children, and their relationship with lexical development, readers can refer to Bortolini (1995), Bortolini et al. (1996), Zmarich and Bonifacio (2004), Pinton et al. (2014), and Viterbori et al. (2018).

In addition to the number of acquired phonemes, clinicians must also consider how their realization varies depending on their position within the word (Bortolini et al., 1995, 1996; Bortolini & Leonard, 1996), as context-dependent diversification of outcomes may indicate phonological impairments (Bortolini & Leonard, 1991a, 1991b; Pinton et al., 2014). Moreover, phonological reorganization and coarticulation phenomena are also relevant in the rehabilitation of adults with dysarthria or apraxia of speech (Hardcastle & Tjaden, 2008; Ziegler & Von Cramon, 1985).

Aside from words segmental features, the suprasegmental level also impacts language acquisition. Suprasegmental factors that may affect acquisition are:

  • Phonotactic distributions: sequences of sounds, licit in a language, do not all occur with the same frequency. Thus, the number of inputs children receive for these sequences might differ. This aspect has also been considered influential for SSDs (Coady & Aslin, 2004; Zamuner et al., 2004).

  • The syllabic structure (e.g., length and type of onset or coda, tautosyllabicity of segments, number of syllables in the word, alternation between weak and strong syllables). For instance, the relevance of word length for SSDs has been thoroughly proven (Dollaghan & Campbell, 1998; Weismer et al., 2000). The number of syllables is also important in acquired language disorders; the output buffer deficits are affected by the word length effect (Caramazza et al., 1986; Patterson, 1986).

Evidence of a connection between complexity in acquisition and SSDs can be found among speech errors or simplification processes, that are indicators of speech impairments (e.g., symptoms of apraxia of speech) (Bortolini, 1995; Bortolini & Leonard, 1991a; Galluzzi et al., 2015; Zanobini et al., 2012). Both complex phonemes and complex suprasegmental features play a role in these processes. Segments that appear later in the acquisition trajectories, such as fricatives and affricates, present difficulties for patients; weak syllable deletion, metathesis, epenthesis, diphthong reduction, or consonant/vowel harmony are all sensitive to the complexity of the phonotactic structure of the word.

In people with SSDs, these difficulties with both phoneme-related and suprasegmental aspects can also affect other linguistic abilities, e.g., the marking of verbal agreement: an example is described by Leonard et al. (1992) and Bortolini et al. (1996). In neurotypical conditions, Italian is a dominant penultimate stress pattern language. Penultimate-stressed words are the most frequent; the second most common type of primary stress is on the antepenultimate (Nespor & Bafile, 2008). The distribution of different kinds of stress in the Italian lexicon is reflected by the words in DILLo, as Table 1 shows:

Table 1 Distribution of words with different types of stress in the DILLo database
Full size table

Consequently, Italian children with a speech impairment may tend to pronounce antepenultimate-stressed words atypically to avoid less frequent syllabic sequences such as strong–weak-weak—as in cantano [ˈkan.ta.no], ‘they sing’, where they might delete the final weak syllable. However, in doing so, they also lose morphological information: by pronouncing canta [ˈkan.ta], ‘he/she sings’, they change the person of the verb.

A clinical application where both the frequency of phonotactic patterns and the length of the words are taken into account is the non-word repetition task (NRT). The NRT is considered one of the leading tests of speech perception, phonological encoding and assembly, and articulation; in fact, it has been frequently used as a diagnostic tool for a variety of SSDs (Munson et al., 2005; Coady & Evans, 2008; cf. D’Amico, 2000 on Italian).

Due to the centrality of segmental and suprasegmental aspects for speech impairments, each word in DILLo’s database was phonologically transcribed in Standard Italian (cf. §2.3.) and coded for its phonotactic structure. Transcriptions were manually carried out by two linguists and then manually re-examined by three other linguists to reduce human errors. The same procedure was applied to the phonotactic structure, extracting information regarding the type of stress and syllabic length (cf. Table 2) of words in DILLo.

Table 2 Distribution of words of different syllabic lengths in the DILLo database
Full size table

2.2 Word frequency

Language processing and acquisition have been shown to be sensitive to frequency at any level of representation (Bybee & Hopper, 2001; Diessel & Hilpert, 2013; Ellis, 2002). Psycholinguistic research has shown the pivotal role of frequency and distributional information in the acquisition of grammatical patterns (Goldberg, 2006; Tomasello, 2003) and word classes (Diessel, 2007), configuring a generalization process on statistical grounds. Focusing on the lexicon, besides the earlier acquisition of more frequent words by children (Goodman et al., 2008; Swingley & Humphrey, 2018), word frequency also influences the speed of lexical retrieval in typical children, children with SSDs, and adults (Jescheniak & Levelt, 1994; Newman & German, 2002). This effect could be due to the repetitive use of a word, which makes it more entrenched, i.e., more easily and automatically accessible for the speaker, as shown by cognitive linguistics research (Bybee, 2007; Gries, 1999).

As expected from this brief overview, word frequency is relevant in treating language-impaired patients. For example, research on children with phonological delays has revealed that treating an erred sound in a high-frequency word enables a systemwide improvement of their phonological accuracy, thus affecting non-treated sounds (Gierut & Morrisette, 2012). Furthermore, Core Vocabulary Therapy, a treatment based on lists of highly frequent words, has proved effective in children with phonological planning deficits (Crosbie et al., 2005). Word frequency is a crucial parameter in lexical retrieval since high-frequency words have lower recognition thresholds (Morton, 1969). The main treatments for phonological and lexical deficits are based on this assumption (Basso, 2005; Marangolo, 2012). Because of the significant role of frequency in language acquisition, processing, and therapy, it was given a central function in the design of DILLo.

Word frequencies are treated according to the Nuovo Vocabolario di Base dell’Italiano (NVdB, i.e., ‘New Basic Vocabulary of Italian’) (De Mauro, 1999, 2016), that is based on frequency counts from an 18-million-word corpus of spoken and written Italian and on psycholinguistic experiments (Chiari & De Mauro, 2014). It divides the lexicon into three categories (De Mauro, 2003):

  • Fundamental vocabulary (FO—Vocabolario Fondamentale): The 2000 most frequent lexemes in the Italian language, covering 86% of the corpus tokens (e.g., albero ‘tree’, avere ‘to have’, essere ‘to be’, mangiare ‘to eat’, felice ‘happy’, e ‘and’, o ‘or’).

  • High-usage vocabulary (AU—Alto Uso): An additional 3000 frequently used lexemes, accounting for another 6% of the corpus tokens (e.g., aeroporto ‘airport’, biscotto ‘cookie’, cucire ‘to sew’, amaro ‘bitter’).

  • High-availability vocabulary (AD—Alta Disponibilità): 2500 lexemes, less frequent than those in the two previous categories, yet essential in the everyday language of speakers (e.g., alluce ‘big toe’, peperoncino ‘chili pepper’, sganciare ‘unhook’, offeso ‘hurt’, prepotente ‘overbearing’).

Around 450 additional words were added, including frequent inflected forms of some of the lexemes. Currently, DILLo contains 7651 wordsFootnote 4 classified with De Mauro’s frequency labels plus five additional categories. Table 3 shows the categories used and their distribution in the database.

Table 3 Labels used to code word frequency in the DILLo database
Full size table

Additionally, to facilitate the query of the database, it is possible to filter the query according to three levels of word frequency: Alta (‘high’), Media (‘medium’), and Bassa (‘low’). This frequency-based subdivision of the words was carried out by coding them for their rank, according to the frequency list of the spoken corpus LIP available in De Mauro et al. (1993). Approximately, the level “high” includes the words ranked 1–3000 (n = 2602), the level “medium” those ranked 3001–20 000 (n = 3787), and the level “low” those from 20 000 onwards (n = 926).

2.3 Regional variants

DILLo also features a list of regional phonetic variants for each lexical entry. The Italian language situation can be characterized as a diaglossic continuum between Italo-Romance dialects, regiolects (regional Italians), and Standard Italian (Auer, 2005; Berruto, 2006). In particular, regiolects identifying features mainly come from local dialects (Grassi et al., 1997).

Regional Italians are the varieties currently employed in spoken communication across the peninsula (Poggi Salani, 2010). Standard Italian, instead, is a linguistic abstraction with no native speakers (Cerruti et al., 2017). Its phonetic and phonological features are not taught in schools, but they have nonetheless been codified. Several pronunciation models have succeeded over the years (Crocco, 2017); however, all these standard models are only followed by voice professionals, while the rest of the population uses pronunciations with different degrees of regionality and idiosyncrasy.

Therefore, including regional variants seemed a necessary step in creating a realistic and useful linguistic resource. Speech pathologists will likely encounter patients adhering to different regional standards, with variable rates of stigmatization, whose pronunciations could be different from those described by Standard Italian phonological transcriptions (cf. §2.1.). These regional features should not be regarded as speech errors or deviant pronunciations, as they are the expression of natural variation within language that ought to be normalized rather than stigmatized. This fact must be considered to avoid over- or under-identification of language impairment and confusion between difference and disorder (Ball & Bernhardt, 2008; Clark et al., 2020; Easton & Verdon, 2021). Instead, “[a]n unbiased approach towards working with culturally and linguistically diverse clients is imperative in speech pathology practice” (Clark et al., 2020). DILLo database includes regional variants to help speech pathologists discard inflectional differences and focus on pathological aspects.

The realization of the list of regional variants was carried out by a team of three linguists. The first step was the consultation of resources describing the phonological features of Italo-Romance dialects and regiolects. For this purpose, the following works were used: Canepari (1980, 1992, 1999), Loporcaro (2009), and De Blasi (2014). By crossing data from these sources, the team looked for horizontal processes of convergence among regiolects (Cerruti & Regis, 2015), i.e., for phonological features shared by different regiolects (despite them being independently emerged in each variety or spread due to contact – cf. koinè: Siegel, 2001; Trudgill, 2004). The set of most common and spread features of regional variation determined which phenomena were to be represented by the regional variants included in DILLo. The team tried to avoid being too specific with this task since the resource is addressed to clinicians from all parts of Italy. An excessive amount of characterization may have been too grounded in extremely local features, lacking adaptability to speakers of different communities, and too chaotic for easy consultation. The phenomena that have thus so far been considered in the regional variants are listed as follows:

  • Variation between mid-open and mid-closed stressed vowels in open syllables (e.g., [ˈbeːne] and [ˈbɛːne], ‘well’; [ˈkoːsa] and [ˈkɔːsa], ‘thing’).

  • Variation between open and closed stressed diphthongs in open syllables (e.g., [ˈfjoːre] and [ˈfjɔːre], ‘flower’; [kameˈrjeːre] and [kameˈrjɛːre], ‘waiter’; [ˈwoːmo] and [ˈwɔːmo], ‘man’).

  • Variation between mid-open and mid-closed stressed vowels in syllables closed by non-geminate /s/ or by a nasal consonant (e.g., [ˈfesta] and [ˈfɛsta], ‘party’; [komˈmento] and [komˈmɛnto], ‘comment’).

  • Variation between mid-open and mid-closed stressed vowels in the following suffixes: -ett*, -ezz*, -ott*, -ozz*Footnote 5 (e.g., [stanˈʦetta] and [stanˈʦɛtta], ‘little room’).

  • Variation between voiced and voiceless intervocalic /s/ (e.g., [ˈkaːsa] and [ˈkaːza], ‘home’).

Dubious cases were individually checked on the Italian dictionary provided as a handbook by Canepari (1992), which includes, for almost every entry, information about its variants in the main regional standards.

Finally, the team has manually transcribed the regional variants of each database entry based on the list above of phenomena. We hope to be able, in the future, to expand the list of considered phenomena while still maintaining a balanced amount of regional characterization.

2.4 Part of speech

The words in the database were also coded for their grammatical category, i.e., the “Part of Speech”—PoS (e.g., noun, verb, adjective, conjunction, etc.). This property has been proven relevant in the lexical acquisition, development, and processing in typical children and children with SSDs (e.g., Conti-Ramsden & Jones, 1997; Hansen, 2016; Sheng & McGregor, 2010). Furthermore, aphasiological data have led researchers to postulate a lexical class effect that affects word production and comprehension independently of the semantic system (Rapp & Caramazza, 1997; Voghera & Laudanna, 2003). Most studies on the topic deal with nouns and verbs, and less attention has been devoted to other word classes (Tribushinina & Dubinkina, 2012), probably due to class frequency. Research has shown that nouns and verbs are more frequently used in parental input than other open-word classes (e.g., adjectives and adverbs) (Sandhofer et al., 2000). Despite being less investigated, closed word classes (e.g., clitic pronouns and determiners) may still be of interest for research: some studies have found them particularly vulnerable in children with SSDs, suggesting weak mastery of syntax (Thordardottir & Namazi, 2007; Befi-Lopes et al., 2013. For a review of clitics acquisition patterns in Italian, see Caprin & Guasti, 2009; Suozzi & Gagliardi, 2022).

In DILLo, it is possible to filter the query by selecting one of the eleven parts of speech the words were coded for, which include both open and closed word classes. However, it must be pointed out that the choice of the parts of speech to consider and the identification of their members are not straightforward. The amount, nature, and boundaries of word classes are still an open debate in linguistic theory, as they are often theory-dependent (for a discussion on word classes in Italian, cf. Colombo & Graffi, 2017). For this reason, words were coded according to the part of speech specified in their entry in the NVdB (De Mauro, 2016; cf. §2.2) to maintain a theory-agnostic and consistent coding scheme. Table 4 shows the list of parts of speech included in the database and their corresponding number of words.

Table 4 Distribution of words based on their corresponding parts of speech in the DILLo database
Full size table

Finally, as Table 4 shows, a distinction between common and proper nouns was made. This feature can be relevant for specific needs in therapy since research on aphasic patients has suggested that the processing of proper and common nouns relies on distinct mechanisms (Yasuda et al., 2000). Proper nouns have been entered into the database mainly to facilitate the creation of some minimal pairs.

2.5 ARASAAC pictograms

A total of 3745 words in the database present their respective ARASAAC—Aragonese Portal of Augmentative and Alternative CommunicationFootnote 6 pictogram (Palao, 2013). ARASAAC is a free resource under the Creative Commons license, designed to represent a message through images of people, objects, or abstract ideas, without cultural, linguistic, or cognitive boundaries. The portal has become a point of reference for many countries, such as Italy, Spain, France, Brazil, Finland, Germany, and Belgium (Paolieri & Marful, 2018). The worldwide success of this resource has grown since its first implementation in 2007, and today it is equipped with a collection of more than 10,000 pictograms in twenty different languages, representing one of the most widely used symbolic systems for the AAC—Augmentative and Alternative Communication (AAC).Footnote 7 As a matter of fact, contrary to earlier claims, AAC proved to be a useful tool that does not hinder speech development but rather supports the development of communication skills (Cress & Marvin, 2003; cf. Romski et al., 2015, for a review on the topic). Early AAC interventions provide long-term benefits, and a multimodal strategy that combines existing oral language interventions with AAC may improve the quality of speech impairment treatment and language learning (Olive et al., 2006; Wright et al., 2013). Thus, pictograms may play a crucial role in the rehabilitation process, offering broader possibilities for linguistic and communicative comprehension for patients (Costantino et al., 2011). Additionally, this multimodal approach supports not only patients with severe communication disorders but also young children and illiterate individuals (Beukelman & Mirenda, 2013). The images also allow SLPs to easily create games and activities to stimulate target words or elicit the naming of certain items. These latter uses of ARASAAC images are part of common rehabilitation practice. For this reason, DILLo allows visualization of ARASAAC symbol(s) associated with words whenever a query is carried out. For example, Table 5 reports the pictograms of a set of words searched in the database as representatives of each part of speech.

Table 5 Examples of ARASAAC pictograms associated with words of each part of speech considered in DILLo
Full size table

Paolieri and Marful (2018) pointed out that the associations between words and pictograms in the ARASAAC database were made according to subjective criteria. This aspect could lead to a misunderstanding between the professional and the patient during a diagnostic or clinical treatment. Therefore, even though Paolieri and Marful’s (2018) psychometric analyses have shown a high degree of validity and reliability of ARASAAC pictograms, we would like to emphasize the need for a very cautious use of this tool.

3 DILLo’s design and search options

DILLo is a web application composed of a relational database and a modern HTML5 front-end. The tool supports live updates whenever new linguistic material is made available. The web interface is user-friendly and has a responsive design to be used by any device. Currently, the tool provides three search options: phoneme-based, grapheme-based, and a minimal pair extractor (cf. Fig. 1).

Fig. 1
figure 1

DILLo search interface

Full size image

The phoneme-based query enables users to search for segments that the patient should have either acquired or lost based on scientific descriptions of its specific disorder (e.g., developmental trajectories of phonological skills, articulatory complexity). The user is also able to select the position of the phoneme within the word (initial, medial, final) and to specify the phonological context (preceding and following phoneme), as clusters or specific types of vowels may have an impact on phoneme production.Footnote 8

When searching by grapheme, it is possible to enter a grapheme (or more than one) or a string (or more than one) to find all the matching elements in the database; as in the phoneme-based research, the position in the word can be selected. Furthermore, in both search options, results can be filtered by choosing the number of syllables, the part of speech, and the frequency of the words (cf. Fig. 2).

Fig. 2
figure 2

DILLo results page

Full size image

The user can also extract two words that form a minimal pair so that the patient can be exposed to the phonemic relevance of the sound they are failing to use, discriminate or identify (Aimar et al., 2009; Barlow & Gierut, 2002; Gierut, 1989, 1991; Williams, 2000). This can be done by choosing the dedicated search option and selecting the contrasting phonemes; it is also possible to specify their position in the word.

For each element of the resulting word list, whichever search option has been used, the clinician can choose to show or hide several linguistic and non-linguistic information based on the features DILLo’s words have been coded for, namely: its phonological transcription, part of speech, regional variant(s) (if present), ARASAAC pictogram(s) (if present), and grammatical information associated with the word (e.g., number, gender, tense, person) if the word is an inflected form of a lexeme.

4 Future directions and conclusion

In summary, the DILLo database can be exploited to retrieve lexical units by SLTs with peculiar characteristics, empowering a customized logopaedic intervention. It allows SLTs to support the development or rehabilitation of specific language competences, such as lexical and phonological skills, which, as part of the overall rehabilitative approach, may contribute to a comprehensive improvement in language difficulties. Nevertheless, it is crucial to note that the tool may not fully address the complex needs of therapists, as patient difficulties often extend beyond the realm of isolated words.

The resource is in continuous development to improve the quality of the service provided to SLPs during their professional activity. In 2021, the tool underwent a two-month testing period by 13 Italian SLPs for the rehabilitation of clinically relevant cases (i.e., DLD – Developmental Language Disorder, SSDs, CAS, SpLDs, aphasia, dysarthria, dysphonia, auditory training), achieving an excellent satisfaction level with 92.30% of positive feedback (Usardi, 2021). However, the resource will welcome further advice from the professionals using it to enhance as many functions as required.

Interesting future developments, some of which have successfully been included in lexical resources from other languages, may include:

  • The addition of a query option to filter phonemes according to the articulatory movements required for their production (Hoch et al., 1986).

  • The addition of an interrogation option allowing the extraction of “pseudo-minimal pairs” based on consonant length (e.g., [ˈka:ne] ‘dog’ ~ [ˈkanne] ‘reeds’, [ˈfa:to] ‘destiny’ ~ [ˈfatto] ‘fact’/ ‘made’) and lexical accent position (e.g., [ˈaŋkora] ‘anchor’ ~ [aŋˈko:ra] ‘again’/ ‘still’, [ˈpa:pa] ‘pope’ ~ [paˈpa] ‘dad’). As a matter of fact, these phonological phenomena play a significant role in Italian. Specifically, consonantal quantity is contrastive on a regular basis and, as a result, many words may differ by only one geminated consonant (for a comprehensive review, the reader may refer to Bertinetto, 1981; Loporcaro, 1992 and Di Benedetto et al., 2021). Moreover, this feature holds significant typological relevance (e.g., compared to English, Spanish, and Chinese). Therefore, as an illustration, discrimination and production tasks can be administered to bilingual speakers with a first language different from Italian.

  • The implementation of a virtual phonetic keyboard to facilitate entering IPA characters.

  • The extension of the phonological context, in the phoneme-based query, to ± 2 phonemes adjacent to the selected one. For example, this option would allow comparing clusters based on the vowel following them (e.g., tra vs. tro).

  • The possibility for the user to visualize the phonotactic structure of the word and its phonological transcription to assess syllable boundaries.

  • The integration of other psycholinguistic information regarding semantic and lexical variables (Barca et al., 2002; Montefinese et al., 2014; Navarrete et al., 2019; Repetto et al., 2023; Spataro et al., 2019; Vergallito et al., 2020), i.e., the age of acquisition (Caselli et al., 2007; Marconi et al., 1993; Montefinese et al., 2019) and imageability of the word, or the frequency of the word in written production, both during childhood (Marconi et al., 1993) and adulthood (Bertinetto et al., 2005).

  • The possibility to print a customized table with selected words and pictograms.

  • The English translation of the browser interface to allow non-Italian researchers to access the resource.

In conclusion, DILLo represents a valuable resource for treating speech impairments, as it is the fruitful result of a cooperative effort between linguists and speech pathologists. The contribution of linguists to the creation of language rehabilitation tools is crucial to increasing and enhancing the database by providing controlled linguistic materials based on theoretical and methodological frameworks. In parallel, the participation of speech pathologists is necessary to ideate personalized treatments and design new features based on rehabilitation requirements.