Words and their behaviour

It is quite worrisome that syntactic norms and values, which ideally guide the arrangement of words, phrases, clauses and other lexical categories, are not strictly upheld and followed by many researchers and even research bodies. That implies violating some basics of research ethics. This study rises to make a critical exposition of syntactic rules in research ethics, which are often violated. Relying on observation and secondary data, the study demonstrates that poor knowledge and negligence of syntactic rules largely accounts for the violation of syntactic rules in research works by researchers and many concerned research bodies. It shows that unacceptable grammatical structures have grave implications for the overall given research work. Also, phrase structure rules correlate with interpretative rules, just as syntax and semantics also correlate. The study concludes that since language is rule-governed, the violation of syntactic rules in research work implies the violation of language rules and research ethics. The study is anchored on the syntactic theory of Universal Grammar, which explains how standard syntactic rules and research ethics are universal, and must be followed in order to have acceptable and correct sentences and research.

We addressed an understudied topic in the literature of language disorders, that is, processing of derivational morphology, a domain which requires integration of semantic and syntactic knowledge. Current psycholinguistic literature suggests that word processing involves morpheme recognition, which occurs immediately upon encountering a complex word. Subsequent processes take place in order to interpret the combination of stem and affix. We investigated the abilities of individuals with agrammatic (PPA-G) and logopenic (PPA-L) variants of primary progressive aphasia (PPA) and individuals with stroke-induced agrammatic aphasia (StrAg) to process pseudowords which violate either the syntactic (word class) rules (*reheavy) or the semantic compatibility (argument structure specifications of the base form) rules (*reswim). To this end, we quantified aspects of word knowledge and explored how the distinct deficits of the populations under investigation affect their performance. Thirty brain-damaged individuals and 10 healthy controls participated in a lexical decision task. We hypothesized that the two agrammatic groups (PPA-G and StrAg) would have difficulties detecting syntactic violations, while no difficulties were expected for PPA-L. Accuracy and Reaction Time (RT) patterns indicated: the PPA-L group made fewer errors but yielded slower RTs compared to the two agrammatic groups which did not differ from one another. Accuracy rates suggest that individuals with PPA-L distinguish *reheavy from *reswim, reflecting access to and differential processing of syntactic vs. semantic violations. In contrast, the two agrammatic groups do not distinguish between *reheavy and *reswim. The lack of difference stems from a particularly impaired performance in detecting syntactic violations, as they were equally unsuccessful at detecting *reheavy and *reswim. Reduced grammatical abilities assessed through language measures are a significant predictor for this performance, suggesting that the “hardware” to process syntactic information is impaired. Therefore, they can only judge violations semantically where both *reheavy and *reswim fail to pass as semantically ill-formed. This finding further suggests that impaired grammatical knowledge can affect word level processing as well. Results are in line with the psycholinguistic literature which postulates the existence of various stages in accessing complex pseudowords, highlighting the contribution of syntactic/grammatical knowledge. Further, it points to the worth of studying impaired language performance for informing normal language processes.

We addressed an understudied topic in the literature of language disorders, i.e. processing of derivational morphology, a domain which requires integration of semantic and syntactic knowledge. Current psycholinguistic literature suggests that word processing involves morpheme recognition occurring immediately upon encountering a complex word. Subsequent processes take place in order to interpret the combination of stem and affix. We investigated the abilities of individuals with agrammatic (PPA-G) and logopenic (PPA-L) variants of primary progressive aphasia and individuals with stroke-induced agrammatic aphasia (StrAg) to process pseudowords which violate either the syntactic (word class) rules (*reclever) or the semantic compatibility (argument structure specifications of the base form) rules (*resmile). To this end, we quantified aspects of word knowledge and explored how the distinct deficits of the populations under investigation affect their performance. Thirty brain-damaged individuals and 10 healthy controls participated in a lexical decision task. We hypothesized that the two agrammatic groups (PPA-G and StrAg) would have difficulties detecting syntactic violations, while no difficulties were expected for PPA-L. Accuracy and Reaction Time (RT) patterns indicated: the PPA-L group made fewer errors but yielded slower RTs compared to the two agrammatic groups which did not differ from one another. Accuracy rates suggest that individuals with PPA-L distinguish *reclever from *resmile, reflecting access to and differential processing of syntactic vs. semantic violations. In contrast, the two agrammatic groups do not distinguish between *reclever and *resmile. The lack of difference stems from a particularly impaired performance in detecting syntactic violations, as they were equally unsuccessful at detecting *reclever and *resmile. Reduced grammatical abilities assessed through language measures are a significant predictor for this performance, suggesting that the “hardware” to process syntactic information is impaired. Therefore, they can only judge violations semantically where both *reclever and *resmile fail to pass as semantically ill-formed. This finding further suggests that impaired grammatical knowledge can affect word level processing as well. Results are in line with the literature which postulates the existence of stages in accessing complex pseudowords, highlighting the contribution of syntactic/grammatical knowledge. Further, it points to the worth of studying impaired language performance for informing normal language processes.

Structuralists and generativists define word classes distributionally (Palmer 1971, Baker 2003, Aarts 2007), while cognitive linguists take a semantic (Langacker 1987a) or semantic-pragmatic approach (Croft 1991, 2001). Psycholinguistic research, by contrast, has shown that phonological properties also play a role (Kelly 1992, Monaghan et al. 2005). This study reports on a production experiment involving English nonce nouns and verbs. The data confirm the importance of phonology, whilst also suggesting that distributional facts are involved in lexical categorisation. Together with the existing psycholinguistic evidence, the results show that both the generative and cognitive models of word classes are too restricted. However, the usage-based model can accommodate the facts straightforwardly. This was already anticipated by Taylor (2002) but is worked out in more detail here by elaborating on his notion of phonological “sub-schemas” and by bringing together insights from Croft (1991, 2001) related to discourse propositional act constructions and recent suggestions by Langacker (2008b) concerning “summary scanning” and “sequential scanning”.

On-line translation of natural language questions into artificial language queries

Generative modeling for chemistry has advanced rapidly in recent years, but this surge in popularity raises a foundational question: which molecular representation is best suited for modern machine learning models? Despite not being designed for generative tasks, SMILES remains the most commonly used string-based representation. However, while SMILES follows strict syntactic rules, grammatically correct SMILES strings do not always correspond to valid molecules. SELFIES, an alternative grammar, addresses this limitation by ensuring that every string of SELFIES tokens represents a valid molecule. In this study, we comprehensively evaluate the limitations of both SMILES and SELFIES as representations for generative models. We analyze two key criteria for robust molecular generation: viability, which means that generated strings represent novel, unique molecules with correct valence, and fidelity, where the distribution of physicochemical properties from sampled molecules resembles that of the training data. We find that approximately one-fifth of the molecules generated using RDKit default canonical SMILES are invalid, failing the viability criterion. In contrast, all SELFIES-generated molecules are viable, but they deviate significantly from the training distribution, indicating low fidelity. To address these limitations, we develop data augmentation procedures for both representations. While simplifying the SELFIES grammar yields only modest gains in fidelity, our stochastic augmentation method for SMILES, ClearSMILES, significantly improves both viability and fidelity. ClearSMILES simplifies syntax by reducing the vocabulary size and explicitly encoding aromaticity via Kekule SMILES, making the string representations easier for models to process. Using ClearSMILES, the rate of invalid samples decreases by an order of magnitude, from 20 to 2.2%, and fidelity to the training distribution is also moderately improved.